1 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4 rcrt1.o%s/grcrt1.o%s for -static-pie.
6 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
8 * i386.c (ix86_multiplication_cost, ix86_division_cost,
9 ix86_shift_rotate_cost): Break out from ...
10 (ix86_rtx_costs): ... here.
11 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
14 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
16 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
17 when entry block was promoted unlikely.
18 (estimate_bb_frequencies): Increase frequency scale.
19 * profile-count.h (profile_count): Export precision info.
21 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
23 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
24 disturbin profile of entry block.
26 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
28 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
31 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
33 * ipa-cp.c (update_profiling_info): Handle conversion to local
35 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
37 2017-11-17 Jeff Law <law@redhat.com>
39 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
40 from evrp_dom_walker class. Various methods moved into new class.
41 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
42 (evrp_range_analyzer::enter): New method.
43 (evrp_range_analyzer::leave): New method.
44 (evrp_dom_walker): Remove delegators no longer needed by this class.
45 Replace vr_values data member with evrp_range_analyzer
47 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
48 method extracted from evrp_dom_walker::before_dom_children.
49 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
50 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
52 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
53 Add private copy constructor and move assignment operators.
54 Privatize methods and class data where trivially possible.
55 (evrp_dom_walker::cleanup): New function, extracted from
56 execute_early_vrp. Simplify access to class data.
58 * vr-values.h (get_output_for_vrp): Prototype.
59 * vr-values.c (get_output_for_vrp): New function extracted from
60 vrp_visit_assignment_or_call and extract_range_from_stmt.
61 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
63 2017-11-17 Luis Machado <luis.machado@linaro.org>
65 * config/aarch64/aarch64.c
66 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
67 (qdf24xx_tunings) <autoprefetcher_model>: Set to
68 tune_params::AUTOPREFETCHER_WEAK.
70 2017-11-17 Tamar Christina <tamar.christina@arm.com>
73 * config/arm/arm.c (arm_valid_target_attribute_rec):
74 Parse "arch=" and "+<ext>".
75 (arm_valid_target_attribute_tree): Re-init global options.
76 (arm_option_override): Make non-static.
77 (arm_options_perform_arch_sanity_checks): Make errors fatal.
78 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
79 (__ARM_FEATURE_CRC32): Support undef.
80 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
81 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
83 2017-11-17 David Malcolm <dmalcolm@redhat.com>
85 * gdbinit.in (break-on-diagnostic): New command.
87 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
89 * config/i386/i386.c (ix86_expand_epilogue): Change simple
90 return to indirect jump for EH return if control-flow
91 protection is enabled. Change explicit 'false' argument in
92 pro_epilogue_adjust_stack with a value of
94 * config/i386/i386.md (simple_return_indirect_internal):
95 Remove SImode restriction to support 64-bit.
97 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
99 * combine.c (added_notes_insn): New.
100 (try_combine): Handle added_notes_insn like added_links_insn.
101 Rewrite return value code.
102 (distribute_notes): Set added_notes_insn to the earliest insn we added
105 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
107 PR rtl-optimization/82621
108 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
109 dest of one of those SETs is unused.
111 2017-11-17 Richard Biener <rguenther@suse.de>
114 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
115 * tree-pretty-print.c (dump_generic_node): Handle
116 annot_expr_parallel_kind.
117 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
118 * gimplify.c (gimple_boolify): Likewise.
120 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
122 * config.gcc (extra_headers): Add cet.h for x86 targets.
123 * config/i386/cet.h: New file.
124 * doc/install.texi: Add --enable-cet/--disable-cet.
126 2017-11-17 Richard Biener <rguenther@suse.de>
128 PR tree-optimization/83017
129 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
130 (gen_parallel_loop): Properly count iterations.
131 (parallelize_loops): Handle loop->can_be_parallel independent
132 of flag_loop_parallelize_all. Make static profitability test match
134 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
135 * invoke.texi (parloops-min-per-thread): Document.
137 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
139 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
140 upstreaming review comments.
142 2017-11-17 Tamar Christina <tamar.christina@arm.com>
144 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
145 with fast unaligned access.
146 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
148 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
150 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
151 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
153 2017-11-17 Richard Biener <rguenther@suse.de>
155 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
156 folding of references.
158 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
161 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
162 of replacing call to strncmp with corresponding call to strcmp when
165 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
167 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
168 option prefer-avx256 for skylake-avx512 configuration.
169 * config/i386/i386.c (ix86_option_override_internal): Ditto.
170 (get_builtin_code_for_version): Ditto.
172 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
173 Monk Chiang <sh.chiang04@gmail.com>
175 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
176 (FIXED_REGISTERS): Reserve more register numbers.
177 (CALL_USED_REGISTERS): Likewise.
178 (REG_ALLOC_ORDER): Likewise.
179 (REG_CLASS_CONTENTS): Likewise.
180 (REGISTER_NAMES): Likewise.
182 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
183 Kito Cheng <kito.cheng@gmail.com>
185 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
187 * config/nds32/iterators.md: Add vector mode iterators and attributes.
189 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
191 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
192 parameter list for vec_splats.
194 2017-11-16 Joseph Myers <joseph@codesourcery.com>
196 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
198 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
201 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
204 * doc/install.texi: Document incompatibility of Perl >=5.6.26
205 with the required version of automake 1.11.6.
207 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
209 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
211 (DU_C2_3_power9): Correct reservation combinations.
212 (FP_DIV_power9, VEC_DIV_power9): New.
213 (power9-alu): Split out rotate/shift...
214 (power9-rot): ...to here, correct dispatch resource.
215 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
217 (power9-fp): Correct latency.
218 (power9-sdiv): Add div/sqrt resource.
219 (power9-ddiv): Correct latency, add div/sqrt resource.
220 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
221 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
223 (power9-qpdiv, power9-qpmul): Adjust resource usage.
225 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
227 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
228 switch statement mapping KF built-ins to TF built-ins if we don't
229 have the proper ISA 3.0 assembler support.
231 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
233 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
234 (lower_emutls_function_body): Do not compute it.
236 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
238 * ipa-split.c (split_bb_info): Turn time to sreal.
239 (split_point): Likewise.
240 (dump_split_point): Likewise.
241 (fine_split_points): Likewise.
242 (execute_split_functions): Only zero split_bbs; turn time to sreals.
244 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
246 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
249 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
251 * predict.c (combine_predictions_for_bb): Preserve zero predicted
253 (expensive_function_p): Remove useless assert.
254 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
256 2017-11-16 Martin Sebor <msebor@redhat.com>
258 PR tree-optimization/82588
259 PR tree-optimization/82583
260 * tree-vrp.c (check_array_ref): Handle flexible array members,
261 string literals, and inner indices.
262 (search_for_addr_array): Add detail to diagnostics.
264 2017-11-16 Nathan Sidwell <nathan@acm.org>
268 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
269 (SET_DECL_ASSEMBLER_NAME): Forward to
270 overwrite_decl_assembler_name.
271 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
272 (overwrite_decl_assembler_name): Declare.
273 * tree.c (overwrite_decl_assembler_name): New.
274 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
275 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
276 (LANG_HOOKS_INITIALIZER): Add it.
277 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
278 * langhooks.c (lhd_set_decl_assembler_name): Use
279 SET_DECL_ASSEMBLER_NAME.
280 (lhd_overwrite_decl_assembler_name): Default implementation.
282 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
283 Jackson Woodruff <jackson.woodruff@arm.com>
285 PR tree-optimization/71026
286 * match.pd: Canonicalize constant multiplies in division.
288 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
290 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
291 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
292 Remove OPT_fomit_frame_pointer entry.
293 * common/config/arc/arc-common.c: Likewise.
294 * common/config/arm/arm-common.c: Likewise.
295 * common/config/avr/avr-common.c: Likewise.
296 * common/config/c6x/c6x-common.c: Likewise.
297 * common/config/cr16/cr16-common.c: Likewise.
298 * common/config/cris/cris-common.c: Likewise.
299 * common/config/epiphany/epiphany-common.c: Likewise.
300 * common/config/fr30/fr30-common.c: Likewise.
301 * common/config/frv/frv-common.c: Likewise.
302 * common/config/ia64/ia64-common.c: Likewise.
303 * common/config/iq2000/iq2000-common.c: Likewise.
304 * common/config/lm32/lm32-common.c: Likewise.
305 * common/config/m32r/m32r-common.c: Likewise.
306 * common/config/mcore/mcore-common.c: Likewise.
307 * common/config/microblaze/microblaze-common.c: Likewise.
308 * common/config/mips/mips-common.c: Likewise.
309 * common/config/mmix/mmix-common.c: Likewise.
310 * common/config/mn10300/mn10300-common.c: Likewise.
311 * common/config/nios2/nios2-common.c: Likewise.
312 * common/config/pa/pa-common.c: Likewise.
313 * common/config/pdp11/pdp11-common.c: Likewise.
314 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
315 * common/config/riscv/riscv-common.c: Likewise.
316 * common/config/rs6000/rs6000-common.c: Likewise.
317 * common/config/rx/rx-common.c: Likewise.
318 * common/config/s390/s390-common.c: Likewise.
319 * common/config/sh/sh-common.c: Likewise.
320 * common/config/sparc/sparc-common.c: Likewise.
321 * common/config/tilegx/tilegx-common.c: Likewise.
322 * common/config/tilepro/tilepro-common.c: Likewise.
323 * common/config/v850/v850-common.c: Likewise.
324 * common/config/visium/visium-common.c: Likewise.
325 * common/config/xstormy16/xstormy16-common.c: Likewise.
326 * common/config/xtensa/xtensa-common.c: Likewise.
327 * invoke.texi (-fomit-frame-pointer): Update documentation.
329 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
331 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
333 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
335 * cfg.c (scale_bbs_frequencies_int,
336 cale_bbs_frequencies_gcov_type): Remove.
337 * cfg.h (scale_bbs_frequencies_int,
338 cale_bbs_frequencies_gcov_type): Remove.
340 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
342 * tree-ssa-loop-manip.c
343 (scale_dominated_blocks_in_loop): Update to profile counts.
344 (tree_transform_and_unroll_loop): Likewise.
346 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
348 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
349 scale_bbs_frequencies_int.
351 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
353 * final.c (compute_alignments): Use counts rather than frequencies.
355 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
357 * cfgloopanal.c: Include sreal.h
358 (average_num_loop_insns): Use counts and sreal for accounting.
360 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
362 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
365 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
366 Kito Cheng <kito.cheng@gmail.com>
368 * config/nds32/constraints.md: Provide more constraints.
369 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
370 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
371 support constraints usage.
373 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
375 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
377 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
378 Kito Cheng <kito.cheng@gmail.com>
380 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
381 * config/nds32/nds32.opt: Refine the layout.
382 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
383 TARGET_EXT_STRING): Support new options.
384 * config/nds32/nds32.h: Likewise.
385 * config/nds32/nds32.md: Likewise.
386 * config/nds32/nds32-predicates.c: Likewise.
387 * config/nds32/constraints.md: Likewise.
388 * common/config/nds32/nds32-common.c: Likewise.
390 2017-11-16 Julia Koval <julia.koval@intel.com>
393 * config/i386/gfniintrin.h: Add sse check.
394 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
396 2017-11-16 Julia Koval <julia.koval@intel.com>
398 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
399 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
400 (ix86_handle_option): Handle -mavx512vbmi2.
401 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
402 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
403 * config/i386/i386-c.c (__AVX512VBMI2__): New.
404 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
405 (ix86_valid_target_attribute_inner_p): Ditto.
406 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
407 * config/i386/i386.opt (mavx512vbmi2): New option.
408 * doc/invoke.texi: Add new option.
410 [2017-11-16 Julia Koval <julia.koval@intel.com>
412 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
413 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
414 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
415 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
416 _mm512_gf2p8mul_epi8): New intrinsics.
417 * config/i386/i386-builtin-types.def
418 (V64QI_FTYPE_V64QI_V64QI): New type.
419 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
420 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
421 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
422 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
423 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
424 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
426 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
428 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
429 explicitly as a stream of bytes.
431 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
433 * config/rs6000/altivec.h (vec_xst_be): New #define.
434 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
435 and externalize from *altivec_vperm_<mode>_internal.
436 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
438 (XL_BE_V8HI): Likewise.
439 (XL_BE_V4SI): Likewise.
440 (XL_BE_V4SI): Likewise.
441 (XL_BE_V2DI): Likewise.
442 (XL_BE_V4SF): Likewise.
443 (XL_BE_V2DF): Likewise.
444 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
445 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
446 all array entries with these keys: VSX_BUILTIN_VEC_XL,
447 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
448 VSX_BUILTIN_VEC_XST_BE.
449 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
450 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
452 (altivec_init_builtins): Replace conditional calls to def_builtin
453 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
454 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
455 with unconditional calls. Remove calls to def_builtin for
456 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
457 __builtin_vec_xst_be.
458 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
459 to define_expand, and add alternate RTL generation for P8.
460 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
462 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
463 add alternate RTL generation for P8.
464 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
465 vsx_ld_elemrev_v16qi.
466 (vsx_st_elemrev_v8hi): Convert define_insn
467 to define_expand, and add alternate RTL generation for P8.
468 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
470 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
471 add alternate RTL generation for P8.
472 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
473 vsx_st_elemrev_v16qi.
475 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
478 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
479 TARGET_AVX512ER check.
480 (ix86_option_override_internal): Set MASK_VZEROUPPER if
481 neither -mzeroupper nor -mno-zeroupper is used and
482 TARGET_EMIT_VZEROUPPER is set.
483 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
484 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
486 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
488 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
489 folding of vector compares.
490 (fold_build_vec_cmp): New helper function.
491 (fold_compare_helper): New helper function.
492 (builtin_function_type): Add compare builtins to the list of functions
493 having unsigned arguments. Cosmetic updates to comment indentation.
494 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
495 the not+eq combination.
497 2017-11-15 Bin Cheng <bin.cheng@arm.com>
499 PR tree-optimization/82726
500 PR tree-optimization/70754
501 * tree-predcom.c (order_drefs_by_pos): New function.
502 (combine_chains): Move code setting has_max_use_after to...
503 (try_combine_chains): ...here. New parameter. Sort combined chains
504 according to position information.
505 (tree_predictive_commoning_loop): Update call to above function.
506 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
508 2017-11-15 Bin Cheng <bin.cheng@arm.com>
510 PR tree-optimization/82726
512 2017-01-23 Bin Cheng <bin.cheng@arm.com>
514 PR tree-optimization/70754
515 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
516 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
517 combined stmt before it if not NULL.
518 (combine_chains): Process refs reversely and compute dominance point
522 2017-02-23 Bin Cheng <bin.cheng@arm.com>
524 PR tree-optimization/79663
525 * tree-predcom.c (combine_chains): Process refs in reverse order
526 only for ZERO length chains, and add explaining comment.
528 2017-11-15 Tamar Christina <tamar.christina@arm.com>
530 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
531 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
532 * doc/invoke.texi (ARM Options): Add armv8.3-a.
534 2017-11-15 Tamar Christina <tamar.christina@arm.com>
536 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
538 2017-11-15 Martin Liska <mliska@suse.cz>
540 * tree-cfg.c (pass_warn_function_return::execute):
541 Compare warn_return_type for greater than zero.
543 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
547 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
548 to return true on Xeon and not on Xeon Phi.
549 (ix86_check_avx256_register): Changed to ...
550 (ix86_check_avx_upper_register): ... this. Add extra check for
551 VALID_AVX512F_REG_OR_XI_MODE.
552 (ix86_avx_u128_mode_needed): Changed
553 ix86_check_avx256_register to ix86_check_avx_upper_register.
554 (ix86_check_avx256_stores): Changed to ...
555 (ix86_check_avx_upper_stores): ... this. Changed
556 ix86_check_avx256_register to ix86_check_avx_upper_register.
557 (ix86_avx_u128_mode_after): Changed
558 avx_reg256_found to avx_upper_reg_found. Changed
559 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
560 (ix86_avx_u128_mode_entry): Changed
561 ix86_check_avx256_register to ix86_check_avx_upper_register.
562 (ix86_avx_u128_mode_exit): Ditto.
563 * config/i386/i386.h: (host_detect_local_cpu): New define.
565 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
567 * config/arm/xgene1.md (xgene1): Split into automatons
568 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
569 (xgene1_f_load): Adjust reservations and/or types.
570 (xgene1_f_store): Likewise.
571 (xgene1_load_pair): Likewise.
572 (xgene1_store_pair): Likewise.
573 (xgene1_fp_load1): Likewise.
574 (xgene1_load1): Likewise.
575 (xgene1_store1): Likewise.
576 (xgene1_move): Likewise.
577 (xgene1_alu): Likewise.
578 (xgene1_simd): Likewise.
579 (xgene1_bfm): Likewise.
580 (xgene1_neon_load1): Likewise.
581 (xgene1_neon_store1): Likewise.
582 (xgene1_neon_logic): Likewise.
583 (xgene1_neon_st1): Likewise.
584 (xgene1_neon_ld1r): Likewise.
585 (xgene1_alu_cond): Added.
586 (xgene1_shift_reg): Likwise.
587 (xgene1_bfx): Likewise.
588 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
590 2017-11-15 Jakub Jelinek <jakub@redhat.com>
593 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
595 (can_widen_mult_without_libcall): New function.
596 (expand_mul_overflow): If only checking unsigned mul overflow,
597 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
598 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
599 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
600 (expand_DIVMOD): Formatting fix.
601 * expmed.h (expand_mult): Add NO_LIBCALL argument.
602 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
603 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
605 PR tree-optimization/82977
606 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
607 constructed temporary to strlen_to_stridx.put.
609 2017-11-15 Martin Liska <mliska@suse.cz>
611 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
612 * configure: Regenerate.
614 2017-11-15 Martin Liska <mliska@suse.cz>
617 * config/sh/sh-mem.cc: Use proper probability for
620 2017-11-14 Jeff Law <law@redhat.com>
622 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
623 the red zone for stack_clash_protection_final_dynamic_probe targets
624 when the total dynamic stack size is zero bytes.
626 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
627 blocks is post order.
629 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
631 * dumpfile.h (TDF_COMPARE_DEBUG): New.
632 * final.c (rest_of_clean_state): Set it for the
633 -fcompare-debug dump.
634 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
635 class when TDF_COMPARE_DEBUG is set.
637 * dwarf2out.c (gen_producer_string): Discard
640 2017-11-15 Joseph Myers <joseph@codesourcery.com>
643 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
644 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
645 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
646 (__TGMATH_REAL_2_3): Remove macros.
647 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
648 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
650 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
652 (remquo): Define using __TGMATH_REAL_3.
654 2017-11-14 Jeff Law <law@redhat.com>
656 * vr-values.c: New file with contents extracted from tree-vrp.c.
657 * Makefile.in (OBJS): Add vr-values.o
658 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
659 (set_value_range, set_and_canonicalize_value_range): Likewise.
660 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
661 (value_range_constant_singleton, symbolic_range_p): Likewise.
662 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
663 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
664 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
665 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
666 (find_case_label_range, find_case_label_index): Likewise.
667 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
668 (range_int_cst_singleton_p, value_inside_range): Likewise.
669 (get_single_symbol): Likewise.
670 (switch_update): Move structure definition here.
671 (to_remove_edges, to_update_switch_stmts): Provide externs.
672 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
673 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
674 (vrp_val_is_min, set_value_range): Likewise.
675 (set_and_canonicalize_value_range, copy_value_range): Likewise.
676 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
677 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
678 (range_is_nonnull, range_int_cst_p): Likewwise.
679 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
680 (get_single_symbol, operand_less_p): Likewise
681 (compare_values_warnv, compare_values): Likewise.
682 (value_inside_range, value_range_constant_singleton): Likewise.
683 (zero_nonzero_bitgs_from_vr): Likewise.
684 (extract_range_from_binary_expr_1): Likewise.
685 (overflow_comparison_p): Likewise.
686 (to_remove_edges, to_update_switch_stmts): Likewise.
687 (find_case_label-index, find_case_label_range): Likewise.
688 (switch_update, set_value_range_to_nonnegative): Remove.
689 (set_value_range_to_truthvalue): Likewise.
690 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
691 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
692 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
693 (find_case_label_ranges, test_for_singularity): Likewise.
694 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
695 (x_vr_values): Move to its remaining use site.
697 2017-11-10 Jeff Law <law@redhat.com>
699 * vr-values.h (VR_INITIALIZER): Move #define here.
700 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
701 * Makefile.in (OBJS): Add tree-evrp.o
702 * tree-vrp.h (assert_info): Move structure definition here.
703 (set_value_range_to_varying): Prototype.
704 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
705 (infer_value_range, register_edge_assert_for): Likewise.
706 (stmt_interesting_for_vrp): Likewise.
707 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
708 (set_value_range_to_varying): No longer static.
709 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
710 (infer_value_range, register_edge_assert_for): Likewise.
712 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
714 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
715 generate the XXBRD instruction.
717 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
718 (rs6000_builtin_type_compatible): Treat _Float128 and long double
719 as being compatible if -mabi=ieeelongdouble.
720 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
721 to setup float128 built-ins with hardware support.
722 (BU_FLOAT128_HW_2): Likewise.
723 (BU_FLOAT128_HW_3): Likewise.
724 (BU_FLOAT128_HW_VSX_1): Likewise.
725 (BU_FLOAT128_HW_VSX_2): Likewise.
726 (scalar_extract_expq): Change float128 built-in functions to
727 accommodate having both KFmode and TFmode functions. Use the
728 KFmode variant as the default.
729 (scalar_extract_sigq): Likewise.
730 (scalar_test_neg_qp): Likewise.
731 (scalar_insert_exp_q): Likewise.
732 (scalar_insert_exp_qp): Likewise.
733 (scalar_test_data_class_qp): Likewise.
734 (sqrtf128_round_to_odd): Delete processing the round to odd
735 built-in functions as special built-in functions, and define them
736 as float128 built-ins. Use the KFmode variant as the default.
737 (truncf128_round_to_odd): Likewise.
738 (addf128_round_to_odd): Likewise.
739 (subf128_round_to_odd): Likewise.
740 (mulf128_round_to_odd): Likewise.
741 (divf128_round_to_odd): Likewise.
742 (fmaf128_round_to_odd): Likewise.
743 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
744 support for KFmode and TFmode xststdcqp calls.
745 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
746 point, switch the built-in handlers for the get/set float128
747 exponent, get float128 mantissa, float128 test built-ins, and the
748 float128 round to odd built-in functions. Eliminate creating the
749 float128 round to odd built-in functions as special built-ins.
750 (rs6000_init_builtins): Eliminate special creation of the float128
751 round to odd built-in functions.
752 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
753 function insns to support both TFmode and KFmode variants.
754 (xsxsigqp_<mode>): Likewise.
755 (xsiexpqpf_<mode>): Likewise.
756 (xsiexpqp_<mode>): Likewise.
757 (xststdcqp_<mode>): Likewise.
758 (xststdcnegqp_<mode>): Likewise.
759 (xststdcqp_<mode>): Likewise.
761 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
763 * tree-ssa-threadupdate.c (compute_path_counts): Remove
764 unused path_in_freq_ptr parameter.
765 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
767 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
769 * ipa-inline.c (edge_badness): Dump sreal frequency.
770 (compute_inlined_call_time): Match natural implementaiton ...
771 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
772 forgotten division by CGRAPH_FREQ_BASE.
774 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
776 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
777 Solaris 11. Update comment.
778 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
780 * config/sol2.h (STARTFILE_SPEC): Likewise.
781 * configure: Regenerate.
783 2017-11-14 Carl Love <cel@us.ibm.com>
785 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
786 le_ and be_ prefixes to swap* variables. Remove
787 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
790 2017-11-14 Jason Merrill <jason@redhat.com>
792 Support GTY((cache)) on hash_map.
793 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
794 (ggc_cache_remove): Override it instead of ggc_mx.
795 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
796 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
797 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
798 (simple_cache_map_traits): Override maybe_mx.
799 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
800 (hash_map): Friend gt_cleare_cache.
801 (gt_cleare_cache): New.
802 * tree.h (tree_cache_traits): New hash_map traits class.
803 (tree_cache_map): New typedef.
805 2017-11-14 Richard Biener <rguenther@suse.de>
807 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
808 paramter and handling.
809 (cleanup_control_flow_bb): Likewise.
810 (cleanup_control_flow_pre): New helper performing a DFS walk
811 to call cleanup_control_flow_bb in PRE order.
812 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
813 via cleanup_control_flow_pre.
815 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
817 * config/aarch64/aarch64-simd.md
818 (aarch64_simd_bsl<mode>_internal): Remove DImode.
819 (*aarch64_simd_bsl<mode>_alt): Likewise.
820 (aarch64_simd_bsldi_internal): New.
821 (aarch64_simd_bsldi_alt): Likewise.
823 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
825 * tracer.c (better_p): Do not compare frequencies.
826 * reg-stack.c (better_edge): Likewise.
827 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
830 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
832 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
833 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
834 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
835 * lto-streamer-in.c (input_function): Use update_max_bb_count.
836 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
837 * predict.c (maybe_hot_frequency_p): Inline to ...
838 (maybe_hot_count_p): ... here; rewrite to counts.
839 (counts_to_freqs): Rename to ...
840 (update_max_bb_count): ... this one.
841 (expensive_function_p): Use counts.
842 (estimate_bb_frequencies): Update.
843 (rebuild_frequencies): Update.
844 * predict.h (counts_to_freqs): Rename to ...
845 (update_max_bb_count): ... this one.
846 * profile.c (compute_branch_probabilities): Add debug info
847 * tree-inline.c (expand_call_inline): Update debug info.
848 (optimize_inline_calls): Use update_max_bb_count..
849 (tree_function_versioning): Use update_max_bb_count..
850 * value-prof.c (gimple_value_profile_transformations):
851 Do not use update_max_bb_count.
853 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
855 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
856 always use frequencies.
858 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
860 * bb-reorder.c: Remove frequencies from comments.
861 (better_edge_p): Use profile counts.
862 (find_traces): Dump profile counts.
863 (rotate_loop): Use profile counts.
864 (find_traces_1_round): Likewise.
865 (connect_better_edge_p): Use counts instead of probabilities for
867 (copy_bb_p): Drop early check for non-0 frequency.
868 (sanitize_hot_paths): Update comments.
870 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
872 * ipa-split.c (struct split_point): Add count.
873 (consider_split): Do not compute incoming frequency; compute incoming
874 count and store it to split_point.
875 (split_function): Set count of the call to split part correctly.
877 2017-11-13 Carl Love <cel@us.ibm.com>
879 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
881 2017-11-13 Tom Tromey <tom@tromey.com>
883 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
885 2017-11-13 Carl Love <cel@us.ibm.com>
887 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
888 Add support for builtins:
889 unsigned int vec_first_{,miss}_match_{,or_eos}index,
890 vector {un,}signed {char,int,short},
891 vector {un,}signed {char,int,short}) arguments.
892 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
893 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
894 Add BU_P9V_AV_2 expansions for the builtins.
895 * config/rs6000/altivec.h (vec_first_match_index,
896 vec_first_mismatch_index, vec_first_match_or_eos_index,
897 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
898 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
899 new extern declaration.
900 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
901 * config/rs6000/vsx.md (first_match_index_<mode>,
902 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
903 first_mismatch_or_eos_index_<mode>): Add define expand.
904 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
905 * doc/extend.texi: Update the built-in documenation file for the new
908 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
910 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
911 into the min/max operations for _Float<N> and _Float<N>X types.
913 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
916 * dwarf2out.c (do_eh_frame): New static variable.
917 (dwarf2out_begin_prologue): Set it.
918 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
920 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
922 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
924 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
926 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
929 * debug.h (dwarf2out_do_eh_frame): Declare.
930 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
931 (dwarf2out_do_frame): Use it.
932 (dwarf2out_do_cfi_asm): Likewise.
933 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
934 (dwarf2out_assembly_start): Likewise.
935 (dwarf2out_begin_prologue): Fix comment.
936 * toplev.c (compile_file): Always call dwarf2out_frame_finish
937 if the target needs either debug or unwind DWARF2 info.
938 * lto-opts.c (lto_write_options): Do not save -fexceptions,
939 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
940 -ftrapping-math, -ftrapv and -fwrapv.
942 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
944 * cgraph.c (cgraph_edge::sreal_frequency): New function.
945 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
946 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
947 (estimate_edge_size_and_time): Likewise.
948 (ipa_merge_fn_summary_after_inlining): Likewise.
949 * ipa-inline.c (cgraph_freq_base_rec): Remove.
950 (compute_uninlined_call_time): Use sreal_frequency.
951 (compute_inlined_call_time): Likewise.
952 (ipa_inline): Do not initialize cgraph_freq_base_rec.
953 * profile-count.c: Include sreal.h.
954 (profile_count::to_sreal_scale): New.
955 * profile-count.h: Forward declare sreal.
956 (profile_count::to_sreal_scale): Declare.
958 2017-11-13 Nathan Sidwell <nathan@acm.org>
960 * diagnostic.c (maybe_line_and_column): New.
961 (diagnostic_get_location_text): Use it.
962 (diagnostic_report_current_module): Likewise.
963 (test_diagnostic_get_location_text): Add tests.
965 2017-11-13 Luis Machado <luis.machado@linaro.org>
967 * doc/md.texi (Specifying processor pipeline description): Fix
968 incorrect latency for the div instruction example.
970 2017-11-13 Jakub Jelinek <jakub@redhat.com>
972 PR tree-optimization/78821
973 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
974 that bit_not_p is the same.
975 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
976 (split_group): Count precisely bit_not_p bits in each statement.
977 (invert_op): New function.
978 (imm_store_chain_info::output_merged_store): Use invert_op to
979 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
980 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
982 2017-11-13 Martin Liska <mliska@suse.cz>
984 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
985 (struct source_info): Likewise.
986 (add_branch_counts): Likewise.
987 (add_line_counts): Likewise.
988 (function_summary): Likewise.
989 (output_intermediate_line): Likewise.
990 (generate_results): Likewise.
992 2017-11-13 Martin Liska <mliska@suse.cz>
994 * gcov.c (struct block_info): Remove typedef for block_t.
995 (struct line_info): Likewise.
996 (line_info::has_block): Likewise.
997 (EXIT_BLOCK): Likewise.
1000 (get_cycles_count): Likewise.
1001 (process_file): Likewise.
1002 (read_graph_file): Likewise.
1003 (solve_flow_graph): Likewise.
1004 (find_exception_blocks): Likewise.
1005 (add_line_counts): Likewise.
1006 (accumulate_line_info): Likewise.
1007 (output_line_details): Likewise.
1009 2017-11-13 Martin Liska <mliska@suse.cz>
1011 * gcov.c (struct arc_info): Remove typedef for arc_t.
1012 (struct line_info): Likewise.
1013 (add_branch_counts): Likewise.
1014 (output_branch_count): Likewise.
1015 (function_info::~function_info): Likewise.
1016 (circuit): Likewise.
1017 (output_intermediate_line): Likewise.
1018 (read_graph_file): Likewise.
1019 (solve_flow_graph): Likewise.
1020 (find_exception_blocks): Likewise.
1021 (add_line_counts): Likewise.
1022 (accumulate_line_info): Likewise.
1023 (output_line_details): Likewise.
1024 (output_function_details): Likewise.
1026 2017-11-13 Martin Liska <mliska@suse.cz>
1028 * gcov.c (struct function_info): Remove typedef for function_t.
1029 (struct source_info): Likewise.
1030 (source_info::get_functions_at_location): Likewise.
1031 (solve_flow_graph): Likewise.
1032 (find_exception_blocks): Likewise.
1033 (add_line_counts): Likewise.
1034 (output_intermediate_file): Likewise.
1035 (process_file): Likewise.
1036 (generate_results): Likewise.
1037 (release_structures): Likewise.
1038 (read_graph_file): Likewise.
1039 (read_count_file): Likewise.
1040 (accumulate_line_counts): Likewise.
1041 (output_lines): Likewise.
1043 2017-11-13 Martin Liska <mliska@suse.cz>
1045 * gcov.c (function_info::function_info): Remove num_counts
1046 and add vector<gcov_type>.
1047 (function_info::~function_info): Use the vector.
1048 (process_file): Likewise.
1049 (read_graph_file): Likewise.
1050 (read_count_file): Likewise.
1051 (solve_flow_graph): Likewise.
1053 2017-11-13 Martin Liska <mliska@suse.cz>
1055 * gcov.c (function_info::is_artificial): New function.
1056 (process_file): Erase all artificial early.
1057 (generate_results): Skip as all artificial are already
1060 2017-11-13 Martin Liska <mliska@suse.cz>
1062 * gcov.c (read_graph_file): Store to global vector of functions.
1063 (read_count_file): Iterate the vector.
1064 (process_file): Likewise.
1065 (generate_results): Likewise.
1066 (release_structures): Likewise.
1068 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1070 PR tree-optimization/82954
1071 * gimple-ssa-store-merging.c
1072 (imm_store_chain_info::coalesce_immediate_stores): If
1073 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1075 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1077 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1078 Upddate call to ENDIAN_LANE_N.
1079 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1080 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1081 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1082 and use aarch64_endian_lane_rtx.
1083 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1085 2017-11-12 Tom de Vries <tom@codesourcery.com>
1087 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1089 2017-11-12 Tom de Vries <tom@codesourcery.com>
1091 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1093 2017-11-12 Tom de Vries <tom@codesourcery.com>
1095 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1097 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1098 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1099 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1101 2017-11-11 Martin Sebor <msebor@redhat.com>
1104 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1107 * prefic.c (translate_name): Replace strncpy with memcpy to
1108 avoid -Wstringop-truncation.
1110 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1112 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1114 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1116 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1118 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1120 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1123 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1125 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1126 enable generating XXBRH if the value is in a vector register.
1127 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1128 value is in a vector register.
1129 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1130 register to register bswap64's instead of doing the GPR sequence
1131 used on previous machines.
1132 (bswapdi2_xxbrd): New insn.
1133 (bswapdi2_reg): Disallow on ISA 3.0.
1134 (register to register bswap64 splitter): Do not split the insn on
1135 ISA 3.0 systems that use XXBRD.
1137 2017-11-10 Martin Sebor <msebor@redhat.com>
1140 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1142 (find_subframework_file): Same.
1144 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1146 * auto-profile.c (afdo_indirect_call): Drop frequency.
1147 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1148 (cgraph_node::create_edge): Drop frequency argument.
1149 (cgraph_node::create_indirect_edge): Drop frequency argument.
1150 (cgraph_edge::make_speculative): Drop frequency arguments.
1151 (cgraph_edge::resolve_speculation): Do not update frequencies
1152 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1153 (cgraph_node::dump): Check consistency in IPA mode.
1154 (cgraph_edge::maybe_hot_p): Use IPA counter.
1155 (cgraph_edge::verify_count_and_frequency): Rename to ...
1156 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1157 (cgraph_node::verify_node): Update.
1158 * cgraph.h (struct cgraph_edge): Drop frequency.
1159 (cgraph_edge::frequency): New function.
1160 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1162 (cgraph_edge::rebuild_edges): Likewise.
1163 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1164 (duplicate_thunk_for_node): Do not pass frequency.
1165 (cgraph_node::create_clone): Scale only counts.
1166 (cgraph_node::create_virtual_clone): Do not pass frequency.
1167 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1168 (cgraph_node::create_version_clone): Do not pass frequency.
1169 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1170 (cgraph_node::expand_thunk): Do not pass frequency.
1171 (cgraph_node::create_wrapper): Do not pass frequency.
1172 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1174 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1175 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1176 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1177 (gather_caller_stats): Use frequency function.
1178 (ipcp_cloning_candidate_p): Use frequency function.
1179 (ipcp_propagate_stage): Use frequency function.
1180 (get_info_about_necessary_edges): Use frequency function.
1181 (update_profiling_info): Update only IPA profile.
1182 (update_specialized_profile): Use frequency functoin.
1183 (perhaps_add_new_callers): Update only IPA profile.
1184 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1185 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1186 (dump_ipa_call_summary): Use frequency function.
1187 (estimate_edge_size_and_time): Use frequency function.
1188 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1189 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1190 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1191 (update_noncloned_counts): ... ths one; scale counts only.
1192 (clone_inlined_nodes): Do not scale frequency.
1193 (inline_call): Do not pass frequency.
1194 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1195 (compute_inlined_call_time): Use IPA profile.
1196 (want_inline_small_function_p): Use IPA profile.
1197 (want_inline_self_recursive_call_p): Use IPA profile.
1198 (edge_badness): Use IPA profile.
1199 (lookup_recursive_calls): Use IPA profile.
1200 (recursive_inlining): Do not pass frequency.
1201 (resolve_noninline_speculation): Do not update frequency.
1202 (inline_small_functions): Collect max of IPA profile.
1203 (dump_overall_stats): Dump IPA porfile.
1204 (dump_inline_stats): Dump IPA porfile.
1205 (ipa_inline): Collect IPA stats.
1206 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1207 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1208 (ipa_propagate_frequency): Use frequency function.
1209 (ipa_profile): Cleanup.
1210 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1211 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1212 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1213 (input_node): Do not stream frequency.
1214 (input_edge): Do not stream frequency.
1215 (merge_profile_summaries): Scale only IPA profiles.
1216 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1217 * predict.c (drop_profile): Do not recompute frequency.
1218 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1219 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1220 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1221 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1222 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1223 * tree-inline.c (copy_bb): Do not scale frequency.
1224 (expand_call_inline): Do not scale frequency.
1225 (tree_function_versioning): Do not scale frequency.
1226 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1228 2017-11-10 Julia Koval <julia.koval@intel.com>
1230 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1231 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1232 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1233 (_mm256_maskz_gf2p8affine_epi64_epi8)
1234 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1235 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1236 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1237 (__builtin_ia32_vgf2p8affineqb_v32qi)
1238 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1239 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1241 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1245 (arm_option_override): Refactor.
1246 (arm_option_reconfigure_globals): New.
1247 (arm_options_perform_arch_sanity_checks): New.
1248 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1250 (arm_options_perform_arch_sanity_checks): Likewise
1252 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1254 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1255 (power9-qpmul): New.
1256 * rs6000/rs6000.md ("type" attr): Add qmul.
1257 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1258 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1259 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1261 2017-11-10 Martin Sebor <msebor@redhat.com>
1264 * builtins.c (compute_objsize): Handle arrays that
1265 compute_builtin_object_size likes to fail for. Make extern.
1266 * builtins.h (compute_objsize): Declare.
1267 (check_strncpy_sizes): New function.
1268 (expand_builtin_strncpy): Call check_strncpy_sizes.
1269 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1270 -Wstringop-truncation.
1271 (gimple_fold_builtin_strncat): Same.
1272 * gimple.c (gimple_build_call_from_tree): Set call location.
1273 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1274 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1275 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1276 (handle_builtin_strlen): Use strlen_to_stridx.
1277 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1279 Use strlen_to_stridx.
1280 (pass_strlen::execute): Release strlen_to_stridx.
1281 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1282 (-Wstringop-truncation): Document new option.
1284 2017-11-10 Martin Liska <mliska@suse.cz>
1286 PR gcov-profile/82702
1287 * gcov.c (main): Handle intermediate files in a different
1289 (get_gcov_intermediate_filename): New function.
1290 (output_gcov_file): Remove support of intermediate files.
1291 (generate_results): Allocate intermediate file.
1292 (release_structures): Clean-up properly fn_end.
1293 (output_intermediate_file): Start iterating with line 1.
1295 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1297 PR tree-optimization/82929
1298 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1299 ops_swapped_p non-static data member.
1300 (store_immediate_info::store_immediate_info): Clear it.
1301 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1302 ops set ops_swapped_p.
1303 (count_multiple_uses): Handle ops_swapped_p.
1305 2017-11-10 Martin Liska <mliska@suse.cz>
1307 * coverage.c (coverage_init): Stream information about
1308 support of has_unexecuted_blocks.
1309 * doc/gcov.texi: Document that.
1310 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1311 * gcov.c (read_graph_file): Likewise.
1312 (output_line_beginning): Fix a small issue with
1315 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1317 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1318 reference of trivial component.
1320 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1323 * gimple-ssa-store-merging.c
1324 (pass_store_merging::terminate_all_aliasing_chains): For
1325 gimple_store_p stmts also call refs_output_dependent_p.
1327 PR rtl-optimization/82913
1328 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1331 2017-11-09 Jeff Law <law@redhat.com>
1333 * vr-values.h: New file with vr_values class.
1334 * tree-vrp.c: Include vr-values.h
1335 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1336 data objects into the vr_values class.
1337 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1338 (get_value_range): Make it a member function within vr_values class.
1339 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1340 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1341 (op_with_constant_singleton_value_range): Likewise.
1342 (extract_range_for_var_from_comparison_expr): Likewise.
1343 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1344 (extract_range_from_binary_expr): Likewise.
1345 (extract_range_from_unary_expr): Likewise.
1346 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1347 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1348 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1349 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1350 (compare_name_with_value, compare_names): Likewise.
1351 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1352 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1353 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1354 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1355 (extract_range_from_phi_node): Likewise.
1356 (simplify_truth_ops_using_ranges): Likewise.
1357 (simplify_div_or_mod_using_ranges): Likewise.
1358 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1359 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1360 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1361 (simplify_float_conversion_using_ranges): Likewise.
1362 (simplify_internal_call_using_ranges): Likewise.
1363 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1364 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1365 poitner via x_vr_values for calls into gimple folder.
1366 (vrp_initialize_lattice): Make this the vr_values ctor.
1367 (vrp_free_lattice): Make this the vr_values dtor.
1368 (set_vr_value): New function.
1369 (class vrp_prop): Add vr_values data member. Add various member
1370 functions as well as member functions that delegate to vr_values.
1371 (check_array_ref): Make a member function within vrp_prop class.
1372 (search_for_addr_array, vrp_initialize): Likewise.
1373 (vrp_finalize): Likewise. Revamp to avoid direct access to
1374 vr_value, values_propagated, etc.
1375 (check_array_bounds): Extract vrp_prop class instance pointer from
1376 walk info structure. Use it to call member functions.
1377 (check_all_array_refs): Make a member function within vrp_prop class.
1378 Smuggle class instance pointer via walk info structure.
1379 (x_vr_values): New local static.
1380 (vrp_valueize): Use x_vr_values to get class instance.
1381 (vr_valueize_1): Likewise.
1382 (class vrp_folder): Add vr_values data member. Add various member
1383 functions as well as member functions that delegate to vr_values.
1384 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1385 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1386 class instance from vr_values. Use it to call member functions.
1387 (vrp_dom_walker): Add vr_values data member.
1388 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1389 instance via x_vr_values.
1390 (identify_jump_threads): Accept vr_values as argument. Store
1391 it into the walker structure.
1392 (evrp_dom_walker): Add vr_values class data member. Add various
1394 (evrp_dom_walker::try_find_new_range): Use vr_values data
1395 member to access the memory allocator.
1396 (evrp_dom_walker::before_dom_children): Store vr_values class
1397 instance into the vrp_folder class.
1398 (evrp_dom_walker::push_value_range): Rework to avoid direct
1399 access to num_vr_values and vr_value.
1400 (evrp_dom_walker::pop_value_range): Likewise.
1401 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1402 Use vr_values to get to dump_all_value_ranges member function.
1403 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1404 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1405 Pass vr_values class instance down to identify_jump_threads.
1406 Remove call to vrp_free_lattice.
1407 (debug_all_value_ranges): Remove.
1409 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1410 (vrp_folder): Likewise.
1412 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1413 Get it from the existing bitmap instead.
1414 (vrp_intersect_ranges_1): Likewise.
1416 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1418 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1420 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1421 set bit_not_p to it.
1422 (imm_store_chain_info::coalesce_immediate_stores): Break group
1423 if bit_not_p is different.
1424 (count_multiple_uses, split_group,
1425 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1426 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1427 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1428 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1429 to store_immediate_info ctor.
1431 2017-11-09 Jim Wilson <jimw@sifive.com>
1433 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1434 (scan_prog_file): Likewise.
1436 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1438 * bb-reorder.c (max_entry_frequency): Remove.
1439 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1440 connect_traces, push_to_next_round_p): Remove prototypes.
1441 (find_traces_1_round): Use counts only.
1442 (push_to_next_round_p): Likewise.
1443 (find_traces): Likewise.
1444 (rotate_loop): Likewise.
1445 (find_traces_1_round): Likewise.
1446 (connect_traces): Likewise.
1447 (edge_order): Likewise.
1449 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1451 * config/arm/arm.c (output_return_instruction): Add comments to
1452 indicate requirement for cmse_nonsecure_entry return to account
1453 for the size of clearing instruction output here.
1454 (thumb_exit): Likewise.
1455 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1456 return in hardfloat mode.
1458 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1460 * config/rs6000/rs6000.c (machine_function): Add a bool,
1461 "toc_is_wrapped_separately".
1462 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1463 if it wasn't explicitly set or unset, we are optimizing for speed, and
1464 doing separate shrink-wrapping.
1465 (rs6000_get_separate_components): Enable the TOC component if
1466 saving the TOC register in the prologue.
1467 (rs6000_components_for_bb): Handle the TOC component.
1468 (rs6000_emit_prologue_components): Store the TOC register where needed.
1469 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1470 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1472 2017-11-09 Martin Jambor <mjambor@suse.cz>
1474 * ipa-param-manipulation.c: New file.
1475 * ipa-param-manipulation.h: Likewise.
1476 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1477 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1478 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1479 (ipa_parm_adjustment): Likewise.
1480 (ipa_parm_adjustment_vec): Likewise.
1481 (ipa_get_vector_of_formal_parms): Moved declaration to
1482 ipa-param-manipulation.h.
1483 (ipa_get_vector_of_formal_parm_types): Likewise.
1484 (ipa_modify_formal_parameters): Likewise.
1485 (ipa_modify_call_arguments): Likewise.
1486 (ipa_combine_adjustments): Likewise.
1487 (ipa_dump_param_adjustments): Likewise.
1488 (ipa_modify_expr): Likewise.
1489 (ipa_get_adjustment_candidate): Likewise.
1490 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1491 ipa-param-manipulation.c.
1492 (ipa_get_vector_of_formal_parm_types): Likewise.
1493 (ipa_modify_formal_parameters): Likewise.
1494 (ipa_modify_call_arguments): Likewise.
1495 (ipa_modify_expr): Likewise.
1496 (get_ssa_base_param): Likewise.
1497 (ipa_get_adjustment_candidate): Likewise.
1498 (index_in_adjustments_multiple_times_p): Likewise.
1499 (ipa_combine_adjustments): Likewise.
1500 (ipa_dump_param_adjustments): Likewise.
1501 * tree-sra.c: Also include ipa-param-manipulation.h
1502 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1505 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * doc/sourcebuild.texi (vect_masked_store): Document.
1511 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1512 Alan Hayward <alan.hayward@arm.com>
1513 David Sherwood <david.sherwood@arm.com>
1515 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1517 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1518 Alan Hayward <alan.hayward@arm.com>
1519 David Sherwood <david.sherwood@arm.com>
1521 * doc/sourcebuild.texi (vect_variable_length): Document.
1523 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1524 Alan Hayward <alan.hayward@arm.com>
1525 David Sherwood <david.sherwood@arm.com>
1527 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1529 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1530 Alan Hayward <alan.hayward@arm.com>
1531 David Sherwood <david.sherwood@arm.com>
1533 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1535 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1536 Alan Hayward <alan.hayward@arm.com>
1537 David Sherwood <david.sherwood@arm.com>
1539 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1540 previously undocumented selectors.
1541 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1543 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1545 * doc/rtl.texi (const_vector): Say that elements can be
1546 const_wide_ints too.
1547 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1548 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1549 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1550 * optabs.c (expand_vector_broadcast): Likewise.
1552 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1553 Alan Hayward <alan.hayward@arm.com>
1554 David Sherwood <david.sherwood@arm.com>
1556 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1557 scaled index even if the unscaled address was invalid.
1558 Don't increase the complexity of using a scale in that case.
1560 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1561 Alan Hayward <alan.hayward@arm.com>
1562 David Sherwood <david.sherwood@arm.com>
1564 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1565 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1566 UNITS_PER_WORD bytes.
1567 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1568 based on REGMODE_NATURAL_SIZE of the inner mode.
1569 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1570 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1571 * expmed.c (lowpart_bit_field_p): Divide the value up into
1572 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1573 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1574 whether something is likely to occupy more than one register.
1576 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1579 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1580 profile_count::adjust_for_ipa_scaling.
1581 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1582 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1584 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1586 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1588 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1589 (split_group): Add total_orig and total_new arguments, estimate the
1590 number of statements related to the store group without store merging
1591 and with store merging.
1592 (imm_store_chain_info::output_merged_store): Adjust split_group
1593 callers, punt if estimated number of statements with store merging
1594 is not smaller than estimated number of statements without it.
1596 (handled_load): Remove has_single_use checks.
1597 (pass_store_merging::process_store): Likewise.
1599 2017-11-09 Richard Biener <rguenther@suse.de>
1601 PR tree-optimization/82902
1602 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1604 2017-11-09 Martin Liska <mliska@suse.cz>
1607 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1610 2017-11-09 Martin Liska <mliska@suse.cz>
1612 PR tree-optimization/82669
1613 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1615 2017-11-09 Martin Liska <mliska@suse.cz>
1617 PR gcov-profile/48463
1618 * coverage.c (coverage_begin_function): Output also end locus
1619 of a function and information whether the function is
1621 * gcov-dump.c (tag_function): Parse and print the information.
1622 * gcov.c (INCLUDE_MAP): Add include.
1623 (INCLUDE_SET): Likewise.
1624 (struct line_info): Move earlier in the source file because
1625 of vector<line_info> in function_info structure.
1626 (line_info::line_info): Likewise.
1627 (line_info::has_block): Likewise.
1628 (struct source_info): Add new member index.
1629 (source_info::get_functions_at_location): New function.
1630 (function_info::group_line_p): New function.
1631 (output_intermediate_line): New function.
1632 (output_intermediate_file): Use the mentioned function.
1633 (struct function_start): New.
1634 (struct function_start_pair_hash): Likewise.
1635 (process_file): Add code that identifies group functions.
1636 Assign lines either to global or function scope.
1637 (generate_results): Skip artificial functions.
1638 (find_source): Assign index for each source file.
1639 (read_graph_file): Read new flag artificial and end_line.
1640 (add_line_counts): Assign it either to global of function scope.
1641 (accumulate_line_counts): Isolate core of the function to
1642 accumulate_line_info and call it for both function and global
1644 (accumulate_line_info): New function.
1645 (output_line_beginning): Fix GNU coding style.
1646 (print_source_line): New function.
1647 (output_line_details): Likewise.
1648 (output_function_details): Likewise.
1649 (output_lines): Iterate both source (global) scope and function
1651 (struct function_line_start_cmp): New class.
1652 * doc/gcov.texi: Reflect changes in documentation.
1654 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1657 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1658 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1659 and similarly for not instead of neg.
1661 2017-11-08 Andi Kleen <ak@linux.intel.com>
1663 * config/i386/i386.opt: Add -mforce-indirect-call.
1664 * config/i386/predicates.md: Check for flag_force_indirect_call.
1665 * doc/invoke.texi: Document -mforce-indirect-call
1667 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1669 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1671 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1672 config/riscv/riscv.c (predict.h): New include.
1673 (riscv_slow_unaligned_access_p): No longer static.
1674 (riscv_block_move_straight): Add require.
1675 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1677 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1680 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1681 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1682 nonimmediate_operand predicate for operand 1 instead of
1685 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1687 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1689 * config/aarch64/constraints.md (Uml): New constraint.
1690 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1693 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1695 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1696 of two vec_duplicates into a vec_concat.
1698 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1700 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1701 Simplify vec_merge of vec_duplicate and vec_concat.
1702 * config/aarch64/constraints.md (Utq): New constraint.
1703 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1706 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1708 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1709 Simplify vec_merge of vec_duplicate and const_vector.
1710 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1712 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1713 mode iterator. Update predicate on operand 1 to
1714 handle non-const_vec constants. Delete constraints.
1715 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1717 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1719 PR tree-optimization/78821
1720 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1722 (store_operand_info::store_operand_info): Initialize it to false.
1723 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1724 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1725 store in the group, and if chain_info is non-NULL, to ignore altogether
1727 (compatible_load_p): Fail if bit_not_p does not match.
1728 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1729 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1730 (pass_store_merging::process_store): Adjust
1731 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1732 call terminate_all_aliasing_chains newly when adding a store into
1733 a chain with non-NULL chain_info.
1735 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1737 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1739 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1741 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1743 (aarch64_layout_frame): Initialise emit_frame_chain.
1744 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1745 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1747 2017-11-08 Martin Liska <mliska@suse.cz>
1749 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1750 of gimple_call_internal_p.
1752 2017-11-07 Tom de Vries <tom@codesourcery.com>
1754 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1756 2017-11-07 Tom de Vries <tom@codesourcery.com>
1758 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1761 2017-11-08 Martin Liska <mliska@suse.cz>
1764 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1766 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1768 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1770 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1773 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1774 (*cmp<mode>_ccz_1): New insn with $k alternative.
1777 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1778 RTX_COMM_COMPARE as commutative as well.
1779 (ix86_binary_operator_ok): Formatting fix.
1780 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1781 *<code><mode>3<mask_name><round_saeonly_name>,
1782 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1783 *<s>mul<mode>3_highpart<mask_name>,
1784 *vec_widen_umult_even_v16si<mask_name>,
1785 *vec_widen_umult_even_v8si<mask_name>,
1786 *vec_widen_umult_even_v4si<mask_name>,
1787 *vec_widen_smult_even_v16si<mask_name>,
1788 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1789 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1790 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1791 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1792 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1793 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1794 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1795 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1796 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1797 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1798 ix86_binary_operator_ok. Formatting fixes.
1799 (*<plusminus_insn><mode>3<mask_name><round_name>,
1800 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1803 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1805 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1806 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1807 eq and ne if TARGET_ISEL.
1808 (cmp): New code_iterator.
1809 (UNS, UNSU_, UNSIK): New code_attrs.
1810 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1811 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1813 ("eq<mode>3"): ... this.
1814 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1816 ("ne<mode>3"): ... this.
1818 2017-11-07 Julia Koval <julia.koval@intel.com>
1821 * common/config/i386/i386-common.c
1822 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1823 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1824 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1825 * config/i386/i386.opt: Ditto.
1826 * config/i386/i386.c (ix86_target_string): Ditto.
1827 (ix86_option_override_internal): Ditto.
1828 (ix86_init_mpx_builtins): Move MPX to args2.
1829 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1830 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1831 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1832 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1833 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1834 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1835 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1837 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1840 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1841 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1842 (zero-extendsidi peephole2): Remove peephole.
1844 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1847 * stor-layout.c: Include attribs.h.
1848 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1849 explicit lookup of "aligned" attribute.
1851 2017-11-07 Andrew Waterman <andrew@sifive.com>
1853 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1854 (riscv_expand_block_move): Likewise.
1855 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1857 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1858 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1859 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1860 (riscv_adjust_block_mem): Likewise.
1861 (riscv_block_move_loop): Likewise.
1862 (riscv_expand_block_move): Likewise.
1863 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1865 2017-11-07 Michael Clark <michaeljclark@mac.com>
1867 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1868 (MUSL_DYNAMIC_LINKER): Likewise.
1870 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1872 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1873 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1876 2017-11-07 Richard Biener <rguenther@suse.de>
1878 * match.pd: Fix build.
1880 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1881 Jackson Woodruff <jackson.woodruff@arm.com>
1883 PR tree-optimization/71026
1884 * match.pd: Canonicalize negate in division.
1886 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1889 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1891 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1893 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1894 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1896 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1898 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1899 non-scalar integral types.
1900 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1901 (-(A-B), -(~A)): New transformations.
1903 2017-11-07 Tom de Vries <tom@codesourcery.com>
1905 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1906 semicolon after "do {} while (0)".
1907 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1908 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1909 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1910 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1911 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1913 2017-11-07 Tom de Vries <tom@codesourcery.com>
1915 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1916 after "do {} while (0)".
1917 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1918 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1919 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1920 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1921 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1923 2017-11-07 Tom de Vries <tom@codesourcery.com>
1926 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1928 (arm_rtx_costs_internal): Add missing semicolon after
1929 HANDLE_NARROW_SHIFT_ARITH call.
1931 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1933 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1934 disable isel if it was not set explicitly.
1936 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1938 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1940 (add_type_attribute) likewise.
1942 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1944 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1945 of struct ix86_frame.
1946 (ix86_initial_elimination_offset): Likewise.
1947 (ix86_expand_split_stack_prologue): Likewise.
1949 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1951 * tree-vrp.h (enum value_range_type): Update stale comment.
1953 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1954 Alan Hayward <alan.hayward@arm.com>
1955 David Sherwood <david.sherwood@arm.com>
1957 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1958 (aarch64_expand_vec_perm_const): Take the number of units too.
1959 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1960 (aarch64_expand_vec_perm_const): Likewise.
1961 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1962 (vec_perm<mode>): Update accordingly.
1964 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1965 Alan Hayward <alan.hayward@arm.com>
1966 David Sherwood <david.sherwood@arm.com>
1968 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1969 Take the number of units too.
1970 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1971 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1972 but check for a vector mode before rather than after the call.
1973 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1974 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1975 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1976 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1977 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1978 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1979 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1980 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1981 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1982 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1983 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1984 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1985 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1986 (aarch64_sqdmull2_n<mode>): Update accordingly.
1988 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1989 Alan Hayward <alan.hayward@arm.com>
1990 David Sherwood <david.sherwood@arm.com>
1992 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1993 the number of units too.
1994 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1995 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1996 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1997 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1998 (vec_store_lanesxi<mode>): Update accordingly.
2000 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2001 Alan Hayward <alan.hayward@arm.com>
2002 David Sherwood <david.sherwood@arm.com>
2004 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2005 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2006 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2007 of units rather than the mode.
2008 * config/aarch64/iterators.md (nunits): New mode attribute.
2009 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2010 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2011 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2012 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2013 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2014 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2015 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2016 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2017 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2018 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2019 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2020 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2021 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2022 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2023 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2024 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2025 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2026 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2027 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2028 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2029 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2030 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2031 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2032 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2033 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2034 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2035 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2036 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2037 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2038 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2039 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2040 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2041 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2042 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2043 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2044 (aarch64_simd_vec_setv2di): Likewise.
2046 2017-11-06 Carl Love <cel@us.ibm.com>
2048 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2050 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2051 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2053 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2055 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2056 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2057 Add power 8 macro expansions.
2058 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2059 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2060 power 8 instructions. (VSX_XXBR): Add iterator.
2062 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2064 * config/arm/arm.md (predicable_short_it): Change default to "no",
2065 improve documentation, remove uses that are identical to the default.
2066 (enabled_for_depr_it): Rename to enabled_for_short_it.
2067 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2068 Remove default uses.
2069 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2070 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2071 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2072 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2074 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2077 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2078 float128 helper macros, which are no longer used after deleting
2079 the old 'q' built-in functions, and moving the round to odd
2080 built-in functions to being special built-in functions.
2081 (BU_FLOAT128_2): Likewise.
2082 (BU_FLOAT128_1_HW): Likewise.
2083 (BU_FLOAT128_2_HW): Likewise.
2084 (BU_FLOAT128_3_HW): Likewise.
2085 (FABSQ): Delete old 'q' built-in functions.
2086 (COPYSIGNQ): Likewise.
2087 (SQRTF128_ODD): Move round to odd built-in functions to be
2088 special built-in functions, so that we can handle
2089 -mabi=ieeelongdouble.
2090 (TRUNCF128_ODD): Likewise.
2091 (ADDF128_ODD): Likewise.
2092 (SUBF128_ODD): Likewise.
2093 (MULF128_ODD): Likewise.
2094 (DIVF128_ODD): Likewise.
2095 (FMAF128_ODD): Likewise.
2096 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2097 built-in names to 'f128'.
2098 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2099 old 'q' built-in functions, as the machine independent code for
2100 'f128' built-in functions handles this.
2101 (rs6000_expand_builtin): Add expansion for float128 round to odd
2102 functions, keying off on -mabi=ieeelongdouble of whether to use
2103 the KFmode or TFmode variant.
2104 (rs6000_init_builtins): Initialize the _Float128 round to odd
2106 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2107 _Float128 'q' built-in functions are now mapped into the new
2108 'f128' built-in functions.
2110 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2112 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2113 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2115 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2118 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2119 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2120 * toplev.c: Include "ipa-fnsummary.h".
2121 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2123 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2125 PR tree-optimization/82838
2126 * gimple-ssa-store-merging.c
2127 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2128 on a separate gimple_seq which is then appended to seq.
2130 2017-11-06 Jeff Law <law@redhat.com>
2133 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2134 (get_probe_interval): New functions.
2135 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2136 (ix86_adjust_stack_and_probe): Likewise.
2137 (output_adjust_stack_and_probe): Likewise.
2138 (ix86_emit_probe_stack_range): Likewise.
2139 (ix86_expand_prologue): Likewise.
2141 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2143 PR tree-optimization/82816
2144 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2145 if the modes of the two types are the same.
2146 (convert_plusminus_to_widen): Likewise.
2148 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2150 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2152 (usadv16qi): New define_expand.
2153 (usadv8hi): New define_expand.
2155 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2158 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2160 (inline_transform): Likewise.
2161 * predict.c (counts_to_freqs): Remove useless conditional.
2162 * profile-count.h (profile_count::apply_scale): Move sanity check.
2163 * tree-inline.c (copy_bb): Always scale.
2164 (copy_cfg_body): Likewise.
2166 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2169 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2171 (*cmp_ite0): Add enabled_for_depr_it attribute.
2172 (*cmp_ite1): Likewise.
2174 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2176 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2179 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2181 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2182 return the value by pointer.
2183 (extract_range_from_multiplicative_op_1): Update accordingly.
2184 Return as soon as an operation fails.
2186 2017-11-05 Tom de Vries <tom@codesourcery.com>
2189 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2190 (DEF_SANITIZER_BUILTIN): ... here.
2191 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2192 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2194 2017-11-05 Tom de Vries <tom@codesourcery.com>
2197 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2199 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2200 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2201 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2203 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2204 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2205 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2207 2017-11-05 Tom de Vries <tom@codesourcery.com>
2210 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2213 2017-11-04 Michael Clark <michaeljclark@mac.com>
2215 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2216 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2218 (*addsi3_extended): Likewise.
2219 (*addsi3_extended2): Likewise.
2220 (<optab>si3): Likewise.
2221 (<optab>di3): Likewise.
2222 (<optab><mode>3): Likewise.
2223 (<*optabe>si3_internal): Likewise.
2224 (zero_extendqi<SUPERQI:mode>2): Likewise.
2225 (*add<mode>hi3): Likewise.
2226 (*xor<mode>hi3): Likewise.
2227 (<optab>di3): Likewise.
2228 (*<optab>si3_extend): Likewise.
2229 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2230 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2231 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2233 2017-11-04 Andrew Waterman <andrew@sifive.com>
2235 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2236 TARGET_STRICT_ALIGN based upon -mtune argument.
2238 2017-11-04 Andrew Waterman <andrew@sifive.com>
2240 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2242 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2244 config/i386/i386.c (choose_basereg): Use optional scratch
2245 register and add assertion.
2246 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2247 needed, and don't allocate stack.
2248 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2249 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2250 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2252 2017-11-03 Jeff Law <law@redhat.com>
2254 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2255 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2256 to probe at the start of a noreturn function.
2258 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2260 PR tree-optimization/78821
2261 * gimple-ssa-store-merging.c: Update the file comment.
2262 (MAX_STORE_ALIAS_CHECKS): Define.
2263 (struct store_operand_info): New type.
2264 (store_operand_info::store_operand_info): New constructor.
2265 (struct store_immediate_info): Add rhs_code and ops data members.
2266 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2267 and op1r arguments to the ctor, initialize corresponding data members.
2268 (struct merged_store_group): Add load_align_base and load_align
2270 (merged_store_group::merged_store_group): Initialize them.
2271 (merged_store_group::do_merge): Update them.
2272 (merged_store_group::apply_stores): Pick the constant for
2273 encode_tree_to_bitpos from one of the two operands, or skip
2274 encode_tree_to_bitpos if neither operand is a constant.
2275 (class pass_store_merging): Add process_store method decl. Remove
2276 bool argument from terminate_all_aliasing_chains method decl.
2277 (pass_store_merging::terminate_all_aliasing_chains): Remove
2278 var_offset_p argument and corresponding handling.
2279 (stmts_may_clobber_ref_p): New function.
2280 (compatible_load_p): New function.
2281 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2282 if there is overlap and rhs_code is not INTEGER_CST. For
2283 non-overlapping stores terminate group if rhs is not mergeable.
2284 (get_alias_type_for_stmts): Change first argument from
2285 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2286 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2287 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2289 (get_location_for_stmts): Change first argument from
2290 auto_vec<gimple *> & to vec<gimple *> &.
2291 (struct split_store): Remove orig_stmts data member, add orig_stores.
2292 (split_store::split_store): Create orig_stores rather than orig_stmts.
2293 (find_constituent_stmts): Renamed to ...
2294 (find_constituent_stores): ... this. Change second argument from
2295 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2296 to info structures rather than the statements.
2297 (split_group): Rename ALLOW_UNALIGNED argument to
2298 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2299 it. Adjust find_constituent_stores caller.
2300 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2301 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2302 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2303 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2304 (mem_valid_for_store_merging): New function.
2305 (handled_load): New function.
2306 (pass_store_merging::process_store): New method.
2307 (pass_store_merging::execute): Use process_store method. Adjust
2308 terminate_all_aliasing_chains caller.
2310 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2312 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2313 Return true for more constants, symbols and label references.
2314 (aarch64_valid_floating_const): Remove unused function.
2316 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2319 * config/aarch64/aarch64.c (aarch64_layout_frame):
2320 Undo forcing of LR at bottom of frame.
2322 2017-11-03 Jeff Law <law@redhat.com>
2325 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2326 for int_registers_saved.
2328 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2329 extracted from tree-ssa-dom.c.
2330 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2331 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2332 (record_equivalences_from_incoming_edge): Add additional argument
2333 to single_pred_edge_ignoring_loop_edges call.
2334 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2335 (uncprop_dom_walker::before_dom_children): Add additional argument
2336 to single_pred_edge_ignoring_loop_edges call.
2337 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2338 single_pred_edge_ignoring_loop_edges rather than open coding.
2339 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2341 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2343 * match.pd (-(-A)): Rewrite.
2345 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2347 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2348 (rs6000_emit_int_cmove): New declaration.
2349 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2350 (rs6000_emit_sISEL): Delete.
2351 (rs6000_emit_int_cmove): Make non-static.
2352 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2353 instead of rs6000_emit_sISEL.
2355 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2357 * asan.c (create_cond_insert_point): Maintain profile.
2358 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2360 * basic-block.h (struct basic_block_def): Remove frequency.
2361 (EDGE_FREQUENCY): Use to_frequency
2362 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2364 (find_traces): Update to use to_frequency.
2365 (find_traces_1_round): Likewise; use only IPA counts.
2366 (bb_to_key): Likewise.
2367 (connect_traces): Use IPA counts only.
2368 (copy_bb_p): Update to use to_frequency.
2369 (fix_up_crossing_landing_pad): Likewise.
2370 (sanitize_hot_paths): Likewise.
2371 * bt-load.c (basic_block_freq): Likewise.
2372 * cfg.c (init_flow): Set count_max to uninitialized.
2373 (check_bb_profile): Remove frequencies; check counts.
2374 (dump_bb_info): Do not dump frequencies.
2375 (update_bb_profile_for_threading): Update counts only.
2376 (scale_bbs_frequencies_int): Likewise.
2377 (MAX_SAFE_MULTIPLIER): Remove.
2378 (scale_bbs_frequencies_gcov_type): Update counts only.
2379 (scale_bbs_frequencies_profile_count): Update counts only.
2380 (scale_bbs_frequencies): Update counts only.
2381 * cfg.h (struct control_flow_graph): Add count-max.
2382 (update_bb_profile_for_threading): Update prototype.
2383 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2384 (find_many_sub_basic_blocks): Likewise.
2385 * cfgcleanup.c (try_forward_edges): Likewise.
2386 (try_crossjump_to_edge): Likewise.
2387 * cfgexpand.c (expand_gimple_cond): Likewise.
2388 (expand_gimple_tailcall): Likewise.
2389 (construct_init_block): Likewise.
2390 (construct_exit_block): Likewise.
2391 * cfghooks.c (verify_flow_info): Check consistency of counts.
2392 (dump_bb_for_graph): Do not dump frequencies.
2393 (split_block_1): Do not update frequencies.
2394 (split_edge): Do not update frequencies.
2395 (make_forwarder_block): Do not update frequencies.
2396 (duplicate_block): Do not update frequencies.
2397 (account_profile_record): Do not update frequencies.
2398 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2399 for global heuristics.
2400 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2401 (expected_loop_iterations_unbounded): Use counts only.
2402 * cfgloopmanip.c (scale_loop_profile): Simplify.
2403 (create_empty_loop_on_edge): Simplify
2405 (duplicate_loop_to_header_edge): Simplify
2406 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2407 (update_br_prob_note): Take care of removing note when profile
2409 (relink_block_chain): Do not dump frequency.
2410 (rtl_account_profile_record): Use to_frequency.
2411 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2412 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2413 (cgraph_update_edges_for_call_stmt_node): Likewise.
2414 (cgraph_edge::verify_count_and_frequency): Update.
2415 (cgraph_node::verify_node): Temporarily disable frequency verification.
2416 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2417 to_cgraph_frequency.
2418 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2419 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2421 (cgraph_node::expand_thunk): Update profile.
2422 * except.c (dw2_build_landing_pads): Do not update frequency.
2423 * final.c (compute_alignments): Use to_frequency.
2424 (dump_basic_block_info): Do not dump frequency.
2425 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2426 (dump_gimple_bb_header): Do not dump frequency.
2427 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2429 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2430 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2431 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2432 (init_before_recovery): Do not update frequency.
2433 (sched_create_recovery_edges): Do not update frequency.
2434 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2435 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2436 (ipa_cp_c_finalize): Set max_count to uninitialized.
2437 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2438 (param_change_prob): Use counts.
2439 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2441 * ipa-split.c (consider_split): Use to_frequency.
2442 (split_function): Use to_frequency.
2443 * ira-build.c (loop_compare_func): Likewise.
2444 (mark_loops_for_removal): Likewise.
2445 (mark_all_loops_for_removal): Likewise.
2446 * loop-doloop.c (doloop_modify): Do not update frequency.
2447 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2449 * lto-streamer-in.c (input_function): Update count_max.
2450 * omp-expand.c (expand_omp_taskreg): Update count_max.
2451 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2452 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2453 (maybe_hot_count_p): Use ipa counts only.
2454 (maybe_hot_bb_p): Simplify.
2455 (maybe_hot_edge_p): Simplify.
2456 (probably_never_executed): Do not take frequency argument.
2457 (probably_never_executed_bb_p): Do not pass frequency.
2458 (probably_never_executed_edge_p): Likewise.
2459 (combine_predictions_for_bb): Check that profile is nonzero.
2460 (propagate_freq): Do not set frequency.
2461 (drop_profile): Simplify.
2462 (counts_to_freqs): Simplify.
2463 (expensive_function_p): Use to_frequency.
2464 (propagate_unlikely_bbs_forward): Simplify.
2465 (determine_unlikely_bbs): Simplify.
2466 (estimate_bb_frequencies): Add hack to silence graphite issues.
2467 (compute_function_frequency): Use ipa counts.
2468 (pass_profile::execute): Update.
2469 (rebuild_frequencies): Use counts only.
2470 (force_edge_cold): Use counts only.
2471 * profile-count.c (profile_count::dump): Dump new count types.
2472 (profile_count::differs_from_p): Check compatiblity.
2473 (profile_count::to_frequency): New function.
2474 (profile_count::to_cgraph_frequency): New function.
2475 * profile-count.h (struct function): Declare.
2476 (enum profile_quality): Add profile_guessed_local and
2477 profile_guessed_global0.
2478 (class profile_proability): Decrease number of bits to 29;
2479 update from_reg_br_prob_note and to_reg_br_prob_note.
2480 (class profile_count: Update comment; decrease number of bits
2481 to 61. Check compatibility.
2482 (profile_count::compatible_p): New private member function.
2483 (profile_count::ipa_p): New member function.
2484 (profile_count::operator<): Handle global zero correctly.
2485 (profile_count::operator>): Handle global zero correctly.
2486 (profile_count::operator<=): Handle global zero correctly.
2487 (profile_count::operator>=): Handle global zero correctly.
2488 (profile_count::nonzero_p): New member function.
2489 (profile_count::force_nonzero): New member function.
2490 (profile_count::max): New member function.
2491 (profile_count::apply_scale): Handle IPA scalling.
2492 (profile_count::guessed_local): New member function.
2493 (profile_count::global0): New member function.
2494 (profile_count::ipa): New member function.
2495 (profile_count::to_frequency): Declare.
2496 (profile_count::to_cgraph_frequency): Declare.
2497 * profile.c (OVERLAP_BASE): Delete.
2498 (compute_frequency_overlap): Delete.
2499 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2500 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2501 * sched-ebb.c (rank): Use counts only.
2502 * shrink-wrap.c (handle_simple_exit): Use counts only.
2503 (try_shrink_wrapping): Use counts only.
2504 (place_prologue_for_one_component): Use counts only.
2505 * tracer.c (find_best_predecessor): Use to_frequency.
2506 (find_trace): Use to_frequency.
2507 (tail_duplicate): Use to_frequency.
2508 * trans-mem.c (expand_transaction): Do not update frequency.
2509 * tree-call-cdce.c: Do not update frequency.
2510 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2511 (gimple_merge_blocks): Likewise.
2512 (gimple_split_edge): Likewise.
2513 (gimple_duplicate_sese_region): Likewise.
2514 (gimple_duplicate_sese_tail): Likewise.
2515 (move_sese_region_to_fn): Likewise.
2516 (gimple_account_profile_record): Likewise.
2517 (insert_cond_bb): Likewise.
2518 * tree-complex.c (expand_complex_div_wide): Likewise.
2519 * tree-eh.c (lower_resx): Update profile.
2520 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2522 (initialize_cfun): Do not initialize frequencies
2523 (freqs_to_counts): Delete.
2524 (copy_cfg_body): Ignore count parameter.
2525 (copy_body): Update.
2526 (expand_call_inline): Update count_max.
2527 (optimize_inline_calls): Update count_max.
2528 (tree_function_versioning): Update count_max.
2529 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2530 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2532 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2533 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2534 (try_peel_loop): Likewise.
2535 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2537 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2538 (tree_transform_and_unroll_loop): Do not use frequencies
2539 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2540 Use reliable prediction only.
2541 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2542 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2543 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2544 probability scaling.
2545 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2546 not update frequency
2547 (any_remaining_duplicated_blocks): Likewise.
2548 (update_profile): Likewise.
2549 (estimated_freqs_path): Delete.
2550 (freqs_to_counts_path): Delete.
2551 (clear_counts_path): Delete.
2552 (ssa_fix_duplicate_block_edges): Likewise.
2553 (duplicate_thread_path): Likewise.
2554 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2555 * tree-tailcall.c (decrease_profile): Do not update frequency.
2556 (eliminate_tail_call): Likewise.
2557 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2558 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2559 (optimize_mask_stores): Likewise.
2560 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2561 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2562 (ubsan_expand_ptr_ifn): Update profile.
2563 * value-prof.c (gimple_ic): Simplify.
2564 * value-prof.h (gimple_ic): Update prototype.
2565 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2566 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2568 (want_inline_self_recursive_call_p): Likewise.
2569 (resolve_noninline_speculation): Only cummulate defined counts.
2570 (inline_small_functions): Use nonzero_p.
2571 (ipa_inline): Do not access freed node.
2573 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2575 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2576 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2578 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2580 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2581 non-legitimate address.
2583 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2585 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2586 (*lt0_<mode>di, *lt0_<mode>si): New.
2588 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2590 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2591 TARGET_PAIRED_FLOAT.
2593 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2594 Jim Wilson <jim.wilson@linaro.org>
2596 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2597 * config/aarch64/aarch64-tune.md: Regenerated.
2598 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2599 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2601 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2603 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2604 (arc_expand_prologue): Restore blink for millicode.
2605 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2607 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2610 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2611 gen_vec_duplicate after forcing the scalar into a register.
2613 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2615 * combine (try_combine): Print the insns input to try_combine to the
2618 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2621 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2622 Remove second argument from aarch64_process_target_attr call.
2623 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2625 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2627 (aarch64_handle_attr_arch): Remove second argument.
2628 (aarch64_handle_attr_cpu): Ditto.
2629 (aarch64_handle_attr_tune): Ditto.
2630 (aarch64_handle_attr_isa_flags): Ditto.
2631 (aarch64_process_one_target_attr): Ditto.
2632 (aarch64_process_target_attr): Ditto.
2633 (aarch64_option_valid_attribute_p): Remove second argument.
2634 on aarch64_process_target_attr call.
2636 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2638 * diagnostic.c: Include "selftest-diagnostic.h".
2639 (selftest::assert_location_text): New function.
2640 (selftest::test_diagnostic_get_location_text): New function.
2641 (selftest::diagnostic_c_tests): Call it.
2643 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2645 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2646 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2647 (class selftest::test_diagnostic_context): Move to...
2648 * selftest-diagnostic.c: New file.
2649 * selftest-diagnostic.h: New file.
2651 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2653 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2654 offset range for FT32B.
2655 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2656 * config/ft32/ft32.md: Add TARGET_NOPM.
2657 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2658 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2660 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2662 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2664 2017-11-02 Jeff Law <law@redhat.com>
2666 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2667 virtual keyword on FINAL OVERRIDE members.
2669 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2670 virtuals together. Add virtual destructor.
2671 (substitute_and_fold_engine): Similarly.
2673 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2675 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2677 2017-11-02 Richard Biener <rguenther@suse.de>
2679 PR tree-optimization/82795
2680 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2682 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2684 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2686 (gcc_GAS_CHECK_FEATURE): Remove.
2687 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2688 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2689 * configure: Regenerate.
2691 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2693 * config/arc/arc.c (hwloop_optimize): Account for empty
2696 2017-11-02 Richard Biener <rguenther@suse.de>
2699 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2700 Truncate ARRAY_REF index and element size.
2702 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2704 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2706 2017-11-01 Jeff Law <law@redhat.com>
2708 * tree-ssa-ccp.c (ccp_folder): New class derived from
2709 substitute_and_fold_engine.
2710 (ccp_folder::get_value): New member function.
2711 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2712 (ccp_fold_stmt): Remove prototype.
2713 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2714 * tree-ssa-copy.c (copy_folder): New class derived from
2715 substitute_and_fold_engine.
2716 (copy_folder::get_value): Renamed from get_value.
2717 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2718 * tree-vrp.c (vrp_folder): New class derived from
2719 substitute_and_fold_engine.
2720 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2721 (vrp_folder::get_value): New member function.
2722 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2723 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2724 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2725 provide a class interface to folder/substitute routines.
2726 (ssa_prop_fold_stmt_fn): Remove typedef.
2727 (ssa_prop_get_value_fn): Likewise.
2728 (subsitute_and_fold): Remove prototype.
2729 (replace_uses_in): Likewise.
2730 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2731 Renamed from replace_uses_in. Call the virtual member function
2732 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2733 (substitute_and_fold_dom_walker): Remove initialization of
2734 data member entries for calbacks. Add substitute_and_fold_engine
2735 member and initialize it.
2736 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2737 member functions for get_value, replace_phi_args_in c
2738 replace_uses_in, and fold_stmt calls.
2739 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2740 substitute_and_fold. Remove assert. Update ctor call.
2742 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2743 (ssa_prop_visit_phi_fn): Likewise.
2744 (class ssa_propagation_engine): New class to provide an interface
2746 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2748 (ssa_prop_visit_phi): Likewise.
2749 (ssa_propagation_engine::simulate_stmt): Moved into class.
2750 Call visit_phi/visit_stmt from the class rather than via
2751 file scoped static variables.
2752 (ssa_propagation_engine::simulate_block): Moved into class.
2753 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2754 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2755 set file scoped statics for the visit_stmt/visit_phi callbacks.
2756 * tree-complex.c (complex_propagate): New class derived from
2757 ssa_propagation_engine.
2758 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2759 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2760 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2762 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2763 ssa_propagation_engine.
2764 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2765 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2766 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2767 * tree-ssa-copy.c (copy_prop): New class derived from
2768 ssa_propagation_engine.
2769 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2770 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2771 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2772 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2773 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2774 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2775 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2777 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2779 PR rtl-optimization/82778
2780 PR rtl-optimization/82597
2781 * compare-elim.c (struct comparison): Add in_a_setter field.
2782 (find_comparison_dom_walker::before_dom_children): Remove killed
2783 bitmap and df_simulate_find_defs call, instead walk the defs.
2784 Compute last_setter and initialize in_a_setter. Merge definitions
2785 with first initialization for a few variables.
2786 (try_validate_parallel): Use insn_invalid_p instead of
2787 recog_memoized. Return insn rather than just the pattern.
2788 (try_merge_compare): Fix up comment. Don't uselessly test if
2789 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2791 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2794 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2795 Alan Hayward <alan.hayward@arm.com>
2796 David Sherwood <david.sherwood@arm.com>
2798 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2799 aarch64_hard_regno_nregs to get the number of registers
2802 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2803 Alan Hayward <alan.hayward@arm.com>
2804 David Sherwood <david.sherwood@arm.com>
2806 * config/aarch64/constraints.md (Upl): Rename to...
2808 * config/aarch64/aarch64.md
2809 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2812 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2813 Alan Hayward <alan.hayward@arm.com>
2814 David Sherwood <david.sherwood@arm.com>
2816 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2817 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2820 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2821 Alan Hayward <alan.hayward@arm.com>
2822 David Sherwood <david.sherwood@arm.com>
2824 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2825 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2826 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2828 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2829 of a single element.
2830 * config/aarch64/iterators.md: Add a comment above the permute
2831 unspecs to say that they are generated directly by
2832 aarch64_expand_vec_perm_const.
2833 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2835 2017-11-01 Nathan Sidwell <nathan@acm.org>
2837 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2839 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2841 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2842 and medany code models, and describe what they do.
2844 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2846 Revert accidental duplicate:
2848 * combine.c (can_change_dest_mode): Reject changes in
2849 REGMODE_NATURAL_SIZE.
2851 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2853 PR rtl-optimization/64682
2854 PR rtl-optimization/69567
2855 PR rtl-optimization/69737
2856 PR rtl-optimization/82683
2857 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2858 register mentioned in the note, drop the note, unless it came from I3,
2859 in which case it should go to I3 again.
2861 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2863 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2864 and return false if not.
2865 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2867 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2869 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2870 range is known to be empty.
2872 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2873 Alan Hayward <alan.hayward@arm.com>
2874 David Sherwood <david.sherwood@arm.com>
2876 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2877 and CONST_VECTOR_NUNITS instead of computing the number of units from
2878 the byte sizes of the vector and element.
2879 (simplify_binary_operation_1): Likewise.
2880 (simplify_const_binary_operation): Likewise.
2881 (simplify_ternary_operation): Likewise.
2883 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2884 Alan Hayward <alan.hayward@arm.com>
2885 David Sherwood <david.sherwood@arm.com>
2887 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2888 (int_mem_offset): ...this new function.
2889 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2890 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2891 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2894 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2895 Alan Hayward <alan.hayward@arm.com>
2896 David Sherwood <david.sherwood@arm.com>
2898 * lower-subreg.c (interesting_mode_p): New function.
2899 (compute_costs, find_decomposable_subregs, decompose_register)
2900 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2901 (resolve_clobber, dump_choices): Use it.
2903 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2904 Alan Hayward <alan.hayward@arm.com>
2905 David Sherwood <david.sherwood@arm.com>
2907 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2909 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2910 Alan Hayward <alan.hayward@arm.com>
2911 David Sherwood <david.sherwood@arm.com>
2913 * alias.c (find_base_value, find_base_term): Only process integer
2914 truncations. Check the precision rather than the size.
2916 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2917 Alan Hayward <alan.hayward@arm.com>
2918 David Sherwood <david.sherwood@arm.com>
2920 * machmode.h (is_narrower_int_mode): New function
2921 * optabs.c (expand_float, expand_fix): Use it.
2922 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2924 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2925 Alan Hayward <alan.hayward@arm.com>
2926 David Sherwood <david.sherwood@arm.com>
2928 * rtl.h (narrower_subreg_mode): New function.
2929 * ira-color.c (update_costs_from_allocno): Use it.
2931 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2932 Alan Hayward <alan.hayward@arm.com>
2933 David Sherwood <david.sherwood@arm.com>
2935 * optabs-query.h (convert_optab_p): New function, split out from...
2936 (convert_optab_handler): ...here.
2937 (widening_optab_handler): Delete.
2938 (find_widening_optab_handler): Remove permit_non_widening parameter.
2939 (find_widening_optab_handler_and_mode): Likewise. Provide an
2940 override that operates on mode class wrappers.
2941 * optabs-query.c (widening_optab_handler): Delete.
2942 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2943 parameter. Assert that the two modes are the same class and that
2944 the "from" mode is narrower than the "to" mode. Use
2945 convert_optab_handler instead of widening_optab_handler.
2946 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2947 instead of widening_optab_handler.
2948 * expr.c (expand_expr_real_2): Update calls to
2949 find_widening_optab_handler.
2950 * optabs.c (expand_widen_pattern_expr): Likewise.
2951 (expand_binop_directly): Take the insn_code as a parameter.
2952 (expand_binop): Only call find_widening_optab_handler for
2953 conversion optabs; use optab_handler otherwise. Update calls
2954 to find_widening_optab_handler and expand_binop_directly.
2955 Use convert_optab_handler instead of widening_optab_handler.
2956 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2957 find_widening_optab_handler and use scalar_mode rather than
2959 (convert_plusminus_to_widen): Likewise.
2961 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2962 Alan Hayward <alan.hayward@arm.com>
2963 David Sherwood <david.sherwood@arm.com>
2965 * machmode.h (fixed_size_mode): New class.
2966 * rtl.h (get_pool_mode): Return fixed_size_mode.
2967 * gengtype.c (main): Add fixed_size_mode.
2968 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2969 (get_raw_arg_mode): Likewise.
2970 * doc/tm.texi: Regenerate.
2971 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2972 * targhooks.c (default_get_reg_raw_mode): Likewise.
2973 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2974 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2975 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2976 (msp430_get_raw_result_mode): Likewise.
2977 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2978 * dbxout.c (dbxout_parms): Require fixed-size modes.
2979 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2980 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2981 * omp-low.c (lower_oacc_reductions): Likewise.
2982 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2983 (simplify_subreg): Update accordingly.
2984 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2985 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2986 that aren't fixed-size.
2987 (get_pool_mode): Return a fixed_size_mode.
2988 (output_constant_pool_2): Take a fixed_size_mode.
2990 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2991 Alan Hayward <alan.hayward@arm.com>
2992 David Sherwood <david.sherwood@arm.com>
2994 * doc/rtl.texi (vec_series): Document.
2995 (const): Say that the operand can be a vec_series.
2996 * rtl.def (VEC_SERIES): New rtx code.
2997 * rtl.h (const_vec_series_p_1): Declare.
2998 (const_vec_series_p): New function.
2999 * emit-rtl.h (gen_const_vec_series): Declare.
3000 (gen_vec_series): Likewise.
3001 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3002 (gen_vec_series): Likewise.
3003 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3004 * simplify-rtx.c (simplify_unary_operation): Handle negations
3006 (simplify_binary_operation_series): New function.
3007 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3008 (test_vector_ops_series): New function.
3009 (test_vector_ops): Call it.
3010 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3011 gen_const_vec_series.
3012 (altivec_lvsr): Likewise.
3013 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3015 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3016 Alan Hayward <alan.hayward@arm.com>
3017 David Sherwood <david.sherwood@arm.com>
3019 * doc/rtl.texi (const): Update description of address constants.
3020 Say that vector constants are allowed too.
3021 * common.md (E, F): Use CONSTANT_P instead of checking for
3023 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3024 checking for CONST_VECTOR.
3025 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3027 * expr.c (expand_expr_real_2): Check for vector modes instead
3028 of checking for CONST_VECTOR.
3029 * rtl.h (const_vec_p): New function.
3030 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3031 (unwrap_const_vec_duplicate): Handle them here too.
3033 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3034 David Malcolm <dmalcolm@redhat.com>
3035 Alan Hayward <alan.hayward@arm.com>
3036 David Sherwood <david.sherwood@arm.com>
3038 * rtl.h (vec_duplicate_p): New function.
3039 * selftest-rtl.c (assert_rtx_eq_at): New function.
3040 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3041 (assert_rtx_eq_at): Declare.
3042 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3043 * selftest-run-tests.c (selftest::run_tests): Call it.
3044 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3045 (simplify_unary_operation_1): Recursively handle vector duplicates.
3046 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3048 (simplify_subreg): Handle subregs of vector duplicates.
3049 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3050 (selftest::simplify_rtx_c_tests): New functions.
3052 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3053 Alan Hayward <alan.hayward@arm.com>
3054 David Sherwood <david.sherwood@arm.com>
3056 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3057 (gen_vec_duplicate): Likewise.
3058 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3060 (gen_const_vector): ...here.
3061 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3062 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3063 whose elements are all equal.
3064 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3065 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3066 (simplify_relational_operation): Likewise.
3067 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3069 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3070 (aarch64_expand_vector_init): Likewise.
3071 * config/arm/arm.c (neon_vdup_constant): Likewise.
3072 (neon_expand_vector_init): Likewise.
3073 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3074 (arm_block_set_unaligned_vect): Likewise.
3075 (arm_block_set_aligned_vect): Likewise.
3076 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3077 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3078 (expand_vec_perm_even_odd_pack): Likewise.
3079 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3080 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3081 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3082 gen_const_vec_duplicate.
3083 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3084 * config/mips/mips.c (mips_gen_const_int_vector): Use
3085 gen_const_vec_duplicate.
3086 (mips_expand_vector_init): Use CONST0_RTX.
3087 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3088 (define_split): Use gen_const_vec_duplicate.
3089 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3090 (define_split): Use gen_const_vec_duplicate.
3091 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3092 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3093 * config/spu/spu.c (spu_const): Likewise.
3095 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3096 Alan Hayward <alan.hayward@arm.com>
3097 David Sherwood <david.sherwood@arm.com>
3099 * combine.c (can_change_dest_mode): Reject changes in
3100 REGMODE_NATURAL_SIZE.
3102 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3104 * configure.ac (--enable-libssp): New.
3105 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3106 trying to determine target-specific default. Adjust indentation.
3107 * configure: Regenerated.
3108 * doc/install.texi (Configuration): Expand --disable-libssp
3111 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3113 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3116 2017-10-31 Martin Jambor <mjambor@suse.cz>
3119 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3121 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3123 * auto-profile.c (autofdo_source_profile::read): Use
3124 UNKNOWN_LOCATION rather than 0.
3125 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3126 (warning_at): ...this overload.
3127 (warning_at_rich_loc_n): Rename to...
3128 (warning_n): ...this overload.
3129 (error_at_rich_loc): Rename to...
3130 (error_at): ...this overload.
3131 (pedwarn_at_rich_loc): Rename to...
3132 (pedwarn): ...this overload.
3133 (permerror_at_rich_loc): Rename to...
3134 (permerror): ...this overload.
3135 (inform_at_rich_loc): Rename to...
3136 (inform): ...this overload.
3137 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3138 (diagnostic_n_impl_richloc): Rename to...
3139 (diagnostic_n_impl): ...this rich_location *-based decl.
3140 (inform_at_rich_loc): Rename to...
3141 (inform): ...this, and add an assertion.
3142 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3143 (warning_at_rich_loc): Rename to...
3144 (warning_at): ...this, and add an assertion.
3145 (warning_at_rich_loc_n): Rename to...
3146 (warning_n): ...this, and add an assertion.
3147 (warning_n): Update location_t-based implementation for removal of
3148 location_t-based diagnostic_n_impl.
3149 (pedwarn_at_rich_loc): Rename to...
3150 (pedwarn): ...this, and add an assertion.
3151 (permerror_at_rich_loc): Rename to...
3152 (permerror): ...this, and add an assertion.
3153 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3154 (error_at_rich_loc): Rename to...
3155 (error_at): ...this, and add an assertion.
3156 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3157 (driver::do_spec_on_infiles): Likewise.
3158 * substring-locations.c (format_warning_va): Update for renaming
3159 of inform_at_rich_loc.
3161 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3163 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3164 _Float<N>X built-in functions so that the variant without the
3165 "__builtin_" prefix is only enabled for the GNU C and Objective C
3166 languages when they are in non-strict ANSI/ISO mode.
3167 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3168 * target.def (floatn_builtin_p): Add a target hook to control
3169 whether _Float<N> and _Float<N>X built-in functions without the
3170 "__builtin_" prefix are enabled, and return true for C and
3171 Objective C in the default hook. Include langhooks.h in
3173 * targhooks.h (default_floatn_builtin_p): Likewise.
3174 * targhooks.c (default_floatn_builtin_p): Likewise.
3175 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3176 floatn_builtin_p target hook.
3177 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3179 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3180 Eric Botcazou <ebotcazou@adacore.com>
3182 PR rtl-optimization/81803
3183 * lra-constraints.c (curr_insn_transform): Also reload the whole
3184 register for a strict subreg no wider than a word if this is for
3185 a WORD_REGISTER_OPERATIONS target.
3187 2017-10-31 Jason Merrill <jason@redhat.com>
3189 * gdbinit.in: Skip over inlines from timevar.h.
3191 2017-10-31 Martin Liska <mliska@suse.cz>
3193 * doc/gcov.texi: Document new option.
3194 * gcov.c (print_usage): Likewise print it.
3195 (process_args): Support the argument.
3196 (format_count): New function.
3197 (format_gcov): Use the function.
3199 2017-10-31 Martin Liska <mliska@suse.cz>
3201 * gcov.c (struct name_map): do not use typedef.
3202 Define operator== and operator<.
3203 (name_search): Remove.
3204 (name_sort): Remove.
3205 (main): Do not allocate names.
3206 (process_file): Add vertical space.
3207 (generate_results): Use std::find.
3208 (release_structures): Do not release memory.
3209 (find_source): Use std::find.
3211 2017-10-31 Martin Liska <mliska@suse.cz>
3213 * gcov.c (struct line_info): Remove it's typedef.
3214 (line_info::line_info): Add proper ctor.
3215 (line_info::has_block): Do not use a typedef.
3216 (struct source_info): Do not use typedef.
3217 (circuit): Likewise.
3218 (get_cycles_count): Likewise.
3219 (output_intermediate_file): Iterate via vector iterator.
3220 (add_line_counts): Use std::vector methods.
3221 (accumulate_line_counts): Likewise.
3222 (output_lines): Likewise.
3224 2017-10-31 Martin Liska <mliska@suse.cz>
3226 * gcov.c (struct source_info): Remove typedef.
3227 (source_info::source_info): Add proper ctor.
3228 (accumulate_line_counts): Use struct, not it's typedef.
3229 (output_gcov_file): Likewise.
3230 (output_lines): Likewise.
3231 (main): Do not allocate an array.
3232 (output_intermediate_file): Use size of vector container.
3233 (process_file): Resize the vector.
3234 (generate_results): Do not preallocate, use newly added vector
3236 (release_structures): Do not release sources.
3237 (find_source): Use vector methods.
3238 (add_line_counts): Do not use typedef.
3240 2017-10-31 Martin Liska <mliska@suse.cz>
3242 * doc/gcov.texi: Document that.
3243 * gcov.c (add_line_counts): Mark lines with a non-executed
3245 (output_line_beginning): Handle such lines.
3246 (output_lines): Pass new argument.
3247 (output_intermediate_file): Print it in intermediate format.
3249 2017-10-31 Martin Liska <mliska@suse.cz>
3251 * color-macros.h: New file.
3252 * diagnostic-color.c: Factor out color related to macros to
3254 * doc/gcov.texi: Document -k option.
3255 * gcov.c (INCLUDE_STRING): Include string.h.
3256 (print_usage): Add -k option.
3257 (process_args): Parse it.
3258 (pad_count_string): New function.
3259 (output_line_beginning): Likewise.
3260 (DEFAULT_LINE_START): New macro.
3261 (output_lines): Support color output.
3263 2017-10-31 Martin Liska <mliska@suse.cz>
3265 PR gcov-profile/82633
3266 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3267 their interaction with GCOV infrastructure.
3268 * configure.ac: Add -fkeep-{inline,static}-functions to
3270 * configure: Regenerate.
3272 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3275 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3277 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3280 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3281 into a register if it does not fit into an immediate offset field.
3283 2017-10-31 Olivier Hainque <hainque@adacore.com>
3285 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3287 2017-10-31 Julia Koval <julia.koval@intel.com>
3289 * config.gcc: Add gfniintrin.h.
3290 * config/i386/gfniintrin.h: New.
3291 * config/i386/i386-builtin-types.def
3292 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3293 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3294 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3295 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3296 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3297 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3298 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3299 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3300 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3301 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3302 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3303 * config/i386/immintrin.h: Include gfniintrin.h.
3304 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3306 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3308 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3310 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3312 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3313 (arm_ashldi3_1bit): Remove pattern.
3314 (ashrdi3): Remove shift by 1 expansion.
3315 (arm_ashrdi3_1bit): Remove pattern.
3316 (lshrdi3): Remove shift by 1 expansion.
3317 (arm_lshrdi3_1bit): Remove pattern.
3318 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3319 cost of ashldi3 by 1.
3320 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3321 (<shift>di3_neon): Likewise.
3323 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3325 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3326 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3327 and (*aarch64_simd_mov<VQ:mode>).
3328 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3329 pattern alternative.
3330 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3331 attributes to match pattern alternative.
3333 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3335 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3336 * config/rs6000/emmintrin.h: New file.
3337 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3339 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3341 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3342 * (movdi_vfp_cortexa8): Remove pattern.
3344 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3346 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3348 (Specific, alpha*-dec-osf5.1): Remove.
3349 (Specific, mips-sgi-irix5): Remove.
3350 (Specific, mips-sgi-irix6): Remove.
3352 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3355 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3356 arguments to clear_bit_region_be.
3358 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3360 * gimplify.c: Include memmodel.h.
3362 2017-10-30 Martin Jambor <mjambor@suse.cz>
3364 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3365 condition whether loop should be executed at all.
3367 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3369 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3370 gimple folding of vec_madd() intrinsics.
3371 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3372 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3373 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3375 2017-10-30 Richard Biener <rguenther@suse.de>
3377 PR tree-optimization/82762
3379 2017-10-23 Richard Biener <rguenther@suse.de>
3381 PR tree-optimization/82129
3383 2017-08-01 Richard Biener <rguenther@suse.de>
3385 PR tree-optimization/81181
3386 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3387 (compute_antic): ... end of iteration here.
3389 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3391 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3392 -std=iso9899:2017 and -std=gnu17.
3393 * doc/standards.texi (C Language): Document C17 support.
3394 * doc/cpp.texi (Overview): Mention -std=c17.
3395 (Standard Predefined Macros): Document C11 and C17 values of
3396 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3397 * doc/extend.texi (Inline): Do not list individual options for
3398 standards newer than C99.
3399 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3401 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3404 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3406 * asan.c (asan_finish_file): Align asan globals array by shadow
3409 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3412 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3413 (struct store_immediate_info): Add bitregion_start and bitregion_end
3415 (store_immediate_info::store_immediate_info): Add brs and bre
3416 arguments and initialize bitregion_{start,end} from those.
3417 (struct merged_store_group): Add bitregion_start, bitregion_end,
3418 align_base and mask fields. Drop unnecessary struct keyword from
3419 struct store_immediate_info. Add do_merge method.
3420 (clear_bit_region_be): Use memset instead of loop storing zeros.
3421 (merged_store_group::do_merge): New method.
3422 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3423 stores as long as the surrounding bitregions have no gaps.
3424 (merged_store_group::merge_overlapping): Use do_merge.
3425 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3426 is byte aligned, rather than requiring that start and width are
3427 byte aligned. Drop unnecessary struct keyword from
3428 struct store_immediate_info. Allocate and populate also mask array.
3429 Make start of the arrays relative to bitregion_start rather than
3430 start and size them according to bitregion_{end,start} difference.
3431 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3432 struct store_immediate_info.
3433 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3434 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3435 struct keyword from struct store_immediate_info.
3436 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3437 between stores as long as the surrounding bitregions have no gaps.
3439 (struct split_store): Add orig non-static data member.
3440 (split_store::split_store): Initialize orig to false.
3441 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3442 if there is exactly a single original stmt. Change stmts argument
3443 to pointer from reference, if NULL, don't push anything to it. Add
3444 first argument, use it to optimize skipping over orig stmts that
3445 are known to be before bitpos already. Simplify.
3446 (split_group): Return unsigned int count how many stores are or
3447 would be needed rather than a bool. Add allow_unaligned argument.
3448 Change split_stores argument from reference to pointer, if NULL,
3449 only do a dry run computing how many stores would be produced.
3450 Rewritten algorithm to use both alignment and misalign if
3451 !allow_unaligned and handle bitfield stores with gaps.
3452 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3453 from bitregion_start instead of start. Compute allow_unaligned
3454 here, if true, do 2 split_group dry runs to compute which one
3455 produces fewer stores and prefer aligned if equal. Punt if
3456 new count is bigger or equal than original before emitting any
3457 statements, rather than during that. Remove no longer needed
3458 new_ssa_names tracking. Replace num_stmts with
3459 split_stores.length (). Use 32-bit stack allocated entries
3460 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3461 if possible. Handle bitfields with gaps.
3462 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3463 Compute bitregion_{start,end} for the stores and construct
3464 store_immediate_info with that. Formatting fixes.
3466 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3469 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3470 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3472 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3474 * gimplify.c: Include tm_p.h.
3476 * common.opt (gcoff): Re-add as ignored option.
3477 (gcoff1, gcoff2, gcoff3): Likewise.
3479 * Makefile.in (OBJS): Delete sdbout.o.
3480 (GTFILES): Delete $(srcdir)/sdbout.c.
3481 * debug.h: Delete sdb_debug_hooks.
3482 * final.c: Delete sdbout.h include.
3483 (final_scan_insn): Delete SDB_DEBUG check.
3484 (rest_of_clean_state): Likewise.
3485 * output.h: Delete sdb_begin_function_line.
3488 * toplev.c: Delete sdbout.h include.
3489 (process_options): Delete SDB_DEBUG check.
3490 * tree-core.h (tree_type_common): Delete pointer field of
3492 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3493 TYPE_SYMTAB_POINTER.
3494 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3495 (TYPE_SYMTAB_IS_POINTER): Delete.
3496 (TYPE_SYMTAB_IS_DIE): Renumber.
3497 * xcoffout.c: Refer to former sdbout.c file.
3498 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3500 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3501 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3502 (Debugging Options): Delete -gcoff.
3503 (-gstabs): Delete SDB reference.
3505 (-gcoff@var{level}): Delete.
3506 * doc/passes.texi (Debugging information output): Delete SDB and
3507 sdbout.c references.
3508 * doc/tm.texi: Regenerate.
3509 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3510 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3512 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3513 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3514 (SDB_DEBUGGING_INFO): Delete.
3515 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3516 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3517 * target.def (output_source_filename): Delete COFF reference.
3519 * common.opt (gcoff): Delete.
3520 (gxcoff+): Update Negative chain.
3521 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3523 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3524 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3525 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3527 (expand_function_start): Change sdb reference to past tense.
3528 (expand_function_end): Change sdb reference to past tense.
3529 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3530 * opts.c (debug_type_names): Delete coff entry.
3531 (common_handle_option): Delete OPT_gcoff case.
3532 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3534 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3535 * config/cris/cris.h: Delete SDB reference in comment.
3536 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3537 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3538 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3539 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3541 (ix86_expand_prologue): Likewise.
3542 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3543 * config/ia64/ia64.h: Likewise.
3544 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3545 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3547 * config/mmix/mmix.h: Likewise.
3548 * config/nds32/nds32.c: Likewise.
3549 * config/stormy/storym16.h: Likewise.
3550 * config/visium/visium.h: Likewise.
3551 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3553 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3555 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3556 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3557 FRAME_POINTER_REGNUM point at high end of local var area.
3559 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3561 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3562 Move comment around. Do not reset best_edge for a copiable
3563 destination if the copy would cause a partition change.
3564 (better_edge_p): Remove redundant check.
3566 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3568 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3570 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3572 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3573 for math functions that have _Float<N> and _Float<N>X variants.
3574 (mathfn_built_in_2): Add support for math functions that have
3575 _Float<N> and _Float<N>X variants.
3576 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3577 (expand_builtin_mathfn_ternary): Add support for fma with
3578 _Float<N> and _Float<N>X variants.
3579 (expand_builtin): Likewise.
3580 (fold_builtin_3): Likewise.
3581 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3582 create math function _Float<N> and _Float<N>X variants as external
3584 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3585 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3586 the __builtin_ prefix and if not strict ansi, without the prefix.
3587 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3588 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3589 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3590 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3591 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3592 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3593 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3594 function signatures for fma _Float<N> and _Float<N>X variants.
3595 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3596 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3597 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3598 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3599 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3600 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3601 * gencfn-macros.c (print_case_cfn): Add support for math functions
3602 that have _Float<N> and _Float<N>X variants.
3603 (print_define_operator_list): Likewise.
3604 (fltfn_suffixes): Likewise.
3606 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3607 for math functions that have _Float<N> and _Float<N>X variants.
3608 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3609 and _Float<N>X variants.
3610 (COPYSIGN): Likewise.
3613 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3614 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3616 (integer_valued_read_call_p): Likewise.
3617 * fold-const-call.c (fold_const_call_ss): Likewise.
3618 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3619 _Float<N> and _Float<N>X variants.
3620 (fold_const_call_ssss): Add support for fma _Float<N> and
3621 _Float<N>X variants.
3622 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3623 support for copysign and fma _Float<N> and _Float<N>X variants.
3624 (backprop::process_builtin_call_use): Likewise.
3625 * tree-call-cdce.c (can_test_argument_range); Add support for
3626 sqrt _Float<N> and _Float<N>X variants.
3627 (edom_only_function): Likewise.
3628 (get_no_error_domain): Likewise.
3629 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3630 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3631 copysign _Float<N> and _Float<N>X variants.
3632 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3633 handled by machine independent code.
3634 (FMAF128): Likewise.
3635 * doc/cpp.texi (Common Predefined Macros): Document defining
3636 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3637 fma _Float<N> and _Float<N>X variants.
3639 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3642 * config/i386/i386-modes.def (CCFPU): Remove definition.
3643 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3644 (ix86_cc_modes_compatible): Ditto.
3645 (ix86_expand_carry_flag_compare): Ditto.
3646 (ix86_expand_int_movcc): Ditto.
3647 (ix86_expand_int_addcc): Ditto.
3648 (ix86_reverse_condition): Ditto.
3649 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3650 Return true/false for unordered/ordered fp comparisons.
3651 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3652 (ix86_prepare_fp_compare_args): Update for rename.
3653 (ix86_expand_fp_compare): Update for rename. Generate unordered
3654 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3655 (ix86_expand_sse_compare_and_jump): Ditto.
3656 * config/i386/predicates.md (fcmov_comparison_operator):
3657 Remove CCFPU mode handling.
3658 (ix86_comparison_operator): Ditto.
3659 (ix86_carry_flag_operator): Ditto.
3660 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3661 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3662 (*cmpu<mode>_cc_i387): Ditto.
3663 (FPCMP): Remove mode iterator.
3664 (unord): Remove mode attribute.
3665 (unord_subst): New define_subst transformation
3666 (unord): New define_subst attribute.
3668 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3669 (*cmpi<unord>xf_i387): Ditto.
3670 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3671 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3672 using unord_subst transformation.
3673 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3674 (round_saeonly): Also handle CCFP mode.
3675 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3676 Remove UNSPEC_SAHF unspec handling.
3678 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3680 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3682 2017-10-27 Jeff Law <law@redhat.com>
3684 * gimple-ssa-sprintf.c: Include domwalk.h.
3685 (class sprintf_dom_walker): New class, derived from dom_walker.
3686 (sprintf_dom_walker::before_dom_children): New function.
3687 (struct call_info): Moved into sprintf_dom_walker class
3688 (compute_formath_length, handle_gimple_call): Likewise.
3689 (sprintf_length::execute): Call the dominator walker rather
3690 than walking the statements.
3692 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3693 gimple statement locations.
3694 (check_array_bounds): Corresponding changes. Get the statement's
3695 location directly from wi->stmt.
3697 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3700 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3702 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3704 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3705 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3707 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3710 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3711 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3712 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3713 maybe_get_pool_constant.
3714 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3717 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3719 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3720 2.26 caveat. Update gas and gld versions.
3721 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3724 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3726 * cgraph.h (set_malloc_flag): Declare.
3727 * cgraph.c (set_malloc_flag_1): New function.
3728 (set_malloc_flag): Likewise.
3729 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3730 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3732 (read_ipa_call_summary): Add support for reading is_return_callee.
3733 (write_ipa_call_summary): Stream is_return_callee.
3734 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3735 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3736 ipa-prop.h, ipa-fnsummary.h.
3737 (pure_const_names): Change to static.
3738 (malloc_state_e): Define.
3739 (malloc_state_names): Define.
3740 (funct_state_d): Add field malloc_state.
3741 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3742 (check_retval_uses): New function.
3743 (malloc_candidate_p): Likewise.
3744 (analyze_function): Add support for malloc attribute.
3745 (pure_const_write_summary): Stream malloc_state.
3746 (pure_const_read_summary): Add support for reading malloc_state.
3747 (dump_malloc_lattice): New function.
3748 (propagate_malloc): New function.
3749 (warn_function_malloc): New function.
3750 (ipa_pure_const::execute): Call propagate_malloc and
3751 ipa_free_fn_summary.
3752 (pass_local_pure_const::execute): Add support for malloc attribute.
3753 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3754 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3756 2017-10-27 Martin Liska <mliska@suse.cz>
3758 PR gcov-profile/82457
3759 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3760 for fork-like functions to be properly instrumented.
3762 2017-10-27 Richard Biener <rguenther@suse.de>
3765 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3766 info when we redirected EH.
3768 2017-10-26 Michael Collison <michael.collison@arm.com>
3770 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3772 (<optab>_trunchf<GPI:mode>2: New pattern.
3773 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3774 * config/aarch64/iterators.md (wv): New mode attribute.
3775 (vf, VF): New mode attributes.
3776 (vgp, VGP): New mode attributes.
3777 (s): Update attribute with SImode and DImode prefixes.
3779 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3781 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3782 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3783 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3784 (nios2_option_overide): Initialize it. Don't allow R0-relative
3785 addressing with PIC.
3786 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3787 (nios2_symbolic_constant_p): Likewise.
3788 (nios2_legitimate_address_p): Likewise.
3789 (nios2_r0rel_section_name_p): New.
3790 (nios2_symbol_ref_in_r0rel_data_p): New.
3791 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3792 (r0rel_constant_p): New.
3793 (nios2_print_operand_address): Handle r0rel_constant_p.
3794 (nios2_cdx_narrow_form_p): Likewise.
3795 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3796 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3797 (Nios II Options): Document -mr0rel-sec.
3799 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3801 * config/nios2/nios2.c: Include xregex.h.
3802 (nios2_gprel_sec_regex): New.
3803 (nios2_option_overide): Initialize it. Don't allow GP-relative
3804 addressing with PIC.
3805 (nios2_small_section_name_p): Check for regex match.
3806 * config/nios2/nios2.opt (mgprel-sec=): New option.
3807 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3808 (Nios II Options): Document -mgprel-sec.
3810 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3812 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3814 2017-10-26 Tom de Vries <tom@codesourcery.com>
3816 PR tree-optimization/82707
3817 * gimple.c (gimple_copy): Fix unsharing of
3818 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3820 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3822 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3823 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3824 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3825 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3826 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3827 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3828 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3829 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3830 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3831 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3832 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3833 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3834 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3835 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3836 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3837 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3838 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3840 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3842 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3844 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3845 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3846 warning to rs6000.c. Remove the Undocumented flag, since it has
3848 (-mabi=ibmlongdouble): Likewise.
3849 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3850 already set, set the default format for long double.
3851 (rs6000_debug_reg_global): Print whether long double is IBM or
3853 (rs6000_option_override_internal): Rework setting long double
3854 format. Only warn if the user is changing the long double default
3855 and they did not use -Wno-psabi.
3856 * doc/invoke.texi (PowerPC options): Update the documentation for
3857 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3859 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3860 Alan Hayward <alan.hayward@arm.com>
3861 David Sherwood <david.sherwood@arm.com>
3863 * rtl.h (wider_subreg_mode): New function.
3864 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3865 rather than an unsigned int *.
3866 * ira-color.c (regno_max_ref_width): Replace with...
3867 (regno_max_ref_mode): ...this new variable.
3868 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3869 Use wider_subreg_mode.
3870 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3871 rather than an unsigned int *.
3872 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3873 (process_alt_operands): Likewise.
3874 (invariant_p): Likewise.
3875 * lra-spills.c (assign_mem_slot): Likewise.
3876 (add_pseudo_to_slot): Likewise.
3877 * lra.c (collect_non_operand_hard_regs): Likewise.
3878 (add_regs_to_insn_regno_info): Likewise.
3879 * reload1.c (regno_max_ref_width): Replace with...
3880 (regno_max_ref_mode): ...this new variable.
3881 (reload): Update accordingly. Update call to
3882 ira_sort_regnos_for_alter_reg.
3883 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3884 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3885 (scan_paradoxical_subregs): Likewise.
3887 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3889 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3890 (aarch64_frame): Add emit_frame_chain boolean.
3891 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3892 Move eh_return case to aarch64_layout_frame.
3893 (aarch64_layout_frame): Initialize emit_frame_chain.
3894 (aarch64_expand_prologue): Use emit_frame_chain.
3896 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3898 * config/aarch64/aarch64.c (aarch64_layout_frame):
3899 Ensure LR is always stored at the bottom of the callee-saves.
3900 Remove rarely used frame layout which saves callee-saves at top of
3901 frame, so the store of LR can be used as a valid probe in all cases.
3903 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3905 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3906 Improve unaligned TImode/TFmode base/offset split.
3908 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3909 Alan Hayward <alan.hayward@arm.com>
3910 David Sherwood <david.sherwood@arm.com>
3912 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3913 * combine.c (find_single_use_1): Likewise.
3914 (expand_field_assignment): Likewise.
3915 (move_deaths): Likewise.
3916 * lra-constraints.c (simplify_operand_subreg): Likewise.
3917 (curr_insn_transform): Likewise.
3918 * lra.c (collect_non_operand_hard_regs): Likewise.
3919 (add_regs_to_insn_regno_info): Likewise.
3920 * rtlanal.c (reg_referenced_p): Likewise.
3921 (covers_regno_no_parallel_p): Likewise.
3923 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3925 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3926 Don't print any bits outside the precision of the value.
3927 * wide-int.cc (test_printing): Add some new tests.
3929 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3931 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3932 supports -xbrace_comment option.
3933 * configure: Regenerate.
3934 * config.in: Regenerate.
3935 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3936 (ASM_CPU_SPEC): Use it.
3938 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3940 * target.def (static_rtx_alignment): New hook.
3941 * targhooks.h (default_static_rtx_alignment): Declare.
3942 * targhooks.c (default_static_rtx_alignment): New function.
3943 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3944 * doc/tm.texi: Regenerate.
3945 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3946 instead of targetm.constant_alignment. Remove call to
3948 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3949 (cris_preferred_mininum_alignment): New function, split out from...
3950 (cris_constant_alignment): ...here.
3951 (cris_static_rtx_alignment): New function.
3952 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3954 (ix86_constant_alignment): ...here.
3955 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3956 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3957 (mmix_static_rtx_alignment): New function.
3958 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3959 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3961 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3964 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3965 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3967 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3969 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3970 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3971 mark as a sign-extending load.
3972 (local_pic_load_u): Define.
3974 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3977 * fold-const.c (operand_equal_for_comparison_p): Also return true
3978 if ARG0 is a simple variant of ARG1 with narrower precision.
3979 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3981 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3983 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3985 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3986 scatter_static, scatter_per_elt.
3987 * x86-tune-costs.h: Add new cost entries.
3989 2017-10-25 Richard Biener <rguenther@suse.de>
3991 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3992 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3993 class pass_fre): Move to ...
3994 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3995 class pass_fre): ... here and adjust for statistics.
3997 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4000 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4001 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4002 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4003 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4006 2017-10-25 Richard Biener <rguenther@suse.de>
4008 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4009 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4010 eliminate_push_avail, eliminate_insert): Move inside...
4011 (class eliminate_dom_walker): ... this class in preparation
4013 (fini_eliminate): Remove by merging with ...
4014 (eliminate): ... this function. Adjust for class changes.
4015 (pass_pre::execute): Remove fini_eliminate call.
4016 (pass_fre::execute): Likewise.
4018 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4021 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4022 (VPERMI2, VPERMI2I): New mode iterators.
4023 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4024 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4026 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4027 mode iterator. Remove 3 old define_insn patterns.
4028 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4029 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4030 VPERMI2 mode iterator, remove the other two expanders.
4031 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4032 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4033 instructions, remove the other two patterns.
4034 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4035 mode iterator, remove the other two patterns.
4036 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4037 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4038 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4039 and adjust argument order accordingly.
4040 (ix86_expand_vec_perm): Adjust caller.
4041 (expand_vec_perm_1): Likewise.
4042 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4043 (expand_vec_perm_vpermt2_vpshub2): ... this.
4044 (ix86_expand_vec_perm_const_1): Adjust caller.
4045 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4048 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4049 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4050 (vec_shl_<mode>): Remove unused expander.
4051 (avx512bw_<shift_insn><mode>3): New define_insn.
4052 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4053 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4055 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4058 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4061 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4063 PR rtl-optimization/82396
4064 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4065 (autopref_multipass_init): Simplify initialization.
4066 (autopref_rank_data): Simplify sort order.
4067 * gcc/sched-int.h (autopref_multipass_data_): Remove
4068 multi_mem_insn_p, min_offset and max_offset.
4070 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4073 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4074 Check special value of flag_omit_frame_pointer.
4075 (aarch64_can_eliminate): Likewise.
4076 (aarch64_override_options_after_change_1): Simplify handling of
4077 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4079 2017-10-24 Richard Biener <rguenther@suse.de>
4081 PR tree-optimization/82697
4082 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4083 zero for conditional load and unconditional store.
4085 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4087 * doc/install.texi: Document bootstrap-cet.
4089 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4092 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4093 ENDBR instruction at function entrance if function is only
4096 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4099 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4100 patterns to better describe from which operation the CF is computed.
4101 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4102 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4103 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4104 is 0, use _0 suffixed expanders instead of emitting a comparison
4107 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4109 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4110 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4112 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4115 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4116 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4117 * loop-iv.c (iv_get_reaching_def): Likewise.
4118 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4119 variable is promoted and the partition contains undefined values.
4121 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4123 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4125 (nios2_address_cost): Define.
4126 (nios2_legitimize_address): Recognize (exp + constant) directly.
4127 (TARGET_ADDRESS_COST): Define.
4129 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4131 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4132 (nios2_symbolic_memory_operand_p): Declare.
4133 (nios2_split_large_constant): Declare.
4134 (nios2_split_symbolic_memory_operand): Declare.
4135 * config/nios2/nios2.c: Adjust includes.
4136 (nios2_symbolic_constant_allowed): New.
4137 (nios2_symbolic_constant_p): New.
4138 (nios2_plus_symbolic_constant_p): New.
4139 (nios2_valid_addr_expr_p): Recognize addresses involving
4141 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4142 (nios2_symbolic_memory_operand_p): New.
4143 (nios2_large_constant_p): New.
4144 (nios2_split_large_constant): New.
4145 (nios2_split_plus_large_constant): New.
4146 (nios2_split_symbolic_memory_operand): New.
4147 (nios2_legitimize_address): Code refactoring. Handle addresses
4148 involving symbolic constants.
4149 (nios2_emit_move_sequence): Likewise.
4150 (nios2_print_operand): Improve error output.
4151 (nios2_print_operand_address): Handle LO_SUM.
4152 (nios2_cdx_narrow_form_p): Likewise.
4153 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4154 operands involving symbolic constants.
4155 (movhi_internal, movsi_internal): Likewise.
4156 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4157 (extendhisi2, extendqi<mode>2): Likewise.
4159 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4161 * tree-pass.h (PROP_rtl_split_insns): Define.
4162 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4164 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4166 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4168 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4171 * target.def (const_not_ok_for_debug_p): Default to
4172 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4173 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4174 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4175 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4176 which targetm.const_not_ok_for_debug_p returned true.
4177 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4179 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4181 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4182 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4183 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4185 (ix86_const_not_ok_for_debug_p): New function.
4186 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4187 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4189 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4192 * system.h: Conditionally include "unique-ptr.h" if
4193 INCLUDE_UNIQUE_PTR is defined.
4194 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4195 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4197 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4199 * config/rl78/rl78.md: New define_expand "subdi3".
4201 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4204 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4205 DF_REF_INSN if DF_REF_INSN_INFO is false.
4207 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4209 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4210 xmm_move instead of sse_move.
4211 (sse_store_index): New function.
4212 (ix86_register_move_cost): Be more sensible about mismatch stall;
4213 model AVX moves correctly; make difference between sse->integer and
4215 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4216 moves; make difference between SSE and AVX.
4217 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4218 and zmm_move. Increase size of sse load and store tables;
4219 add unaligned load and store tables; add ssemmx_to_integer.
4220 * x86-tune-costs.h: Update all entries according to real
4221 move latencies from Agner Fog's manual and chip documentation.
4223 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4226 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4227 * config/i386/constraints.md (Wf): New constraint.
4228 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4229 (cmp<dwi>_doubleword): Removed.
4230 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4231 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4232 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4233 expand with cmp<dwi>_doubleword. For LTU and GEU use
4234 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4236 * common.opt (gcolumn-info): Enable by default.
4237 * doc/invoke.texi (gcolumn-info): Document new default.
4239 2017-10-23 Richard Biener <rguenther@suse.de>
4241 PR tree-optimization/82672
4242 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4243 Fold the stmt if we propagated into it.
4245 2017-10-23 Richard Biener <rguenther@suse.de>
4247 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4248 (bitmap_remove_expr_from_set): ... this. All callers call this
4249 for non-constant values.
4250 (bitmap_set_subtract): Rename to...
4251 (bitmap_set_subtract_expressions): ... this. Adjust and
4253 (bitmap_set_contains_value): Remove superfluous check.
4254 (bitmap_set_replace_value): Inline into single caller ...
4255 (bitmap_value_replace_in_set): ... here and simplify.
4256 (dependent_clean): Merge into ...
4257 (clean): ... this using an overload. Adjust.
4258 (prune_clobbered_mems): Adjust.
4259 (compute_antic_aux): Likewise.
4260 (compute_partial_antic_aux): Likewise.
4262 2017-10-23 Richard Biener <rguenther@suse.de>
4264 PR tree-optimization/82129
4266 2017-08-01 Richard Biener <rguenther@suse.de>
4268 PR tree-optimization/81181
4269 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4270 (compute_antic): ... end of iteration here.
4272 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4274 * target.def (starting_frame_offset): New hook.
4275 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4276 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4277 * doc/tm.texi.in: Regenerate.
4278 * hooks.h (hook_hwi_void_0): Declare.
4279 * hooks.c (hook_hwi_void_0): New function.
4280 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4281 STARTING_FRAME_OFFSET.
4282 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4283 * reload1.c (reload): Likewise.
4284 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4285 instead of STARTING_FRAME_OFFSET.
4286 * function.c (try_fit_stack_local): Likewise.
4287 (assign_stack_local_1): Likewise
4288 (instantiate_virtual_regs): Likewise.
4289 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4290 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4291 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4292 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4293 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4294 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4295 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4296 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4297 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4298 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4299 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4300 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4301 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4302 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4303 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4304 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4305 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4306 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4307 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4308 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4309 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4310 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4311 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4312 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4313 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4314 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4315 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4316 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4317 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4318 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4319 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4320 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4321 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4322 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4323 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4324 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4325 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4326 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4327 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4328 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4329 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4330 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4331 return a HOST_WIDE_INT.
4332 (avr_builtin_setjmp_frame_value): Use it instead of
4333 STARTING_FRAME_OFFSET.
4334 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4335 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4336 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4338 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4339 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4340 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4341 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4342 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4343 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4344 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4345 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4346 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4347 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4348 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4349 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4351 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4352 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4353 * config/mips/mips.c (mips_compute_frame_info): Refer to
4354 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4355 (mips_starting_frame_offset): New function.
4356 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4357 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4358 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4359 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4360 and return a HOST_WIDE_INT.
4361 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4362 (mmix_initial_elimination_offset): Refer to
4363 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4364 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4365 * config/pa/pa.c (pa_starting_frame_offset): New function.
4366 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4367 (pa_expand_prologue): Likewise.
4368 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4369 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4370 !FRAME_GROWS_DOWNWARD handling to...
4371 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4372 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4373 !FRAME_GROWS_DOWNWARD handling to...
4374 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4375 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4376 !FRAME_GROWS_DOWNWARD handling to...
4377 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4378 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4380 (rs6000_starting_frame_offset): New function.
4381 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4382 !FRAME_GROWS_DOWNWARD handling to...
4383 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4384 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4385 !FRAME_GROWS_DOWNWARD handling to...
4386 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4387 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4388 !FRAME_GROWS_DOWNWARD handling to...
4389 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4390 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4391 (rs6000_starting_frame_offset): New function.
4392 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4393 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4394 * config/vax/vax.c (vax_starting_frame_offset): New function.
4395 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4396 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4397 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4398 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4399 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4400 * system.h (STARTING_FRAME_OFFSET): Poison.
4402 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4404 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4405 SCALAR_TYPE_MODE instead of TYPE_MODE.
4407 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4408 Alan Hayward <alan.hayward@arm.com>
4409 David Sherwood <david.sherwood@arm.com>
4411 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4413 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4414 Alan Hayward <alan.hayward@arm.com>
4415 David Sherwood <david.sherwood@arm.com>
4417 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4419 2017-10-23 Richard Biener <rguenther@suse.de>
4421 PR tree-optimization/82129
4422 * tree-ssa-pre.c (bitmap_set_and): Remove.
4423 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4424 canonicalizing expressions in the set to those with lowest
4425 ID rather than taking that from the first edge.
4427 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4429 * combine.c (rtx_equal_for_field_assignment_p): Use
4430 byte_lowpart_offset.
4432 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4433 Alan Hayward <alan.hayward@arm.com>
4434 David Sherwood <david.sherwood@arm.com>
4436 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4437 to a SUBREG_PROMOTED_VAR.
4439 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4440 Alan Hayward <alan.hayward@arm.com>
4441 David Sherwood <david.sherwood@arm.com>
4443 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4444 (expand_debug_source_expr): Likewise.
4445 * combine.c (combine_simplify_rtx): Likewise.
4446 * cse.c (fold_rtx): Likewise.
4447 * optabs.c (expand_float): Likewise.
4448 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4449 (simplify_binary_operation_1): Likewise.
4451 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4452 Alan Hayward <alan.hayward@arm.com>
4453 David Sherwood <david.sherwood@arm.com>
4455 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4456 (record_promoted_value): Likewise.
4457 * expr.c (expand_expr_real_2): Likewise.
4458 * ree.c (update_reg_equal_equiv_notes): Likewise.
4459 (combine_set_extension): Likewise.
4460 * rtlanal.c (low_bitmask_len): Likewise.
4461 * simplify-rtx.c (neg_const_int): Likewise.
4462 (simplify_binary_operation_1): Likewise.
4464 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4465 Alan Hayward <alan.hayward@arm.com>
4466 David Sherwood <david.sherwood@arm.com>
4468 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4469 * regcprop.c (maybe_mode_change): Likewise.
4470 * reload1.c (alter_reg): Likewise.
4472 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4474 * inchash.h (inchash::hash::add_wide_int): New function.
4475 * lto-streamer-out.c (hash_tree): Use it.
4477 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4479 * inchash.h (inchash::hash::add_wide_int): Rename to...
4480 (inchash::hash::add_hwi): ...this.
4481 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4482 (polymorphic_call_target_hasher::hash): Likewise.
4483 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4484 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4485 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4486 * lto-streamer-out.c (hash_tree): Likewise.
4487 * optc-save-gen.awk: Likewise.
4488 * tree.c (add_expr): Likewise.
4490 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4493 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4494 for ordered inequality comparisons even with TARGET_IEEE_FP.
4496 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4499 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4500 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4501 Expand with cmp<dwi>_doubleword.
4503 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4505 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4506 List CET intrinsics.
4507 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4508 specific to -fcf-protection option.
4510 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4512 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4513 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4514 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4515 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4516 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4517 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4518 (extra_objs): Add cet.o for Linux/x86 targets.
4519 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4520 * config/i386/cet.c: New file.
4521 * config/i386/cetintrin.h: Likewise.
4522 * config/i386/t-cet: Likewise.
4523 * config/i386/cpuid.h (bit_SHSTK): New.
4524 (bit_IBT): Likewise.
4525 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4526 pass IBT and SHSTK bits.
4527 * config/i386/i386-builtin-types.def
4528 (VOID_FTYPE_UNSIGNED_PVOID): New.
4529 (VOID_FTYPE_UINT64_PVOID): Likewise.
4530 * config/i386/i386-builtin.def: Add CET intrinsics.
4531 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4532 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4533 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4534 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4536 * config/i386/i386.c (rest_of_insert_endbranch): New.
4537 (pass_data_insert_endbranch): Likewise.
4538 (pass_insert_endbranch): Likewise.
4539 (make_pass_insert_endbranch): Likewise.
4540 (ix86_notrack_prefixed_insn_p): Likewise.
4541 (ix86_target_string): Add -mibt, -mshstk flags.
4542 (ix86_option_override_internal): Add flag_cf_protection
4544 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4545 (ix86_print_operand): Add 'notrack' prefix output.
4546 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4547 (ix86_expand_builtin): Expand CET intrinsics.
4548 (x86_output_mi_thunk): Add 'endbranch' instruction.
4549 * config/i386/i386.h (TARGET_IBT): New.
4550 (TARGET_IBT_P): Likewise.
4551 (TARGET_SHSTK): Likewise.
4552 (TARGET_SHSTK_P): Likewise.
4553 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4554 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4555 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4556 (builtin_setjmp_setup): New pattern.
4557 (builtin_longjmp): Likewise.
4558 (rdssp<mode>): Likewise.
4559 (incssp<mode>): Likewise.
4560 (saveprevssp): Likewise.
4561 (rstorssp): Likewise.
4562 (wrss<mode>): Likewise.
4563 (wruss<mode>): Likewise.
4564 (setssbsy): Likewise.
4565 (clrssbsy): Likewise.
4566 (nop_endbr): Likewise.
4567 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4569 * config/i386/immintrin.h: Include <cetintrin.h>.
4570 * config/i386/linux-common.h
4571 (file_end_indicate_exec_stack_and_cet): New prototype.
4572 (TARGET_ASM_FILE_END): New.
4574 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4576 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4577 latencies instead of having separate table; make difference between
4578 integer and float costs.
4579 * i386.h (processor_costs): Remove scalar_stmt_cost,
4580 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4581 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4583 * x86-tune-costs.h: Remove entries which has been removed in
4584 procesor_costs from all tables; make cond_taken_branch_cost
4585 and cond_not_taken_branch_cost COST_N_INSNS based.
4587 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4589 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4591 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4593 * config/i386/i386.md (isa): Remove fma_avx512f.
4594 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4595 <avx512>_fmadd_<mode>_mask3<round_name>,
4596 <avx512>_fmsub_<mode>_mask<round_name>,
4597 <avx512>_fmsub_<mode>_mask3<round_name>,
4598 <avx512>_fnmadd_<mode>_mask<round_name>,
4599 <avx512>_fnmadd_<mode>_mask3<round_name>,
4600 <avx512>_fnmsub_<mode>_mask<round_name>,
4601 <avx512>_fnmsub_<mode>_mask3<round_name>,
4602 <avx512>_fmaddsub_<mode>_mask<round_name>,
4603 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4604 <avx512>_fmsubadd_<mode>_mask<round_name>,
4605 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4606 (*vec_widen_umult_even_v16si<mask_name>,
4607 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4608 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4610 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4612 * extend.texi: Add 'nocf_check' documentation.
4613 * gimple.texi: Add second parameter to
4614 gimple_build_call_from_tree.
4615 * invoke.texi: Add -fcf-protection documentation.
4616 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4618 2017-10-20 Richard Biener <rguenther@suse.de>
4620 PR tree-optimization/82473
4621 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4622 the largest input type.
4624 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4626 * c-attribs.c (handle_nocf_check_attribute): New function.
4627 (c_common_attribute_table): Add 'nocf_check' handling.
4628 * gimple-parser.c: Add second argument NULL to
4629 gimple_build_call_from_tree.
4630 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4631 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4633 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4634 * common.opt: Add fcf-protection flag.
4635 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4636 * flag-types.h: Add enum cf_protection_level.
4637 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4638 Add 'nocf_check' attribute propagation to gimple call.
4639 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4640 (gimple_build_call_from_tree): Update prototype.
4641 (gimple_call_nocf_check_p): New function.
4642 (gimple_call_set_nocf_check): Likewise.
4643 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4644 * ipa-icf.c: Add nocf_check attribute in statement hash.
4645 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4646 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4647 * toplev.c (process_options): Add flag_cf_protection handling.
4649 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4651 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4653 2017-10-20 Richard Biener <rguenther@suse.de>
4655 PR tree-optimization/82603
4656 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4657 remove false predicated stores.
4659 2017-10-20 Richard Biener <rguenther@suse.de>
4661 * graphite-isl-ast-to-gimple.c
4662 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4663 Remove return value and simplify, dump copied stmt after lhs
4665 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4666 Reduce dump verbosity.
4667 (gsi_insert_earliest): Likewise.
4668 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4669 * graphite.c (print_global_statistics): Adjust dumping.
4670 (print_graphite_scop_statistics): Likewise.
4671 (print_graphite_statistics): Do not dump loops here.
4672 (graphite_transform_loops): But here.
4674 2017-10-20 Nicolas Roche <roche@adacore.com>
4676 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4677 * configure: Regenerate.
4679 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4682 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4683 functions when optimizing replace GIMPLE_RETURN stmts with
4684 calls to __builtin_unreachable ().
4687 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4688 for -fsanitize=thread link of executables.
4689 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4690 link of executables.
4693 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4695 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4696 define_insns for logical vector shifts to use VI248_AVX512BW
4697 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4698 condition, useless isa and prefix attributes. Change the first
4699 2 of these define_insns to ...
4700 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4701 define_insn for avx512vl.
4702 (<shift_insn><mode>3): ... and this, new define_insn without
4703 masking for non-avx512vl.
4706 * config/i386/sse.md (*andnot<mode>3,
4707 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4708 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4709 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4710 not applied use empty suffix even for TARGET_AVX512VL.
4711 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4712 is applied, supply evex,evex or evex,evex,evex instead of just
4715 2017-10-20 Julia Koval <julia.koval@intel.com>
4717 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4718 (OPTION_MASK_ISA_GFNI_UNSET): New.
4719 (ix86_handle_option): Handle OPT_mgfni.
4720 * config/i386/cpuid.h (bit_GFNI): New.
4721 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4722 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4723 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4724 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4725 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4726 * config/i386/i386.opt: Add mgfni.
4728 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4730 * config/msp430/msp430.c (msp430_option_override): Disable
4731 -fdelete-null-pointer-checks.
4732 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4734 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4736 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4737 of x87 and SSE instructions.
4739 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4741 * asan.c (create_cond_insert_point): Do not update edge count.
4742 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4743 (afdo_propagate_circuit): Likewise.
4744 (afdo_calculate_branch_prob): Likewise.
4745 (afdo_annotate_cfg): Likewise.
4746 * basic-block.h (struct edge_def): Remove count.
4747 (edge_def::count): New accessor.
4748 * bb-reorder.c (rotate_loop): Update.
4749 (find_traces_1_round): Update.
4750 (connect_traces): Update.
4751 (sanitize_hot_paths): Update.
4752 * cfg.c (unchecked_make_edge): Update.
4753 (make_single_succ_edge): Update.
4754 (check_bb_profile): Update.
4755 (dump_edge_info): Update.
4756 (update_bb_profile_for_threading): Update.
4757 (scale_bbs_frequencies_int): Update.
4758 (scale_bbs_frequencies_gcov_type): Update.
4759 (scale_bbs_frequencies_profile_count): Update.
4760 (scale_bbs_frequencies): Update.
4761 * cfganal.c (connect_infinite_loops_to_exit): Update.
4762 * cfgbuild.c (compute_outgoing_frequencies): Update.
4763 (find_many_sub_basic_blocks): Update.
4764 * cfgcleanup.c (try_forward_edges): Update.
4765 (try_crossjump_to_edge): Update
4766 * cfgexpand.c (expand_gimple_cond): Update
4767 (expand_gimple_tailcall): Update
4768 (construct_exit_block): Update
4769 * cfghooks.c (verify_flow_info): Update
4770 (redirect_edge_succ_nodup): Update
4771 (split_edge): Update
4772 (make_forwarder_block): Update
4773 (duplicate_block): Update
4774 (account_profile_record): Update
4775 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4776 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4777 * cfgloopmanip.c (scale_loop_profile): Update.
4779 (lv_adjust_loop_entry_edge): Update.
4780 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4781 (force_nonfallthru_and_redirect): Update.
4782 (purge_dead_edges): Update.
4783 (rtl_flow_call_edges_add): Update.
4784 * cgraphunit.c (init_lowered_empty_function): Update.
4785 (cgraph_node::expand_thunk): Update.
4786 * gimple-pretty-print.c (dump_probability): Update.
4787 (dump_edge_probability): Update.
4788 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4789 * haifa-sched.c (sched_create_recovery_edges): Update.
4790 * hsa-gen.c (convert_switch_statements): Update.
4791 * ifcvt.c (dead_or_predicable): Update.
4792 * ipa-inline-transform.c (inline_transform): Update.
4793 * ipa-split.c (split_function): Update.
4794 * ipa-utils.c (ipa_merge_profiles): Update.
4795 * loop-doloop.c (add_test): Update.
4796 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4797 * lto-streamer-in.c (input_cfg): Update.
4798 (input_function): Update.
4799 * lto-streamer-out.c (output_cfg): Update.
4800 * modulo-sched.c (sms_schedule): Update.
4801 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4802 * predict.c (maybe_hot_edge_p): Update.
4803 (unlikely_executed_edge_p): Update.
4804 (probably_never_executed_edge_p): Update.
4805 (dump_prediction): Update.
4806 (drop_profile): Update.
4807 (propagate_unlikely_bbs_forward): Update.
4808 (determine_unlikely_bbs): Update.
4809 (force_edge_cold): Update.
4810 * profile.c (compute_branch_probabilities): Update.
4811 * reg-stack.c (better_edge): Update.
4812 * shrink-wrap.c (handle_simple_exit): Update.
4813 * tracer.c (better_p): Update.
4814 * trans-mem.c (expand_transaction): Update.
4815 (split_bb_make_tm_edge): Update.
4816 * tree-call-cdce.c: Update.
4817 * tree-cfg.c (gimple_find_sub_bbs): Update.
4818 (gimple_split_edge): Update.
4819 (gimple_duplicate_sese_region): Update.
4820 (gimple_duplicate_sese_tail): Update.
4821 (gimple_flow_call_edges_add): Update.
4822 (insert_cond_bb): Update.
4823 (execute_fixup_cfg): Update.
4824 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4825 * tree-complex.c (expand_complex_div_wide): Update.
4826 * tree-eh.c (lower_resx): Update.
4827 (unsplit_eh): Update.
4828 (cleanup_empty_eh_move_lp): Update.
4829 * tree-inline.c (copy_edges_for_bb): Update.
4830 (freqs_to_counts): Update.
4831 (copy_cfg_body): Update.
4832 * tree-ssa-dce.c (remove_dead_stmt): Update.
4833 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4834 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4835 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4836 (unloop_loops): Update.
4837 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4838 * tree-ssa-loop-split.c (connect_loops): Update.
4839 (split_loop): Update.
4840 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4841 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4842 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4843 * tree-ssa-reassoc.c (branch_fixup): Update.
4844 * tree-ssa-tail-merge.c (replace_block_by): Update.
4845 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4846 (compute_path_counts): Update.
4847 (update_profile): Update.
4848 (recompute_probabilities): Update.
4849 (update_joiner_offpath_counts): Update.
4850 (estimated_freqs_path): Update.
4851 (freqs_to_counts_path): Update.
4852 (clear_counts_path): Update.
4853 (ssa_fix_duplicate_block_edges): Update.
4854 (duplicate_thread_path): Update.
4855 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4856 (case_bit_test_cmp): Update.
4857 (collect_switch_conv_info): Update.
4858 (gen_inbound_check): Update.
4859 (do_jump_if_equal): Update.
4860 (emit_cmp_and_jump_insns): Update.
4861 * tree-tailcall.c (decrease_profile): Update.
4862 (eliminate_tail_call): Update.
4863 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4864 (vect_do_peeling): Update.
4865 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4866 * ubsan.c (ubsan_expand_null_ifn): Update.
4867 (ubsan_expand_ptr_ifn): Update.
4868 * value-prof.c (gimple_divmod_fixed_value): Update.
4869 (gimple_mod_pow2): Update.
4870 (gimple_mod_subtract): Update.
4871 (gimple_ic): Update.
4872 (gimple_stringop_fixed_value): Update.
4874 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4877 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4879 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4881 PR rtl-optimization/82395
4882 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4883 based on non_spilled_static_chain_regno_p.
4885 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4887 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4888 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4889 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4891 2017-10-19 Martin Sebor <msebor@redhat.com>
4893 PR tree-optimization/82596
4894 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4896 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4898 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4899 (get_mem_refs_of_builtin_call): Likewise.
4900 * builtins.c (expand_builtin_apply): Adjust call to
4901 allocate_dynamic_stack_space.
4902 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4903 the third argument to allocate_dynamic_stack_space, otherwise -1.
4904 (expand_builtin): Deal with all alloca variants.
4905 (is_inexpensive_builtin): Likewise.
4906 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4907 * calls.c (special_function_p): Deal with all alloca variants.
4908 (initialize_argument_information): Adjust call to
4909 allocate_dynamic_stack_space.
4910 (expand_call): Likewise.
4911 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4912 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4913 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4914 use it for the stack usage computation.
4915 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4916 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4917 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4918 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4919 (in_loop_p): Remove first argument and useless check.
4920 (pass_walloca::execute): Remove useless test and adjust call to above.
4921 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4922 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4923 (gimplify_call_expr): Deal with all alloca variants.
4924 * hsa-gen.c (gen_hsa_alloca): Likewise.
4925 (gen_hsa_insns_for_call): Likewise.
4926 * ipa-pure-const.c (special_builtin_state): Likewise.
4927 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4928 * tree-object-size.c (alloc_object_size): Likewise.
4929 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4930 (call_may_clobber_ref_p_1): Likewise.
4931 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4932 (ccp_fold_stmt): Likewise.
4933 (optimize_stack_restore): Likewise.
4934 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4935 (mark_all_reaching_defs_necessary_1): Likewise.
4936 (propagate_necessity): Likewise.
4937 (eliminate_unnecessary_stmts): Likewise.
4938 * tree.c (build_common_builtin_nodes): Build
4939 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4940 (build_alloca_call_expr): New function.
4941 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4942 (CASE_BUILT_IN_ALLOCA): Likewise.
4943 (build_alloca_call_expr): Declare.
4944 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4946 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4949 * dwarf2out.c (new_die_raw): New static inline function.
4950 (new_die): Use it to create the DIE.
4951 (add_AT_external_die_ref): Likewise.
4952 (clone_die): Likewise.
4953 (clone_as_declaration): Likewise.
4954 (dwarf2out_vms_debug_main_pointer): Likewise.
4955 (base_type_die): Likewise. Remove early return for corner cases.
4956 Do not call add_pubtype on the DIE here.
4957 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4958 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4959 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4960 native order exists for base types, attach the DIE manually and call
4961 add_pubtype on it. Do not equate a reverse order DIE to the type.
4963 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4965 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4966 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4968 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4970 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4972 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4973 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4974 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4975 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4976 * opts.c (sanitizer_opts): Add builtin.
4977 * ubsan.c (instrument_builtin): New function.
4978 (pass_ubsan::execute): Call it.
4979 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4980 * doc/invoke.texi: Document -fsanitize=builtin.
4982 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4983 builtins, store max (log2 (align), 0) into uchar field instead of
4984 align into uptr field.
4985 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4986 store uchar 0 field instead of uptr 0 field.
4987 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4988 instead of passing one address of struct with 2 locations pass
4989 two addresses of structs with 1 location each.
4990 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4991 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4992 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4993 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4994 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4995 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4996 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4997 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4999 2017-10-19 Martin Liska <mliska@suse.cz>
5002 * file-find.c (remove_prefix): Remove.
5003 * file-find.h (remove_prefix): Likewise.
5004 * gcc-ar.c: Remove smartness of lookup.
5006 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5008 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5009 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5010 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5012 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5015 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5016 (setcc + and to xor + setcc): New peephole2.
5018 2017-10-19 Tom de Vries <tom@codesourcery.com>
5020 * doc/sourcebuild.texi (Test Directives, Variants of
5021 dg-require-support): Add dg-require-stack-size.
5023 2017-10-19 Martin Liska <mliska@suse.cz>
5026 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5027 that have a large alignment.
5028 (gimplify_target_expr): Likewise.
5030 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5032 PR rtl-optimization/82602
5033 * ira.c (rtx_moveable_p): Return false for volatile asm.
5035 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5038 * config/i386/i386-modes.def (CCGZ): New CC mode.
5039 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5040 * config/i386/predicates.md (ix86_comparison_operator):
5042 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5043 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5044 with double-word subtraction.
5045 (put_condition_code): Handle CCGZmode.
5047 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5049 * wide-int.cc (debug (const wide_int &)): New.
5050 (debug (const wide_int *)): New.
5051 (debug (const widest_int &)): New.
5052 (debug (const widest_int *)): New.
5054 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5057 * lra-constraints.c (curr_insn_transform): Use non-input operand
5058 instead of output one for matched reload.
5060 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5062 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5063 (tree-ssa-loop-ivopts.h): New header file.
5064 (struct builtin_info): New fields.
5065 (classify_builtin_1): Compute and record base and offset parts for
5066 memset builtin partition by calling strip_offset.
5067 (offset_cmp, fuse_memset_builtins): New functions.
5068 (finalize_partitions): Fuse adjacent memset partitions by calling
5070 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5071 Expose the interface.
5072 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5074 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5076 PR tree-optimization/82574
5077 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5078 that data reference must be executed exactly once per iteration
5079 against the outermost loop in nest.
5080 (classify_partition): Update call to above function.
5082 2017-10-18 Richard Biener <rguenther@suse.de>
5084 PR tree-optimization/82591
5085 * graphite.c (graphite_transform_loops): Move code gen message
5087 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5088 Here. Handle scop_to_isl_ast failing.
5089 (scop_to_isl_ast): Limit the number of ISL operations.
5091 2017-10-18 Richard Biener <rguenther@suse.de>
5093 * graphite-isl-ast-to-gimple.c
5094 (translate_isl_ast_to_gimple::set_rename): Simplify.
5095 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5096 (graphite_copy_stmts_from_block): ... here.
5097 (copy_bb_and_scalar_dependences): Simplify.
5098 (add_parameters_to_ivs_params): Canonicalize.
5099 (generate_entry_out_of_ssa_copies): Simplify.
5100 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5101 by passing in ISL dimension.
5102 (parameter_index_in_region_1): Rename to ...
5103 (parameter_index_in_region): ... this.
5104 (extract_affine): Adjust assert, pass down parameter index.
5105 (add_param_constraints): Use range-info when available.
5106 (build_scop_context): Adjust.
5107 * sese.c (new_sese_info): Adjust.
5108 (free_sese_info): Likewise.
5109 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5110 Remove unused typedefs.
5111 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5113 2017-10-18 Martin Liska <mliska@suse.cz>
5115 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5117 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5119 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5120 (s390_sched_init): Do not reset s390_sched_state if we entered the
5121 current basic block via a fallthru edge and all others are unlikely.
5123 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5125 * config/s390/s390.c (NUM_SIDES): New variable.
5126 (LONGRUNNING_THRESHOLD): New variable.
5127 (LATENCY_FACTOR): New variable.
5128 (s390_sched_score): Decrease score for long-running instructions on
5130 (s390_sched_variable_issue): Perform bookkeeping for long-running
5133 2017-10-18 Richard Biener <rguenther@suse.de>
5135 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5136 Simplify with removal of the parameter rename map.
5137 (set_rename): Likewise.
5138 (should_copy_to_new_region): Likewise.
5139 (graphite_copy_stmts_from_block): Likewise.
5140 (copy_bb_and_scalar_dependences): Remove initialization of
5141 unused copied_bb_map.
5143 (copy_internal_parameters): Likewise.
5144 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5145 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5146 Use INTEGRAL_TYPE_P.
5147 (parameter_index_in_region_1): Rename to ...
5148 (assign_parameter_index_in_region): ... this. Assert we have
5149 a parameter we handle.
5150 (scan_tree_for_params): Adjust.
5151 * sese.h (parameter_rename_map_t): Remove.
5152 (struct sese_info_t): Remove unused parameter_rename_map and
5153 copied_bb_map members.
5154 * sese.c (new_sese_info): Adjust.
5155 (free_sese_info): Likewise.
5157 2017-10-18 Martin Liska <mliska@suse.cz>
5160 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5161 on an abnormal edge.
5163 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5165 * doc/invoke.texi (ffunction-sections and fdata-sections):
5168 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5170 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5171 the use statement can throw internally.
5173 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5175 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5176 any RTX present on the RHS of a SET.
5177 * compare-elim.c (try_eliminate_compare): Restore comment.
5179 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5181 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5182 may be also called on tcc_type nodes.
5183 * langhooks.c (lhd_tree_size): Likewise.
5185 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5187 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5188 format_warning_at_substring.
5189 (maybe_warn): Convert source_range * param to a location_t. Pass
5190 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5191 (format_directive): Remove code to extract source_ranges and
5192 source_range * in favor of just a location_t.
5193 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5195 * substring-locations.c (format_warning_va): Convert
5196 source_range * param to a location_t.
5197 (format_warning_at_substring): Likewise.
5198 * substring-locations.h (format_warning_va): Likewise.
5199 (format_warning_at_substring): Likewise.
5201 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5203 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5205 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5206 and scatter/gather ops.
5208 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5209 vec_gather_load and vec_scatter_store.
5210 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5211 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5212 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5213 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5214 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5215 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5217 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5219 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5220 Detect FTST instruction and handle its register pops. Only pop
5221 second operand if can_pop_second_op is true.
5222 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5223 set can_pop_second_op to false in the compare_for_stack_reg call.
5225 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5226 output_fp_compare for stack register operands.
5227 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5228 instructions here. Do not emit stack register pops here. Assert
5229 that FCOMPP pops next to top stack register. Rewrite function.
5231 2017-10-17 Nathan Sidwell <nathan@acm.org>
5234 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5235 use DECL_ASSEMBLER_NAME_RAW.
5238 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5241 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5242 Wilco Dijkstra <wilco.dijkstra@arm.com>
5244 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5245 converstion to Pmode from the buf_addr.
5247 2017-10-17 Richard Biener <rguenther@suse.de>
5249 * graphite-dependences.c (scop_get_reads_and_writes): Change
5250 output parameters to references.
5252 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5254 PR 71026/tree-optimization
5255 * fold-const.c (distribute_real_division): Removed.
5256 (fold_binary_loc): Remove calls to distribute_real_divison.
5258 2017-10-17 Richard Biener <rguenther@suse.de>
5260 * graphite-scop-detection.c
5261 (scop_detection::stmt_has_simple_data_refs_p): Always use
5262 the full nest as region.
5263 (try_generate_gimple_bb): Likewise.
5264 * sese.c (scalar_evolution_in_region): Simplify now that
5265 SCEV can handle instantiation in regions.
5266 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5267 in the non-loop part of a function if requested.
5269 2017-10-17 Richard Biener <rguenther@suse.de>
5271 PR tree-optimization/82563
5272 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5274 (graphite_regenerate_ast_isl): Call it.
5275 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5277 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5279 PR tree-optimization/82549
5280 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5281 Formatting fixes. Instead of calling make_bit_field_ref with negative
5284 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5286 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5287 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5288 _mm_maskz_reduce_ss): New.
5289 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5290 __builtin_ia32_reducess_mask): Ditto..
5291 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5292 * config/i386/sse.md (reduces<mode>): Renamed to ...
5293 (reduces<mode><mask_scalar_name>): ... this.
5294 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5296 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5297 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5299 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5301 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5302 * selftest-run-tests.c (selftest::run_tests): Call
5303 selftest::unique_ptr_tests_cc_tests.
5304 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5305 * unique-ptr-tests.cc: New file.
5307 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5310 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5312 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5313 (make_hard_regno_born, make_hard_regno_dead): Update
5314 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5316 2017-10-16 Jeff Law <law@redhat.com>
5318 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5320 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5322 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5324 2017-10-16 Olivier Hainque <hainque@adacore.com>
5326 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5327 with_cpu if we were configured for an e500v2 target cpu name.
5329 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5331 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5332 * doc/invoke.texi: Document +nodsp as a valid extension for
5335 2017-10-16 Martin Liska <mliska@suse.cz>
5337 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5338 (test_set_range): Likewise.
5339 (test_range_functions): Rename to ...
5340 (test_bit_in_range): ... this.
5341 (sbitmap_c_tests): Add new test.
5343 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5345 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5347 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5348 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5350 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5352 * config/aarch64/aarch64-builtins.c
5353 (aarch64_types_quadopu_lane_qualifiers): New.
5354 (TYPES_QUADOPU_LANE): New.
5355 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5356 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5357 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5358 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5359 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5360 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5361 (Vdottype, DOTPROD): New.
5362 (sur): Add SDOT and UDOT.
5364 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5366 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5367 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5368 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5370 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5371 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5372 Enable TARGET_DOTPROD.
5373 (cortex-a75.cortex-a55): Likewise.
5374 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5376 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5378 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5379 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5381 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5383 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5384 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5385 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5386 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5387 * config/arm/types.md (neon_dot, neon_dot_q): New.
5388 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5390 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5392 * config/arm/arm.h (TARGET_DOTPROD): New.
5393 * config/arm/arm.c (arm_arch_dotprod): New.
5394 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5395 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5396 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5397 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5398 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5399 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5400 * doc/invoke.texi (armv8.2-a): Document dotprod
5402 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5404 * i386.c (ix86_vec_cost): New function.
5405 (ix86_rtx_costs): Handle vector operations better.
5406 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5407 * x86-tune-costs.h: Add new costs to all tables.
5409 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5411 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5413 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5414 divsd, sqrtss and sqrtsd
5415 * x86-tune-costs.h: Add new entries to all costs.
5416 (znver1_cost): Fix to match real instruction latencies.
5418 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5419 Michael Collison <michael.collison@arm.com>
5421 * compare-elim.c: Include emit-rtl.h.
5422 (can_merge_compare_into_arith): New function.
5423 (try_validate_parallel): Likewise.
5424 (try_merge_compare): Likewise.
5425 (try_eliminate_compare): Call the above when no previous clobber
5427 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5430 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5434 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5435 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5439 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5440 to be any operand_equal_p operands. For & (B - 1) require
5441 B to be power of 2. Recognize
5442 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5444 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5447 * optabs.c (expand_memory_blockage): Fix call of
5448 targetm.have_memory_blockage.
5450 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5453 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5454 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5455 objects to extra_objs instead of overwriting it.
5457 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5459 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5460 Use any_fp_register_operand as operand[3] predicate. Simplify
5461 equality test for operands[2] and operands[4] memory location.
5462 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5463 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5464 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5465 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5466 any_fp_register_operand as operand[1] predicate. Simplify
5467 equality test for operands[0] and operands[3] memory location.
5468 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5469 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5470 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5472 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5474 * target-insns.def: Add memory_blockage.
5475 * optabs.c (expand_memory_blockage): New function.
5476 (expand_asm_memory_barrier): Rename ...
5477 (expand_asm_memory_blockage): ... to this.
5478 (expand_mem_thread_fence): Call expand_memory_blockage
5479 instead of expand_asm_memory_barrier.
5480 (expand_mem_singnal_fence): Ditto.
5481 (expand_atomic_load): Ditto.
5482 (expand_atomic_store): Ditto.
5483 * doc/md.texi (Standard Pattern Names For Generation):
5484 Document memory_blockage instruction pattern.
5486 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5488 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5489 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5490 * config/rl78/rl78.md: New define_expand "adddi3".
5492 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5494 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5497 2017-10-13 Jeff Law <law@redhat.com>
5499 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5501 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5504 * internal-fn.c (expand_mul_overflow): If both operands have
5505 the same highpart of -1 or 0 and the topmost bit of lowpart
5506 is different, overflow is if res <= 0 rather than res < 0.
5508 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5510 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5511 TARGET_P9_VECTOR code for unaligned_load case.
5513 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5515 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5517 2017-10-13 Nathan Sidwell <nathan@acm.org>
5519 * tree-core.h (tree_contains_struct): Make bool.
5520 * tree.c (tree_contains_struct): Likewise.
5521 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5522 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5523 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5524 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5525 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5527 2017-10-13 Richard Biener <rguenther@suse.de>
5529 * graphite-isl-ast-to-gimple.c
5530 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5531 parameters and dominance check.
5532 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5533 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5534 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5535 Do not update SSA form here or do intermediate IL verification.
5536 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5537 (graphite_initialize): Remove check on the number of loops in
5538 the function and inline into graphite_transform_loops.
5539 (graphite_finalize): Inline into graphite_transform_loops.
5540 (graphite_transform_loops): Perform SSA update and IL verification
5542 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5544 2017-10-13 Richard Biener <rguenther@suse.de>
5546 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5547 graphite_expression_type_precision): Avoid global constructor
5549 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5550 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5551 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5552 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5553 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5554 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5556 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5559 * config/i386/i386.h (ix86_red_zone_size): New.
5560 * config/i386/i386.md (push peephole2s): Replace
5561 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5563 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5564 Alan Hayward <alan.hayward@arm.com>
5565 David Sherwood <david.sherwood@arm.com>
5567 * combine.c (can_change_dest_mode): Reject changes in
5568 REGMODE_NATURAL_SIZE.
5570 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5571 Alan Hayward <alan.hayward@arm.com>
5572 David Sherwood <david.sherwood@arm.com>
5574 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5575 (expand_debug_source_expr): Likewise.
5576 * combine.c (combine_simplify_rtx): Likewise.
5577 * cse.c (fold_rtx): Likewise.
5578 * fwprop.c (canonicalize_address): Likewise.
5579 * targhooks.c (default_shift_truncation_mask): Likewise.
5581 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5582 Alan Hayward <alan.hayward@arm.com>
5583 David Sherwood <david.sherwood@arm.com>
5585 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5586 (widened_mode): Likewise.
5587 (expand_unop): Likewise.
5588 * ree.c (transform_ifelse): Likewise.
5589 (merge_def_and_ext): Likewise.
5590 (combine_reaching_defs): Likewise.
5591 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5593 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5594 Alan Hayward <alan.hayward@arm.com>
5595 David Sherwood <david.sherwood@arm.com>
5597 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5598 * combine.c (gen_lowpart_for_combine): Likewise.
5599 * dwarf2out.c (rtl_for_decl_location): Likewise.
5600 * final.c (alter_subreg): Likewise.
5601 * rtlhooks.c (gen_lowpart_general): Likewise.
5602 (gen_lowpart_if_possible): Likewise.
5604 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5605 Alan Hayward <alan.hayward@arm.com>
5606 David Sherwood <david.sherwood@arm.com>
5608 * calls.c (expand_call): Use subreg_lowpart_offset.
5609 * cse.c (cse_insn): Likewise.
5610 * regcprop.c (copy_value): Likewise.
5611 (copyprop_hardreg_forward_1): Likewise.
5613 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5616 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5617 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5618 =Q constraints to +Q and into insn condition add check
5619 that operands[0] and operands[1] are equal.
5620 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5621 =Q constraints to +Q and into insn condition add check
5622 that operands[0] is equal to either operands[1] or operands[2].
5625 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5626 instead of handling MINUS_EXPR twice (once for each argument),
5627 canonicalize operand order and handle just once, use rtype where
5628 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5631 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5632 any values of __C while still being pattern recognizable as a simple
5635 2017-10-13 Richard Biener <rguenther@suse.de>
5637 PR tree-optimization/82451
5639 2017-10-02 Richard Biener <rguenther@suse.de>
5641 PR tree-optimization/82355
5642 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5643 a mapping for the enclosing loop but avoid generating one for
5645 (copy_bb_and_scalar_dependences): Remove premature codegen
5646 error on PHIs in blocks duplicated into multiple places.
5647 * graphite-scop-detection.c
5648 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5649 in the region use it as loop and nest to analyze the DR in.
5650 (try_generate_gimple_bb): Likewise.
5651 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5652 (add_loop_constraints): For blocks in a loop not in the region
5653 create a dimension with a single iteration.
5654 * sese.h (gbb_loop_at_index): Remove assert.
5656 * cfgloop.c (loop_preheader_edge): For the loop tree root
5657 return the single successor of the entry block.
5658 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5659 Reset the SCEV hashtable and niters.
5660 * graphite-scop-detection.c
5661 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5662 assert that we only have POLYNOMIAL_CHREC that vary in loops
5663 contained in the region.
5664 (scop_detection::graphite_can_represent_expr): Adjust.
5665 (scop_detection::stmt_has_simple_data_refs_p): For loops
5666 not in the region set loop to NULL. The nest is now the
5667 entry edge to the region.
5668 (try_generate_gimple_bb): Likewise.
5669 * sese.c (scalar_evolution_in_region): Adjust for
5670 instantiate_scev change.
5671 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5672 Make nest parameter the edge into the region.
5673 (create_data_ref): Likewise.
5674 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5675 entry edge into a region and adjust instantiate_scev calls.
5676 (create_data_ref): Likewise.
5677 (graphite_find_data_references_in_stmt): Likewise.
5678 (find_data_references_in_stmt): Pass the loop preheader edge
5679 from the nest argument.
5680 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5681 parameter the edge into the region.
5682 (instantiate_parameters): Use the loop preheader edge as entry.
5683 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5685 (get_instantiated_value_entry): Make instantiate_below parameter
5686 the edge into the region.
5687 (instantiate_scev_name): Likewise. Adjust dominance checks,
5688 when we cannot use loop-based instantiation instantiate by
5689 walking use-def chains.
5690 (instantiate_scev_poly): Adjust.
5691 (instantiate_scev_binary): Likewise.
5692 (instantiate_scev_convert): Likewise.
5693 (instantiate_scev_not): Likewise.
5694 (instantiate_array_ref): Remove.
5695 (instantiate_scev_3): Likewise.
5696 (instantiate_scev_2): Likewise.
5697 (instantiate_scev_1): Likewise.
5698 (instantiate_scev_r): Do not blindly handle N-operand trees.
5699 Do not instantiate array-refs. Handle all constants and invariants.
5700 (instantiate_scev): Make instantiate_below parameter
5701 the edge into the region.
5702 (resolve_mixers): Use the loop preheader edge for the region
5703 parameter to instantiate_scev_r.
5704 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5706 2017-10-13 Richard Biener <rguenther@suse.de>
5708 PR tree-optimization/82525
5709 * graphite-isl-ast-to-gimple.c
5710 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5712 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5713 Fail code generation when we cannot represent the isl integer.
5714 (binary_op_to_tree): Elide modulo operations that are no-ops
5715 in the type we code generate. Remove now superfluous code
5718 2017-10-13 Richard Biener <rguenther@suse.de>
5720 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5721 (scop_detection::harmful_loop_in_region): Remove premature
5722 IV type restriction.
5723 (scop_detection::graphite_can_represent_scev): We can handle
5724 pointer IVs just fine.
5726 2017-10-13 Alan Modra <amodra@gmail.com>
5728 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5729 "Clobbers and Scratch Registers". Add paragraph on
5730 alternative to clobbers for scratch registers and OpenBLAS
5733 2017-10-13 Alan Modra <amodra@gmail.com>
5735 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5736 example of a memory input for a string of known length. Move
5737 commentary out of table. Add a number of new examples
5738 covering array memory inputs.
5740 2017-10-12 Martin Liska <mliska@suse.cz>
5742 PR tree-optimization/82493
5743 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5744 (test_range_functions): New function.
5745 (sbitmap_c_tests): Likewise.
5746 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5747 * selftest.h (sbitmap_c_tests): New function.
5749 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5751 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5753 * config/rs6000/amo.h: Fix spacing issue.
5755 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5758 * config/i386/i386.md (*ashl<mode>3_mask_1,
5759 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5760 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5763 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5765 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5766 (profile_probability): Set max_probability
5767 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5769 * profile-count.c (profile_probability::differs_from_p): Do not
5770 rely on max_probaiblity == 10000
5772 2017-10-12 Jeff Law <law@redhat.com>
5774 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5777 2017-10-12 Martin Sebor <msebor@redhat.com>
5781 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5782 (handle_alias_pairs): Call it.
5783 * common.opt (-Wattribute-alias): New option.
5784 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5785 * doc/invoke.texi (-Wattribute-alias): Document.
5787 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5790 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5792 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5794 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5795 (make_hard_regno_born, make_hard_regno_dead): Update
5796 bb_killed_pseudos and bb_gen_pseudos.
5798 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5800 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5802 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5804 * config/alpha/alpha.c (alpha_split_conditional_move):
5805 Use std::swap instead of manually swapping.
5806 (alpha_stdarg_optimize_hook): Ditto.
5807 (alpha_canonicalize_comparison): Ditto.
5809 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5811 * tree-loop-distribution.c (struct builtin_info): New struct.
5812 (struct partition): Refactor fields into struct builtin_info.
5813 (partition_free): Free struct builtin_info.
5814 (build_size_arg_loc, build_addr_arg_loc): Delete.
5815 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5816 information from struct builtin_info.
5817 (find_single_drs): New function refactored from classify_partition.
5818 Also moved builtin validity checks to this function.
5819 (compute_access_range, alloc_builtin): New functions.
5820 (classify_builtin_st, classify_builtin_ldst): New functions.
5821 (classify_partition): Refactor code into functions find_single_drs,
5822 classify_builtin_st and classify_builtin_ldst.
5823 (distribute_loop): Don't do runtime alias check when distributing
5825 (find_seed_stmts_for_distribution): New function.
5826 (pass_loop_distribution::execute): Refactor code finding seed
5827 stmts into above function. Support distribution for the innermost
5828 two-level loop nest. Adjust dump information.
5830 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5832 * tree-loop-distribution.c: Adjust the general comment.
5833 (NUM_PARTITION_THRESHOLD): New macro.
5834 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5835 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5836 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5837 in call to build_partition_graph.
5838 (finalize_partitions): New parameter. Make loop distribution more
5839 conservative by fusing more partitions.
5840 (distribute_loop): Don't do runtime alias check in case of loop nest
5842 (find_seed_stmts_for_distribution): New function.
5843 (prepare_perfect_loop_nest): New function.
5844 (pass_loop_distribution::execute): Refactor code finding seed stmts
5845 and loop nest into above functions. Support loop nest distribution.
5846 Adjust dump information accordingly.
5848 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5850 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5851 and set PTYPE_SEQUENTIAL for merged partition.
5853 2017-10-12 Richard Biener <rguenther@suse.de>
5855 PR tree-optimization/69728
5857 2017-09-19 Richard Biener <rguenther@suse.de>
5859 PR tree-optimization/69728
5860 * graphite-sese-to-poly.c (schedule_error): New global.
5861 (add_loop_schedule): Handle empty domain by failing the
5863 (build_original_schedule): Handle schedule_error.
5865 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5866 domain by returning an unchanged schedule.
5868 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5870 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5871 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5873 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5875 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5878 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5881 * expr.c (store_field): Don't optimize away bitsize == 0 store
5882 from CALL_EXPR with addressable return type.
5884 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5886 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5887 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5888 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5889 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5890 TARGET_ISEL instead of TARGET_ISEL<sel>.
5892 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5894 * config/rs6000/rs6000.c
5895 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5897 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5899 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5901 (reg_or_cint_operand): Fix comment.
5902 (reg_or_zero_operand): New predicate.
5903 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5904 * config/rs6000/rs6000.c (output_isel): Delete.
5905 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5906 instead of reg_or_cint_operand. Output instruction directly (not via
5908 (isel_unsigned_<mode>): Ditto.
5909 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5910 gpc_reg_operand. Add an instruction alternative for this. Output
5911 instruction directly.
5912 (*isel_reversed_unsigned_<mode>): Ditto.
5914 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5916 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5917 (TARGET_CANONICALIZE_COMPARISON): Define.
5919 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5922 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5923 Check whether the dest is REG before adding REG_EQUIV note.
5925 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5928 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5930 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5931 (make_hard_regno_born, make_hard_regno_dead): Update
5932 bb_killed_pseudos and bb_gen_pseudos.
5934 2017-10-11 Nathan Sidwell <nathan@acm.org>
5936 * incpath.h (enum incpath_kind): Name enum, prefix values.
5937 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5938 * incpath.c (heads, tails): Use INC_MAX.
5939 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5940 (merge_include_chains, split_quote_chain,
5941 register_include_chains): Update incpath_kind names.
5942 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5943 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5945 (add_framework_path, darwin_register_objc_includes): Likewise.
5946 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5948 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5950 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5951 Do not use float_operator operator predicate.
5952 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5953 * config/i386/predicates.md (float_operator): Remove predicate.
5955 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5957 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5958 (*jccxf_i387): Ditto.
5959 (*jcc<mode>_i387): Ditto.
5960 (*jccu<mode>_i387): Ditto.
5961 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5962 (*jcc_*_i387 splitters): Remove.
5963 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5964 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5965 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5968 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5970 * profile-count.h (slow_safe_scale_64bit): New function.
5971 (safe_scale_64bit): New inline.
5972 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5973 * profile-count.c: Include wide-int.h
5974 (slow_safe_scale_64bit): New.
5976 2017-10-11 Nathan Sidwell <nathan@acm.org>
5978 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5979 HAS_DECL_ASSEMBLER_NAME_P.
5980 * gimple-expr.c (gimple_decl_printable_name: Check
5981 HAS_DECL_ASSEMBLER_NAME_P too.
5982 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5983 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5984 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5985 * passes.c (rest_of_decl_compilation): Check
5986 HAS_DECL_ASSEMBLER_NAME_P too.
5987 * recog.c (verify_changes): Likewise.
5988 * tree-pretty-print.c (dump_decl_name): Likewise.
5989 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5991 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5992 (SET_DECL_ASSEMBLER_NAME): Use it.
5993 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5994 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5995 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5997 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5999 * config.gcc (i386, x86_64): Add extra objects.
6000 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6001 (ix86_min_insn_size): Declare.
6002 (ix86_issue_rate): Declare.
6003 (ix86_adjust_cost): Declare.
6004 (ia32_multipass_dfa_lookahead): Declare.
6005 (ix86_macro_fusion_p): Declare.
6006 (ix86_macro_fusion_pair_p): Declare.
6007 (ix86_bd_has_dispatch): Declare.
6008 (ix86_bd_do_dispatch): Declare.
6009 (ix86_core2i7_init_hooks): Declare.
6010 (ix86_atom_sched_reorder): Declare.
6011 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6012 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6013 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6014 (rip_relative_addr_p): Rename to ...
6015 (ix86_rip_relative_addr_p): ... this one; export.
6016 (memory_address_length): Update.
6017 (ix86_issue_rate): Move to x86-tune-sched.c.
6018 (ix86_flags_dependent): Move to x86-tune-sched.c.
6019 (ix86_agi_dependent): Move to x86-tune-sched.c.
6020 (exact_dependency_1): Move to x86-tune-sched.c.
6021 (exact_store_load_dependency): Move to x86-tune-sched.c.
6022 (ix86_adjust_cost): Move to x86-tune-sched.c.
6023 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6024 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6025 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6026 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6027 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6028 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6029 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6030 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6031 (min_insn_size): Rename to ...
6032 (ix86_min_insn_size): ... this one; export.
6033 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6034 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6035 (core2i7_first_cycle_multipass_backtrack): Move to
6036 x86-tune-sched-core.c.
6037 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6038 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6039 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6040 (ix86_avoid_jump_mispredicts): Update.
6041 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6042 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6043 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6044 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6045 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6046 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6047 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6048 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6049 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6050 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6051 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6052 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6053 (BIG): Move to ix86-tune-sched-bd.c.
6054 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6055 (enum insn_path): Move to ix86-tune-sched-bd.c.
6056 (get_mem_group): Move to ix86-tune-sched-bd.c.
6057 (is_cmp): Move to ix86-tune-sched-bd.c.
6058 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6059 (is_branch): Move to ix86-tune-sched-bd.c.
6060 (is_prefetch): Move to ix86-tune-sched-bd.c.
6061 (init_window): Move to ix86-tune-sched-bd.c.
6062 (allocate_window): Move to ix86-tune-sched-bd.c.
6063 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6064 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6065 (process_end_window): Move to ix86-tune-sched-bd.c.
6066 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6067 (find_constant): Move to ix86-tune-sched-bd.c.
6068 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6069 (has_immediate): Move to ix86-tune-sched-bd.c.
6070 (get_insn_path): Move to ix86-tune-sched-bd.c.
6071 (get_insn_group): Move to ix86-tune-sched-bd.c.
6072 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6073 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6074 (add_insn_window): Move to ix86-tune-sched-bd.c.
6075 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6076 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6077 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6078 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6079 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6080 (do_dispatch): Move to ix86-tune-sched-bd.c.
6081 (has_dispatch): Move to ix86-tune-sched-bd.c.
6082 * config/i386/t-i386: Add new object files.
6083 * config/i386/x86-tune-costs.h: New file.
6084 * config/i386/x86-tune-sched-atom.c: New file.
6085 * config/i386/x86-tune-sched-bd.c: New file.
6086 * config/i386/x86-tune-sched-core.c: New file.
6087 * config/i386/x86-tune-sched.c: New file.
6089 2017-10-11 Liu Hao <lh_mouse@126.com>
6091 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6092 the generic version below instead.
6093 (should_colorize): Recognize Windows consoles as terminals
6095 * pretty-print.c [__MINGW32__] (write_all): New function.
6096 [__MINGW32__] (find_esc_head): Likewise.
6097 [__MINGW32__] (find_esc_terminator): Likewise.
6098 [__MINGW32__] (eat_esc_sequence): Likewise.
6099 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6101 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6104 2017-10-11 Richard Biener <rguenther@suse.de>
6106 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6107 Properly call analyze_scalar_evolution with the loop of the stmt.
6109 2017-10-11 Richard Biener <rguenther@suse.de>
6111 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6112 * tree-core.h (tree_base): Add chrec_var union member.
6113 * tree.h (CHREC_VAR): Remove.
6114 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6115 * tree-chrec.h (build_polynomial_chrec): Adjust.
6116 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6117 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6119 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6121 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6122 * match.pd: ... here.
6123 ((T) X == (T) Y): Relax condition.
6125 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6127 PR tree-optimization/82472
6128 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6130 (break_alias_scc_partitions): Update postorder number.
6132 2017-10-11 Martin Liska <mliska@suse.cz>
6135 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6137 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6139 2017-10-11 Martin Liska <mliska@suse.cz>
6141 * print-rtl.c (print_insn): Move declaration of idbuf
6142 to same scope as name.
6144 2017-10-11 Martin Liska <mliska@suse.cz>
6149 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6152 2017-10-11 Martin Liska <mliska@suse.cz>
6155 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6158 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6160 * config.gcc (default_gnu_indirect_function): Default to yes for
6161 arm*-*-linux* with glibc.
6163 2017-10-11 Richard Biener <rguenther@suse.de>
6165 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6166 default-defs and types we do not want to analyze.
6167 (interpret_loop_phi): Replace unreachable code with an assert.
6168 (compute_scalar_evolution_in_loop): Remove and inline ...
6169 (analyze_scalar_evolution_1): ... here, replacing condition with
6170 what makes the intent clearer. Remove handling of cases
6171 get_scalar_evolution now handles.
6173 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6175 PR rtl-optimization/81434
6176 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6177 comment for main loop. In sched_group_found if, also add checks for
6178 pass and min_cost_group.
6180 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6182 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6183 (rs6000_insn_cost): New function.
6184 * config/rs6000/rs6000.md (cost): New attribute.
6186 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6187 H.J. Lu <hongjiu.lu@intel.com>
6191 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6192 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6193 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6194 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6195 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6196 OPTION_MASK_ISA_MMX and some other bit set require both
6197 mmx and the ISAs without the mmx bit.
6198 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6199 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6200 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6201 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6202 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6203 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6204 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6205 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6206 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6207 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6208 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6209 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6210 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6211 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6212 Add OPTION_MASK_ISA_MMX.
6214 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6216 * config.gcc (armv7*-*-freebsd*): New target.
6217 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6219 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6221 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6222 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6225 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6227 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6228 a template parameter.
6229 (WIDE_INT_REF_FOR): Update accordingly.
6230 * tree.h (wi::int_traits <const_tree>): Delete.
6231 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6232 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6233 (wi::tree_to_wide_ref): New typedef.
6234 (wi::to_wide): New function.
6235 * calls.c (get_size_range): Use wi::to_wide when operating on
6237 * cgraph.c (cgraph_node::create_thunk): Likewise.
6238 * config/i386/i386.c (ix86_data_alignment): Likewise.
6239 (ix86_local_alignment): Likewise.
6240 * dbxout.c (stabstr_O): Likewise.
6241 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6242 * expr.c (const_vector_from_tree): Likewise.
6243 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6244 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6245 (fold_negate_expr_1, int_const_binop_1, const_binop)
6246 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6247 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6248 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6249 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6250 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6251 (fold_not_const, round_up_loc): Likewise.
6252 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6253 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6254 (alloca_call_type): Likewise.
6255 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6256 * godump.c (go_output_typedef): Likewise.
6257 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6258 * internal-fn.c (get_min_precision): Likewise.
6259 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6260 * ipa-polymorphic-call.c
6261 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6262 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6263 (ipa_modify_call_arguments): Likewise.
6264 * match.pd: Likewise.
6265 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6266 * print-tree.c (print_node_brief, print_node): Likewise.
6267 * stmt.c (expand_case): Likewise.
6268 * stor-layout.c (layout_type): Likewise.
6269 * tree-affine.c (tree_to_aff_combination): Likewise.
6270 * tree-cfg.c (group_case_labels_stmt): Likewise.
6271 * tree-data-ref.c (dr_analyze_indices): Likewise.
6272 (prune_runtime_alias_test_list): Likewise.
6273 * tree-dump.c (dequeue_and_dump): Likewise.
6274 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6275 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6276 * tree-pretty-print.c (dump_generic_node): Likewise.
6277 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6278 (simple_iv_with_niters): Likewise.
6279 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6280 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6281 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6282 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6283 (refine_value_range_using_guard, number_of_iterations_ne_max)
6284 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6285 (get_cst_init_from_scev, record_nonwrapping_iv)
6286 (scev_var_range_cant_overflow): Likewise.
6287 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6288 * tree-ssa-pre.c (compute_avail): Likewise.
6289 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6290 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6291 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6292 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6293 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6294 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6295 (dump_case_nodes, try_switch_expansion): Likewise.
6296 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6297 (vect_do_peeling): Likewise.
6298 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6299 * tree-vect-stmts.c (vectorizable_load): Likewise.
6300 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6301 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6302 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6303 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6304 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6305 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6306 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6307 (evrp_dom_walker::before_dom_children): Likewise.
6308 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6309 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6310 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6311 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6312 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6313 (get_range_pos_neg): Likewise.
6314 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6315 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6316 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6317 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6318 * config/avr/avr.c (avr_fold_builtin): Likewise.
6319 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6320 * config/msp430/msp430.c (msp430_attr): Likewise.
6321 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6322 * config/powerpcspe/powerpcspe-c.c
6323 (altivec_resolve_overloaded_builtin): Likewise.
6324 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6325 (rs6000_expand_ternop_builtin): Likewise.
6326 * config/rs6000/rs6000-c.c
6327 (altivec_resolve_overloaded_builtin): Likewise.
6328 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6329 (rs6000_expand_ternop_builtin): Likewise.
6330 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6332 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6334 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6335 when copying loop nest with only one inner loop.
6337 2017-10-10 Richard Biener <rguenther@suse.de>
6339 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6340 blocks if SCEV is active.
6341 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6343 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6344 (scev_initialize): Assert we are not yet initialized.
6346 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6348 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6349 inner loop's exit stmt by making it always exit the loop, otherwise
6350 we would generate an infinite empty loop.
6352 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6354 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6355 renaming variables in new preheader if it's deleted.
6357 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6359 * tree-loop-distribution.c (struct partition): Remove unused field
6360 loops of the structure.
6361 (partition_alloc, partition_free): Ditto.
6362 (build_rdg_partition_for_vertex): Ditto.
6364 2017-10-09 Jeff Law <law@redhat.com>
6366 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6367 return type to match prototype and documentation.
6369 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6371 * config/rs6000/rs6000.c (processor_costs): Move to ...
6372 * config/rs6000/rs6000.h: ... here.
6373 (rs6000_cost): Declare.
6375 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6377 * except.c (setjmp_fn): New global variable.
6378 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6379 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6380 if DONT_USE_BUILTIN_SETJMP is defined.
6382 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6384 * target.def (insn_cost): New hook.
6385 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6386 * doc/tm.texi: Regenerate.
6387 * rtlanal.c (insn_cost): Use the new hook.
6389 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6391 * combine.c (combine_validate_cost): Compute the new insn_cost,
6392 not just pattern_cost.
6393 (try_combine): Adjust comment.
6395 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6397 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6399 * combine.c (uid_insn_cost): Adjust comment.
6400 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6402 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6403 * dse.c (find_shift_sequence): Ditto.
6404 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6405 (bb_valid_for_noce_process_p): Use pattern_cost.
6406 * rtl.h (insn_rtx_cost): Delete.
6407 (pattern_cost): New prototype.
6408 (insn_cost): New prototype.
6409 * rtlanal.c (insn_rtx_cost): Rename to...
6410 (pattern_cost): ... this.
6413 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6415 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6416 (*jcc<mode>_0_r_i387): Ditto.
6417 (*jccxf_r_i387): Ditto.
6418 (*jcc<mode>_r_i387): Ditto.
6419 (*jccu<mode>_r_i387): Ditto.
6420 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6421 (*jcc): Rename from *jcc_1.
6423 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6425 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6426 deferred rescans after the lvx/stvx recombination pre-pass.
6428 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6430 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6431 memory operation instruction support.
6432 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6433 (rs6000-ibm-aix[789]*): Likewise.
6434 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6435 Document new functions.
6437 2017-10-09 Richard Biener <rguenther@suse.de>
6439 PR tree-optimization/82397
6440 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6441 equality only for semantically equal trees.
6443 2017-10-09 Richard Biener <rguenther@suse.de>
6445 PR tree-optimization/82449
6446 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6447 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6448 allow constant addresses.
6449 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6452 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6454 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6457 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6460 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6463 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6466 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6468 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6471 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6472 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6474 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6476 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6477 (WI_BINARY_PREDICATE_RESULT): Likewise.
6478 (wi::binary_traits::operator_result): New type.
6479 (wi::binary_traits::predicate_result): Likewise.
6480 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6481 (generic_wide_int::operator==, generic_wide_int::operator!=)
6482 (generic_wide_int::operator&, generic_wide_int::and_not)
6483 (generic_wide_int::operator|, generic_wide_int::or_not)
6484 (generic_wide_int::operator^, generic_wide_int::operator+
6485 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6487 (operator~, unary operator-, operator==, operator!=, operator&)
6488 (operator|, operator^, operator+, binary operator-, operator*): New
6490 * expr.c (get_inner_reference): Use wi::bit_and_not.
6491 * fold-const.c (fold_binary_loc): Likewise.
6492 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6493 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6494 (bit_value_binop): Likewise.
6495 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6496 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6497 (extract_range_from_binary_expr_1): Likewise.
6498 (masked_increment): Likewise.
6499 (simplify_bit_ops_using_ranges): Likewise.
6501 2017-10-09 Martin Jambor <mjambor@suse.cz>
6504 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6505 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6506 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6508 (hsa_fixup_mov_insn_type): New function.
6509 (hsa_op_with_type::get_in_type): Use it.
6510 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6511 immediates in an assert.
6512 (hsa_op_with_type::extend_int_to_32bit): New method.
6513 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6514 types. Convert to dest type if necessary.
6515 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6516 (reg_for_gimple_ssa): Pass false as min32int to
6517 hsa_type_for_scalar_tree_type.
6518 (gen_hsa_addr): Fixup type when creating addresable temporary.
6519 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6520 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6521 necessary. Call hsa_fixup_mov_insn_type.
6522 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6523 extend operands and convert to dest type if necessary.
6524 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6525 to dest type if necessary.
6526 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6527 if conversion nt necessary and size matches.
6528 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6529 to dest type if necessary.
6530 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6531 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6533 (gen_hsa_clrsb): Likewise.
6534 (gen_hsa_ffs): Likewise.
6535 (gen_hsa_divmod): Extend operands and convert to dest type if
6537 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6539 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6541 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6542 Remove empty default arguments. Use a brace block as output
6544 (conditional return): Ditto.
6546 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6547 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6549 (group_ending_nop): Ditto.
6550 (doloop_end): Ditto.
6551 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6552 (splitters for those): Ditto.
6554 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6556 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6557 a conditional jump (and the compare for it) so that pc_rtx is the
6559 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6560 for the deleted and renamed ctr<mode>_internal[234] patterns.
6561 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6562 Delete second conditional return pattern.
6563 (ctr<mode>_internal2): Delete this second bdnz pattern.
6564 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6565 (ctr<mode>_internal4): Delete this second bdz pattern.
6567 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6569 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6570 (always_initialized_rtx_for_ssa_name_p): New predicate.
6571 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6572 (finish_out_of_ssa): Free new field of SA.
6573 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6574 * tree-ssa-coalesce.c: Include tree-ssa.h.
6575 (get_parm_default_def_partitions): Remove extern keyword.
6576 (get_undefined_value_partitions): New function.
6577 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6578 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6580 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6582 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6584 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6586 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6588 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6589 for targets that preffer 128bit.
6591 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6593 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6595 2017-10-08 Olivier Hainque <hainque@adacore.com>
6597 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6598 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6600 (thumb_set_return_address): Likewise.
6602 2017-10-08 Olivier Hainque <hainque@adacore.com>
6604 * common/config/arm/arm-common.c (arm_except_unwind_info):
6605 Handle DWARF2_UNWIND_INFO.
6607 2017-10-07 Michael Collison <michael.collison@arm.com>
6609 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6612 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6614 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6615 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6616 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6617 defined, force the creation of a new block for a dispatch label.
6619 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6621 * invoke.texi (Wsuggest-attribute=cold): Document.
6622 * common.opt (Wsuggest-attribute=cold): New
6623 * ipa-pure-const.c (warn_function_cold): New function.
6624 * predict.c (compute_function_frequency): Use it.
6625 * predict.h (warn_function_cold): Declare.
6627 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6629 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6632 2017-10-06 Martin Liska <mliska@suse.cz>
6634 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6635 keyword for member functions.
6636 (struct sanopt_tree_couple): New struct.
6637 (struct sanopt_tree_couple_hash): New function.
6638 (struct sanopt_ctx): Add new hash_map.
6639 (has_dominating_ubsan_ptr_check): New function.
6640 (record_ubsan_ptr_check_stmt): Likewise.
6641 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6642 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6643 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6645 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6648 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6649 aarch64_simd_valid_immediate on CONST_VECTORs.
6650 (aarch64_reg_or_bic_imm): Likewise.
6652 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6654 PR rtl-optimization/82396
6655 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6657 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6659 * graphite-dependences.c (scop_get_reads): Move code to...
6660 (scop_get_must_writes): Move code to...
6661 (scop_get_may_writes): Move code to...
6662 (scop_get_reads_and_writes): ... here.
6663 (scop_get_dependences): Call scop_get_reads_and_writes.
6665 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6667 PR tree-optimization/82434
6668 * fold-const.h (can_native_encode_type_p,
6669 can_native_encode_string_p): Remove.
6670 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6671 don't encode anything, just return what would be otherwise returned.
6672 (native_encode_fixed, native_encode_complex, native_encode_vector):
6674 (native_encode_string): Likewise. Inline by hand
6675 can_native_encode_string_p.
6676 (can_native_encode_type_p): Remove.
6677 (can_native_encode_string_p): Remove.
6678 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6679 STRING_CSTs using can_native_encode_string_p, test all
6680 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6681 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6682 argument from native_encode_expr.
6683 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6684 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6687 2017-10-06 Richard Biener <rguenther@suse.de>
6689 PR tree-optimization/82397
6690 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6691 operand_equal_p but rely on data_ref_compare_tree for detecting
6693 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6694 to match up with dr_group_sort_cmp.
6696 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6699 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6701 * config/s390/s390-builtin-types.def: Regenerate.
6703 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6706 * config/s390/s390-builtin-types.def: Regenerate.
6707 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6708 Change flag from B_VXE to B_VX.
6709 (s390_vec_min_dbl): Remove B_VXE flag.
6711 2017-10-06 Richard Biener <rguenther@suse.de>
6713 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6714 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6715 translate_isl_ast_to_gimple::is_valid_rename,
6716 translate_isl_ast_to_gimple::get_rename,
6717 translate_isl_ast_to_gimple::get_def_bb_for_const,
6718 translate_isl_ast_to_gimple::get_new_name,
6719 translate_isl_ast_to_gimple::collect_all_ssa_names,
6720 translate_isl_ast_to_gimple::copy_loop_phi_args,
6721 translate_isl_ast_to_gimple::collect_all_ssa_names,
6722 translate_isl_ast_to_gimple::copy_loop_phi_args,
6723 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6724 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6725 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6726 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6727 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6728 translate_isl_ast_to_gimple::copy_cond_phi_args,
6729 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6730 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6731 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6732 translate_isl_ast_to_gimple::rename_uses,
6733 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6734 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6735 (set_rename_for_each_def): Likewise.
6736 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6737 here. Handle rewriting SCEV analyzable uses here.
6738 (copy_bb_and_scalar_dependences): Generate code for PHI
6740 (graphite_regenerate_ast_isl): Adjust.
6741 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6742 (add_write, add_read): New functions.
6743 (build_cross_bb_scalars_def): Use it and simplify.
6744 (build_cross_bb_scalars_use): Likewise.
6745 (graphite_find_cross_bb_scalar_vars): Inline into...
6746 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6747 simulating out-of-SSA. Compute liveout and add dependencies.
6748 (build_scops): Force an empty entry block.
6749 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6751 (sese_build_liveouts): Declare.
6752 (sese_trivially_empty_bb_p): Likewise.
6753 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6754 compute liveout and debug_liveout.
6755 (sese_bad_liveouts_use): Remove.
6756 (sese_reset_debug_liveouts_bb): Likewise.
6757 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6758 (sese_build_liveouts): Build liveout and debug_liveout and store
6760 (new_sese_info): Adjust.
6761 (free_sese_info): Likewise.
6762 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6763 do not build liveout here.
6764 (move_sese_in_condition): Adjust region entry.
6765 (scev_analyzable_p): Match up with chrec_apply requirements.
6766 (sese_trivially_empty_bb_p): New.
6767 * tree-into-ssa.c (get_reaching_def): Properly support generating
6768 default-defs for incremental rewrite of anonymous names.
6770 2017-10-06 Richard Biener <rguenther@suse.de>
6772 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6773 precision do not perform modulo reduction.
6775 2017-10-06 Richard Biener <rguenther@suse.de>
6777 PR tree-optimization/82436
6778 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6779 conservatively choose the vectorization factor when checking
6780 whether we can perform the required load permutation.
6781 (vect_transform_slp_perm_load): Assert when we may not fail.
6783 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6785 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6786 message for incompatible -msdata=* and -mcall-* options.
6788 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6790 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6791 rate for post-reload scheduling.
6793 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6795 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6797 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6799 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6800 to improve monte carlo in scimark.
6802 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6804 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6805 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6806 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6807 pentium4_cost, nocona_cost): Set reassociation width to 1.
6808 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6809 width to 2 for fp operations and 1 otherwise.
6810 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6812 (atom_cost): Set reassociation width to 2.
6813 (slm_cost, generic_cost): Set fp reassociation width
6814 to 2 and 1 otherwise.
6815 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6816 (core_cost): Set fp reassociation width to 4 and vector to 2.
6817 (ix86_reassociation_width): Rewrite using cost table; special case
6818 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6819 and TARGET_AVX128_OPTIMAL.
6820 * config/i386/i386.h (processor_costs): Add
6821 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6822 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6823 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6824 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6825 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6826 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6828 2017-10-05 Nathan Sidwell <nathan@acm.org>
6830 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6832 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6834 * config/arm/arm.c (arm_test_fpu_data): New.
6835 (arm_run_selftests): Call arm_test_fpu_data.
6837 2017-10-04 Nathan Sidwell <nathan@acm.org>
6839 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6841 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6843 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6845 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6846 check type for aarch64_simd_valid_immediate.
6847 (aarch64_output_simd_mov_immediate): Update prototype.
6848 (aarch64_simd_valid_immediate): Update prototype.
6849 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6850 support for ORR-immediate.
6851 (and<mode>3): modified pattern to add support for BIC-immediate.
6852 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6853 now checks for valid immediate for BIC and ORR based on new enum
6855 (aarch64_output_simd_mov_immediate): Function now used to output
6856 BIC/ORR imm as well based on new enum argument.
6857 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6859 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6860 (aarch64_reg_or_bic_imm): Likewise.
6862 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6864 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6865 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6867 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6871 PR rtl-optimization/82396
6872 * haifa-sched.c (autopref_multipass_init): Simplify
6874 (autopref_rank_data): Simplify sort order.
6875 * sched-int.h (autopref_multipass_data_): Remove
6876 multi_mem_insn_p, min_offset and max_offset.
6878 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6880 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6882 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6884 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6887 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6889 * doc/sourcebuild.texi: Document vect_long_mult.
6891 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6893 PR tree-optimization/82413
6894 * fold-const.c (build_range_check): Use widest_int when comparing
6895 the maximum ETYPE value with HIGH.
6897 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6899 PR rtl-optimization/82396
6900 * haifa-sched.c (autopref_multipass_init): Simplify
6902 (autopref_rank_data): Simplify sort order.
6903 * sched-int.h (autopref_multipass_data_): Remove
6904 multi_mem_insn_p, min_offset and max_offset.
6906 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6908 PR tree-optimization/82381
6909 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6910 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6913 PR tree-optimization/82374
6914 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6915 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6916 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6917 current_function_decl to the new decl.
6919 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6921 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6922 helper macro for IEEE float128 hardware built-in functions.
6923 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6925 (TRUNCF128_ODD): Likewise.
6926 (ADDF128_ODD): Likewise.
6927 (SUBF128_ODD): Likewise.
6928 (MULF128_ODD): Likewise.
6929 (DIVF128_ODD): Likewise.
6930 (FMAF128_ODD): Likewise.
6931 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6932 UNSPEC_TRUNC_ROUND_TO_ODD.
6933 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6934 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6935 floating point round to odd instructions.
6936 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6937 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6938 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6939 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6940 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6941 (trunc<mode>sf2_hw): Change the truncate with round to odd
6942 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6943 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6944 to odd hardware instructions.
6945 (sub<mode>3_odd): Likewise.
6946 (mul<mode>3_odd): Likewise.
6947 (div<mode>3_odd): Likewise.
6948 (sqrt<mode>2_odd): Likewise.
6949 (fma<mode>4_odd): Likewise.
6950 (fms<mode>4_odd): Likewise.
6951 (nfma<mode>4_odd): Likewise.
6952 (nfms<mode>4_odd): Likewise.
6953 (trunc<mode>df2_odd): Change the truncate with round to odd
6954 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6956 * doc/extend.texi (PowerPC built-in functions): Update documentation
6957 for existing IEEE float128-bit built-in functions. Add built-in
6958 functions that generate the IEEE 128-bit floating point round to
6961 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6963 PR rtl-optimization/77729
6964 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6965 to (X&(C1&~C2))|C2 transformations.
6967 2017-10-03 Martin Jambor <mjambor@suse.cz>
6969 PR tree-optimization/82363
6970 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6971 mismatch, mark lacc written regardless of racc.
6973 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6975 PR tree-optimization/82381
6976 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6977 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6978 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6979 fallthrough into reassoc_stmt_dominates_stmt_p.
6982 * combine.c (combine_instructions): Don't combine in unreachable
6985 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6989 function to not use the have_cpu variable. Do not set cpu_index,
6990 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6992 (rs6000_valid_attribute_p): Remove duplicate initializations of
6993 old_optimize and func_optimize.
6994 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6995 (rs6000_activate_target_options): Make global.
6996 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6999 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7001 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7002 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7003 Set *poffset to 0 and *psize and *pmax_size to -1 if
7004 *poffset + *psize overflows in HOST_WIDE_INT.
7006 PR tree-optimization/82387
7007 PR tree-optimization/82388
7008 PR tree-optimization/82389
7009 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7010 instead of live_bytes non-NULL.
7012 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7015 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7018 2017-10-02 Richard Biener <rguenther@suse.de>
7020 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7021 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7022 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7024 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7026 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7027 requested precision matches the type's.
7028 * calls.c (alloc_max_size): Calculate the new candidate size as
7029 a widest_int and use wi::to_widest when comparing it with the
7030 current candidate size.
7031 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7032 zero rather than integer_zero_node.
7033 * match.pd: Check for a no-op conversion before using wi::add
7034 rather than after. Use tree_to_uhwi when summing small shift
7035 counts into an unsigned int.
7037 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7038 Alan Hayward <alan.hayward@arm.com>
7039 David Sherwood <david.sherwood@arm.com>
7042 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7043 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7044 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7045 POINTER_AND_FP_REGS.
7047 2017-10-02 Richard Biener <rguenther@suse.de>
7049 PR tree-optimization/82355
7050 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7051 a mapping for the enclosing loop but avoid generating one for
7053 (copy_bb_and_scalar_dependences): Remove premature codegen
7054 error on PHIs in blocks duplicated into multiple places.
7055 * graphite-scop-detection.c
7056 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7057 in the region use it as loop and nest to analyze the DR in.
7058 (try_generate_gimple_bb): Likewise.
7059 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7060 (add_loop_constraints): For blocks in a loop not in the region
7061 create a dimension with a single iteration.
7062 * sese.h (gbb_loop_at_index): Remove assert.
7064 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7066 * omp-expand.c (adjust_context_scope): New function.
7067 (expand_parallel_call): Call adjust_context_scope.
7069 2017-10-01 Jeff Law <law@redhat.com>
7071 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7072 dom_opt_dom_walker class with direct access to private members.
7073 Add comments. Call test_for_singularity.
7074 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7075 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7076 m_dummy_cond anymore.
7077 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7079 (pass_dominator:execute): Build the dummy_cond here and pass it
7080 to the dom_opt_dom_walker ctor.
7081 (test_for_singularity): New function.
7083 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7084 Maya Rashish <coypu@sdf.org>
7086 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7087 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7088 (alpha*-*-netbsd*) Use nbsd_tm_file.
7089 (arm*-*-netbsdelf*) Likewise.
7090 (i[34567]86-*-netbsdelf*) Likewise.
7091 (x86_64-*-netbsd*) Likewise.
7092 (mips*-*-netbsd*) Likewise.
7093 (powerpc-*-netbsd*) Likewise.
7094 (sh*-*-netbsd*) Likewise.
7095 (sparc-*-netbsdelf*) Likewise.
7096 (sparc64-*-netbsd*) Likewise.
7097 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7099 (vax-*-netbsdelf*) Likewise.
7100 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7101 (UINT_FAST8_TYPE) Likewise.
7102 (INT_FAST16_TYPE) Check CHAR_FAST16.
7103 (UINT_FAST16_TYPE) Likewise.
7105 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7108 * config/i386/i386.md
7109 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7110 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7111 *divmodsi4_zext_2): New define_insn_and_split.
7112 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7113 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7114 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7115 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7116 New define_insn_and_split.
7117 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7118 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7119 operands[1] having DImode when mode is SImode.
7121 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7122 always SImode for DIV and MOD in REG_EQUAL notes.
7124 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7127 * match.pd: Fix handling of NaNs in pattern.
7129 2017-09-29 Jeff Law <law@redhat.com>
7131 * sbitmap.c (bitmap_bit_in_range_p): New function.
7132 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7133 * tree-ssa-dse.c (live_bytes_read): New function.
7134 (dse_classify_store): Ignore reads of dead bytes.
7136 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7137 typos and whitespace errors.
7138 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7139 * config/s390/s390.c (s390_emit_prologue): Likewise.
7141 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7144 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7145 with a symbol for LRA.
7147 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7149 PR rtl-optimization/82338
7150 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7152 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7154 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7155 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7156 (qsort_chk): Declare.
7157 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7158 (qsort_chk): New function.
7160 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7162 PR tree-optimization/82337
7163 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7164 phi definition if the PHI result appears in an abnormal PHI.
7165 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7166 basis appears in an abnormal PHI.
7168 2017-09-29 Richard Biener <rguenther@suse.de>
7170 * graphite-isl-ast-to-gimple.c
7171 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7172 (binary_op_to_tree): Use it.
7173 (get_rename_from_scev): Likewise.
7174 (copy_loop_phi_nodes): Likewise.
7175 (copy_bb_and_scalar_dependences): Likewise.
7176 (translate_pending_phi_nodes): Likewise.
7178 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7181 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7182 for movabsq $(i32 << shift), r64.
7184 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7186 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7187 index when encoding %esp as %rsp to avoid 0x67 prefix.
7189 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7191 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7192 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7194 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7196 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7197 Extensions with more than 16 double VFP registers.
7198 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7199 to_clear_mask and all code related to it. Replace the remaining
7200 entry by a sbitmap and adapt code accordingly.
7202 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7204 * brig-builtins.def: Change pure attributes to const.
7206 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7208 * config.gcc (default_gnu_indirect_function): Default to yes for
7209 sparc*-*-linux* with glibc.
7211 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7213 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7214 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7215 when creating .init_array and .fini_array sections with priority
7218 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7221 * config/aarch64/aarch64.c
7222 (aarch64_builtin_support_vector_misalignment): Always return false
7223 when misalignment is unknown.
7225 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7227 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7228 this function to return false if the definition used by the swap
7229 instruction is artificial, or if the memory address from which the
7230 constant value is loaded is not represented by a base address held
7231 in a register or if the base address register is a frame or stack
7232 pointer. Additionally, return false if the base address of the
7233 loaded constant is a SYMBOL_REF but is not considered to be a
7235 (replace_swapped_load_constant): New function.
7236 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7237 loaded constant vector with a load of a swapped constant vector.
7239 2017-09-27 Carl Love <cel@us.ibm.com>
7241 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7242 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7243 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7246 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7248 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7249 first, always call autopref_rank_data otherwise.
7251 2017-09-27 Richard Biener <rguenther@suse.de>
7253 * graphite-scop-detection.c (find_scop_parameters): Move
7254 loop bound handling ...
7255 (gather_bbs::before_dom_children): ... here, avoiding the need
7256 to build scop_info->loop_nest.
7257 (record_loop_in_sese): Remove.
7258 * sese.h (sese_info_t::loop_nest): Remove.
7259 * sese.c (new_sese_info): Do not allocate loop_nest.
7260 (free_sese_info): Do not free loop_nest.
7262 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7265 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7266 lhs from calls if the lhs has addressable type.
7268 2017-09-27 Richard Biener <rguenther@suse.de>
7270 * graphite.h (scop::max_alias_set): New member.
7271 * graphite-scop-detection.c: Remove references to non-existing
7272 --param in comments.
7273 (build_alias_sets): Record the maximum alias set used for drs.
7274 (build_scops): Support zero as unlimited for
7275 --param graphite-max-arrays-per-scop.
7276 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7278 (build_poly_sr_1): ... here. Compute alias set based on the
7279 maximum alias set used for drs rather than
7280 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7282 2017-09-27 Richard Biener <rguenther@suse.de>
7284 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7285 --param loop-block-tile-size=0 to disable tiling.
7287 2017-09-27 Richard Biener <rguenther@suse.de>
7289 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7290 (graphite-max-nb-scop-params): Document special value zero.
7291 * domwalk.h (dom_walker::STOP): New symbolical constant.
7292 (dom_walker::dom_walker): Add optional parameter for bb to
7294 (dom_walker::~dom_walker): Declare.
7295 (dom_walker::before_dom_children): Document STOP return value.
7296 (dom_walker::m_user_bb_to_rpo): New member.
7297 (dom_walker::m_bb_to_rpo): Likewise.
7298 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7299 mapping here if not provided by the user.
7300 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7301 provided by the user.
7302 (dom_walker::STOP): Define.
7303 (dom_walker::walk): Do not compute bb to RPO mapping here.
7304 Support STOP return value from before_dom_children to stop
7306 * graphite-optimize-isl.c (optimize_isl): If the schedule
7307 is the same still generate code if -fgraphite-identity
7308 or -floop-parallelize-all are given.
7309 * graphite-scop-detection.c: Include cfganal.h.
7310 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7312 (gather_bbs::before_dom_children): Return STOP for BBs
7314 (build_scops): Compute bb to RPO mapping and pass it to
7315 the domwalk. Treat --param graphite-max-nb-scop-params=0
7316 as not limiting the number of params.
7317 * graphite.c (graphite_initialize): Remove limit on the
7318 number of basic-blocks in a function.
7319 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7320 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7321 default value of 10.
7323 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7325 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7326 Adjust code to eliminate needing to do the shift right 32-bits
7327 operation after XSCVDPSPN.
7329 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7331 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7332 ((X / Y) != 0 -> X >= Y): Likewise.
7334 2017-09-26 Carl Love <cel@us.ibm.com>
7336 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7337 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7338 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7339 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7340 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7341 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7342 definitions and overloading.
7343 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7344 statement for P9V_BUILTIN_XST_LEN_R.
7345 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7346 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7347 define_expand and define_insn for the instructions and builtins.
7348 * doc/extend.texi: Update the built-in documentation file for the new
7350 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7351 define_insn for the instructions
7353 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7356 * gcc/config/netbsd-protos.h: New file.
7357 * gcc/config/netbsd.c: New file.
7358 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7359 * gcc/config/t-netbsd: New file.
7360 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7361 (tmake_file) Add t-netbsd.
7362 (extra_objs) Add netbsd.o.
7364 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7368 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7370 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7372 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7373 sign extension from a vector register to a GPR by doing a 32-bit
7374 direct move and then an EXTSW.
7375 (extendsi<mode>2 splitter): Likewise.
7376 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7377 right or vector extract after doing XSCVDPSPN. Use
7378 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7380 (movdi_from_sf_zero_ext): Likewise.
7381 (reload_gpr_from_vsxsf): Likewise.
7382 (p8_mfvsrd_4_disf): Delete, no longer used.
7383 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7384 then needing to move the SFmode to a GPR to use the XSCVDPSP
7385 instruction instead of FRSP and XSCVDPSPN.
7386 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7387 it is adjacent to the other XSCVSPDP insns.
7388 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7389 SFmode to be in traditional Altivec registers.
7390 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7391 (vsx_xscvspdpn): Likewise.
7392 (vsx_xscvspdpn_scalar): Likewise.
7394 2017-09-26 Martin Jambor <mjambor@suse.cz>
7396 * tree-sra.c (compare_access_positions): Put integral types first,
7397 stabilize sorting of integral types, remove conditions putting
7398 non-full-precision integers last.
7399 (sort_and_splice_var_accesses): Disable scalarization if a
7400 non-integert would be represented by a non-full-precision integer.
7402 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7404 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7405 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7406 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7407 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7408 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7409 conditionals inside the function instead of around it. Call
7410 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7411 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7413 2017-09-26 Richard Biener <rguenther@suse.de>
7415 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7417 (scop_detection::build_scop_breadth): ... this. Removed.
7418 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7419 (scop_detection::harmful_stmt_in_bb): Likewise.
7420 (scop_detection::graphite_can_represent_stmt): Likewise.
7421 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7422 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7423 (scop_detection::can_represent_loop_1): ... this. Removed.
7424 (scop_detection::harmful_loop_in_region): Simplify after inlining
7425 the above and remove more quadraticness.
7426 (build_scops): Adjust.
7427 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7430 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7433 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7434 REGNO (base) == SP_REG if base is a REG.
7437 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7438 if it is different SSA_NAME.
7439 (optimize_range_tests_cmp_bitwise): New function.
7440 (optimize_range_tests): Call it.
7442 2017-09-26 Richard Biener <rguenther@suse.de>
7444 PR tree-optimization/82321
7445 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7446 for the def being inside the loop.
7448 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7450 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7452 * config/s390/s390-builtins.def: Fix constraint on op4.
7454 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7456 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7457 independent expanders.
7458 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7459 ("vec_ordered", "vec_unordered"): New expanders.
7461 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7463 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7466 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7468 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7469 vec_unpacks_lo_v16qi.
7470 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7472 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7474 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7475 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7476 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7478 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7480 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7482 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7484 ("*vec_slb<mode>"): New insn pattern.
7485 ("vec_shr_<mode>"): New expander.
7486 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7487 and force the shift count operand to V16QImode.
7488 ("vec_srb<mode>"): Set shift count mode to V16QI.
7490 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7492 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7493 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7494 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7496 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7499 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7501 2017-09-26 Richard Biener <rguenther@suse.de>
7503 PR tree-optimization/82320
7504 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7507 2017-09-25 Jeff Law <law@redhat.com>
7509 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7510 prototype for new argument.
7511 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7512 mostly extracted from rs6000_emit_allocate_stack.
7513 (rs6000_emit_probe_stack_range_stack_clash): New function.
7514 (rs6000_emit_allocate_stack): Call
7515 rs6000_emit_probe_stack_range_stack_clash as needed.
7516 (rs6000_emit_probe_stack_range): Add additional argument
7517 to call to gen_probe_stack_range{si,di}.
7518 (output_probe_stack_range): New.
7519 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7520 (output_probe_stack_range_stack_clash): New.
7521 (rs6000_emit_prologue): Emit notes into dump file as requested.
7522 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7523 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7524 Add additional operand and pass it to output_probe_stack_range.
7526 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7528 PR tree-optimization/82163
7529 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7530 (checking_verify_loop_closed_ssa): New parameter.
7531 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7532 (check_loop_closed_ssa_stmt): Delete.
7533 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7534 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7535 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7538 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7540 * brig-builtins.def: Treat HSAIL barrier builtins as
7541 setjmp/longjump style functions.
7543 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7545 * target.def (constant_alignment): New hook.
7546 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7547 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7548 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7549 * doc/tm.texi: Regenerate.
7550 * targhooks.h (default_constant_alignment): Declare.
7551 (constant_alignment_word_strings): Likewise.
7552 * targhooks.c (default_constant_alignment): New function.
7553 (constant_alignment_word_strings): Likewise.
7554 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7555 instead of CONSTANT_ALIGNMENT.
7556 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7557 (force_const_mem): Likewise.
7558 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7559 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7560 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7561 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7562 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7564 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7565 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7566 constant_alignment_word_strings.
7567 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7568 (CONSTANT_ALIGNMENT): Likewise.
7569 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7570 (arm_constant_alignment): New function.
7571 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7572 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7573 constant_alignment_word_strings.
7574 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7575 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7576 constant_alignment_word_strings.
7577 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7578 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7579 (cris_constant_alignment): New function.
7580 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7581 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7582 (epiphany_constant_alignment): New function.
7583 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7584 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7585 constant_alignment_word_strings.
7586 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7587 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7588 constant_alignment_word_strings.
7589 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7590 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7591 constant_alignment_word_strings.
7592 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7593 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7594 * config/i386/i386.c (ix86_constant_alignment): Make static.
7595 Use the same interface as the target hook.
7596 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7597 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7598 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7599 constant_alignment_word_strings.
7600 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7601 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7602 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7603 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7604 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7605 constant_alignment_word_strings.
7606 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7607 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7608 constant_alignment_word_strings.
7609 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7610 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7611 constant_alignment_word_strings.
7612 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7613 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7615 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7616 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7617 * config/mips/mips.c (mips_constant_alignment): New function.
7618 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7619 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7620 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7621 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7622 (mmix_constant_alignment): Make static. Use the same interface
7624 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7625 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7626 constant_alignment_word_strings.
7627 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7628 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7629 constant_alignment_word_strings.
7630 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7631 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7632 constant_alignment_word_strings.
7633 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7634 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7635 (rs6000_constant_alignment): New function.
7636 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7637 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7638 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7639 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7640 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7641 (rs6000_constant_alignment): New function.
7642 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7643 * config/s390/s390.c (s390_constant_alignment): New function.
7644 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7645 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7646 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7647 constant_alignment_word_strings.
7648 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7649 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7650 (sparc_constant_alignment): New function.
7651 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7652 * config/spu/spu.c (spu_constant_alignment): New function.
7653 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7654 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7655 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7656 constant_alignment_word_strings.
7657 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7658 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7659 constant_alignment_word_strings.
7660 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7661 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7662 constant_alignment_word_strings.
7663 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7664 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7665 (visium_constant_alignment): New function.
7666 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7667 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7668 (xtensa_constant_alignment): New function.
7669 * system.h (CONSTANT_ALIGNMENT): Poison.
7671 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7674 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7675 (rs6000_builtin_valid_without_lhs): New helper function.
7676 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7677 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7679 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7681 * target.h (vec_perm_indices): Use unsigned short rather than
7683 (auto_vec_perm_indices): Likewise.
7684 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7685 Use unsigned int rather than unsigned char.
7686 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7688 2017-09-25 Richard Biener <rguenther@suse.de>
7690 * cfgloop.h (sort_sibling_loops): Declare.
7691 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7692 (sort_sibling_loops): New function sorting the sibling loop list
7694 * graphite.c (graphite_transform_loops): Sort sibling loops.
7696 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7698 * target.def (vec_perm_const_ok): Change sel parameter to
7700 * optabs-query.c (can_vec_perm_p): Update accordingly.
7701 * doc/tm.texi: Regenerate.
7702 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7703 auto_vec_perm_indices and remove separate nelt field.
7704 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7705 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7706 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7707 (aarch64_expand_vec_perm_const): Update accordingly.
7708 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7709 to vec_perm_indices.
7710 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7711 auto_vec_perm_indices and remove separate nelt field.
7712 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7713 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7714 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7716 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7717 to vec_perm_indices.
7718 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7719 sel to vec_perm_indices.
7720 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7721 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7722 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7724 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7727 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7730 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7731 on the FUNCTION_DECL function context if it has a DIE that is a
7734 2017-09-25 Richard Biener <rguenther@suse.de>
7736 PR tree-optimization/82285
7737 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7740 2017-09-25 Tom de Vries <tom@codesourcery.com>
7744 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7747 2017-09-25 Richard Biener <rguenther@suse.de>
7749 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7750 ISL errors other than isl_error_quota happen. Dump if the
7751 schedule is the same.
7752 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7753 errors instead of aborting inside ISL.
7755 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7758 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7759 of libgcc_eh for m64.
7760 * config/i386/darwin64.h: Likewise.
7762 2017-09-25 Richard Biener <rguenther@suse.de>
7765 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7766 attribute for incomplete types nor twice for complete ones.
7768 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7771 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7772 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7775 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7778 * config/i386/i386.opt (mprefer-avx256): Use
7779 ix86_target_flags variable.
7780 * config/i386/i386.c (ix86_target_string): Move
7781 -mprefer-avx256 to flag2_opts.
7783 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7786 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7787 and x != -1 | y != -1 into (x & y) != -1.
7789 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7791 * config.gcc: Add new case statement to set
7792 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7793 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7794 s390x-*-linux* case statements. Added aarch64 to the list of
7795 supported architectures.
7797 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7799 PR tree-optimization/82289
7800 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7801 STMT_VINFO_RELEVANT_P.
7803 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7804 Alan Hayward <alan.hayward@arm.com>
7805 David Sherwood <david.sherwood@arm.com>
7807 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7808 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7809 (extract_range_from_binary_expr_1): Don't call
7810 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7812 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7813 Alan Hayward <alan.hayward@arm.com>
7814 David Sherwood <david.sherwood@arm.com>
7816 * target.def (preferred_vector_alignment): New hook.
7817 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7819 * doc/tm.texi: Regenerate.
7820 * targhooks.h (default_preferred_vector_alignment): Declare.
7821 * targhooks.c (default_preferred_vector_alignment): New function.
7822 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7824 (DR_TARGET_ALIGNMENT): New macro.
7825 (aligned_access_p): Update commentary.
7826 (vect_known_alignment_in_bytes): New function.
7827 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7829 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7830 Calculate the misalignment based on the target alignment rather than
7832 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7833 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7834 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7835 the target alignment, rather than masking the element misalignment
7836 with the number of elements in a vector. Also use the target
7837 alignment when calculating the maximum number of peels.
7838 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7839 instead of TYPE_ALIGN_UNIT.
7840 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7841 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7842 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7843 (vect_create_data_ref_ptr): Likewise.
7844 (vect_setup_realignment): Realign by ANDing with
7845 -DR_TARGET_MISALIGNMENT.
7846 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7847 the number of peels based on DR_TARGET_ALIGNMENT.
7848 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7849 with the guaranteed alignment boundary when deciding whether
7851 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7852 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7853 (ensure_base_align): Remove stmt_info parameter. Get the
7854 target base alignment from DR_TARGET_ALIGNMENT.
7855 (vectorizable_store): Update call accordingly. Interpret
7856 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7858 (vectorizable_load): Likewise.
7860 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7861 Alan Hayward <alan.hayward@arm.com>
7862 David Sherwood <david.sherwood@arm.com>
7864 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7865 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7866 (vect_enhance_data_refs_alignment): Likewise.
7868 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7870 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7871 error. Only quit immediately if parsing is complete.
7872 (BEGIN): Initialize fatal_err and parse_done.
7873 (begin fpu, end fpu): Check number of arguments.
7874 (begin arch, end arch): Likewise.
7875 (begin cpu, end cpu): Likewise.
7876 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7877 (optalias): Likewise.
7879 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7881 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7882 * config/arm/arm-isa.h: Delete. Move definitions to ...
7883 * arm-cpus.in: ... here. Use new feature and fgroup values.
7884 * config/arm/arm.c (arm_option_override): Use lower case for feature
7886 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7887 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7888 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7890 (print_isa_bits_for): New function.
7891 (gen_isa): New function.
7892 (gen_comm_data): Use print_isa_bits_for.
7893 (define feature): New keyword.
7894 (define fgroup): New keyword.
7895 * config/arm/t-arm (TM_H): Remove.
7896 (GTM_H): Add arm-isa.h.
7897 (arm-isa.h): Add rule to generate file.
7898 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7899 case for feature bit names.
7901 2017-09-22 Richard Biener <rguenther@suse.de>
7903 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7905 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7906 print of no dependency loops ...
7907 * graphite.c (graphite_transform_loops): ... here.
7908 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7910 (same_close_phi_node, remove_duplicate_close_phi,
7911 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7912 (canonicalize_loop_closed_ssa): ... here and simplify.
7913 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7914 (optimize_isl): Use dump_printf_loc to tell when we stopped
7915 optimizing because of an ISL timeout.
7917 2017-09-22 Richard Biener <rguenther@suse.de>
7919 PR tree-optimization/82291
7920 * tree-if-conv.c (predicate_mem_writes): Make sure to
7921 remove writes in blocks predicated with false.
7923 2017-09-22 Richard Biener <rguenther@suse.de>
7925 * sese.c: Include cfganal.h.
7926 (if_region_set_false_region): Remove.
7927 (create_if_region_on_edge): Likewise.
7928 (move_sese_in_condition): Re-implement without destroying
7931 2017-09-22 Richard Biener <rguenther@suse.de>
7933 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7934 Verify both BBs contain loop PHI nodes before dispatching to
7936 (graphite_regenerate_ast_isl): Do not recompute dominators,
7937 do not verify three times. Restructure for clarity.
7938 * graphite-scop-detection.c (same_close_phi_node,
7939 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7940 defined_in_loop_p, canonicalize_loop_closed_ssa,
7941 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7942 checking and SSA rewrite, move to ...
7943 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7944 (graphite_initialize): Do not pass in ctx, do not reset the
7945 SCEV cache, compute only dominators.
7946 (graphite_transform_loops): Allocate ISL ctx after
7947 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7948 Maintain post-dominators only around build_scops.
7949 * sese.c (if_region_set_false_region): Make static. Free
7950 and recompute dominators.
7951 (move_sese_in_condition): Assert we don't get called with
7952 post-dominators computed.
7953 * sese.h (if_region_set_false_region): Remove.
7955 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7957 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7958 mode attribute for TARGET_AVX512VL.
7960 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7962 * config/i386/i386.opt (mprefer-avx256): New option.
7963 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7965 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7966 for TARGET_PREFER_AVX256.
7967 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7969 2017-09-21 Jeff Law <law@redhat.com>
7971 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7972 Fix dump output if the only stack space is for pushed registers.
7974 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7976 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7979 2017-09-21 Martin Sebor <msebor@redhat.com>
7982 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7983 code (in C++) or code that triggers warnings.
7985 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7987 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7989 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7991 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7992 * sched-rgn.c: Ditto.
7993 * sel-sched-ir.c: Ditto.
7995 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7997 * toplev.h (set_random_seed): Adjust return type.
7998 * toplev.c (init_local_tick): Move eager initialization of random_seed
7999 to get_random_seed. Adjust comment.
8000 (init_random_seed): Inline to get_random_seed, delete.
8001 (get_random_seed): Initialize random_seed lazily.
8002 (set_random_seed): Do not return previous value.
8003 (print_switch_value): Do not call get_random_seed.
8005 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8007 * cgraph.c (delete_function_version): New, broken out from...
8008 (cgraph_node::delete_function_version): ...here. Rename to
8009 cgraph_node::delete_function_version_by_decl. Update all uses.
8010 (cgraph_node::remove): Call delete_function_version.
8012 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8015 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8016 VAR_DECLs to which addressable non-volatile parameters are mapped
8017 and for id->retvar after the return value assignment. Clear
8018 id->retval and id->retbnd after inlining.
8020 2017-09-21 Richard Biener <rguenther@suse.de>
8022 PR tree-optimization/82276
8023 PR tree-optimization/82244
8024 * tree-vrp.c (build_assert_expr_for): Set
8025 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8027 (remove_range_assertions): Revert earlier change.
8029 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8032 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8034 2017-09-21 Richard Biener <rguenther@suse.de>
8036 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8037 Restore valid IL after code generation errors.
8038 * graphite.c (graphite_transform_loops): Diagnose code
8039 generation issues as MSG_MISSED_OPTIMIZATION and continue
8040 with processing SCOPs.
8042 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8043 Alan Hayward <alan.hayward@arm.com>
8044 David Sherwood <david.sherwood@arm.com>
8046 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8047 rather than choosing between plus_constant and gen_rtx_<CODE>.
8048 * expr.c (emit_push_insn): Likewise.
8049 (expand_expr_real_2): Likewise.
8051 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8052 Alan Hayward <alan.hayward@arm.com>
8053 David Sherwood <david.sherwood@arm.com>
8055 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8057 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8058 Alan Hayward <alan.hayward@arm.com>
8059 David Sherwood <david.sherwood@arm.com>
8061 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8062 calling tree_to_uhwi.
8064 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8065 Alan Hayward <alan.hayward@arm.com>
8066 David Sherwood <david.sherwood@arm.com>
8068 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8069 INTEGER_CST rather than a negative test for ADDR_EXPR.
8071 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8072 Alan Hayward <alan.hayward@arm.com>
8073 David Sherwood <david.sherwood@arm.com>
8075 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8076 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8078 2017-09-21 Richard Biener <rguenther@suse.de>
8080 PR tree-optimization/71351
8081 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8082 graphite_create_new_loop_guard): Remove, fold remaining parts
8084 (translate_isl_ast_node_for): ... here and simplify.
8086 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8089 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8090 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8091 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8092 alternative always use QI mode, for -Os imov (=R,R) alternative
8093 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8096 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8097 Jeff Law <law@redhat.com>
8099 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8100 (allocate_stack_space): New function, partially extracted from
8102 (s390_emit_prologue): Track offset to most recent stack probe.
8103 Code to allocate space moved into allocate_stack_space.
8104 Dump actions when no stack is allocated.
8105 (s390_prologue_plus_offset): New function.
8106 (s390_emit_stack_probe): Likewise.
8108 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8110 * common.opt (Wa, Wl, Wp, g, gz=): Add
8112 (gno-column-info): Remove.
8113 (gcolumn-info): Drop RejectNegative.
8115 (gno-record-gcc-switches): Remove.
8116 (grecord-gcc-switches): Drop RejectNegative.
8117 (gno-split-dwarf): Remove.
8118 (gsplit-dwarf): Drop RejectNegative.
8119 (gno-strict-dwarf): Remove.
8120 (gstrict-dwarf): Drop RejectNegative.
8121 * config/darwin.opt (gfull, gused): Add RejectNegative.
8122 * dwarf2out.c (gen_producer_string): Drop
8123 gno-record-gcc-switches handler.
8124 * optc-gen.awk: Add g to prefixes with negative forms.
8125 * opts-common.c (remapping_prefix_p): New.
8126 (find_opt): Check it.
8127 (generate_canonical_option): Test g prefix.
8128 (option_map): Add -gno- mapping.
8129 (add_misspelling_candidates): Check remapping_prefix_p.
8131 2017-09-20 Jeff Law <law@redhat.com>
8133 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8134 thinko in stack clash protection support.
8136 * explow.c (compute_stack_clash_protection_loop_data): Use
8137 CONST_INT_P instead of explicit test. Verify object is a
8138 CONST_INT_P before looking at INTVAL.
8139 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8140 instead of explicit test.
8142 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8145 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8146 address instead of to r1 and r11.
8148 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8150 * config.gcc: Support "knm".
8151 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8152 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8154 * config/i386/i386.c (m_KNM): Define.
8155 (processor_target_table): Add "knm".
8157 (ix86_option_override_internal): Add "knm".
8158 (ix86_issue_rate): Add PROCESSOR_KNM.
8159 (ix86_adjust_cost): Ditto.
8160 (ia32_multipass_dfa_lookahead): Ditto.
8161 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8162 (fold_builtin_cpu): Add M_INTEL_KNM.
8163 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8164 (processor_type): Add PROCESSOR_KNM.
8165 * config/i386/x86-tune.def: Add m_KNM.
8166 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8168 2017-09-20 Richard Biener <rguenther@suse.de>
8170 PR tree-optimization/80213
8171 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8172 are allowed in empty BBs as well.
8173 (canonicalize_loop_closed_ssa): Also look for other complex
8175 (scop_detection::get_sese): Include the loop-closed PHI block
8177 (scop_detection::merge_sese): Remove code adding extra blocks.
8178 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8179 (build_scops): Assert the final returned scop is invalid.
8181 2017-09-20 Richard Biener <rguenther@suse.de>
8183 PR tree-optimization/82264
8184 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8186 (vn_phi_lookup): Likewise.
8187 (vn_phi_insert): Likewise.
8189 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8191 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8192 that fits into uhwi or shwi, add DW_AT_const_value regardless
8193 of early_dwarf without going through RTL, using add_AT_unsigned
8196 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8197 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8198 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8199 (DEBUG_MACRO_SECTION): Likewise.
8200 (DEBUG_LTO_MACRO_SECTION): Likewise.
8201 (DEBUG_STR_DWO_SECTION): Likewise.
8202 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8203 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8204 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8205 (DEBUG_LTO_LINE_STR_SECTION): Define.
8206 (init_sections_and_labels): Initialize debug_line_str_section
8207 variable. Initialize debug_loc_section for -gdwarf-5 to
8208 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8210 2017-09-20 Richard Biener <rguenther@suse.de>
8212 * graphite-sese-to-poly.c (extract_affine): Properly handle
8213 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8215 2017-09-20 Richard Biener <rguenther@suse.de>
8217 PR tree-optimization/81373
8218 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8219 Force SESE live-out defs to be handled even if they are
8222 2017-09-19 Jeff Law <law@redhat.com>
8224 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8225 nothing for stack adjustments with REG_STACK_CHECK.
8226 * sched-deps.c (parse_add_or_inc): Reject insns with
8227 REG_STACK_CHECK from dependency breaking.
8228 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8229 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8230 * reg-notes.def (STACK_CHECK): New note.
8232 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8233 (ix86_expand_prologue): Dump stack clash info as needed.
8234 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8236 * function.c (dump_stack_clash_frame_info): New function.
8237 * function.h (dump_stack_clash_frame_info): Prototype.
8238 (enum stack_clash_probes): New enum.
8240 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8241 flag_stack_clash_protection.
8242 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8243 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8244 (arm_frame_pointer_required): Likewise.
8245 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8246 (ia64_expand_prologue): Likewise.
8247 * config/mips/mips.c (mips_expand_prologue): Likewise.
8248 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8249 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8250 (sparc_flat_expand_prologue): Likewise.
8251 * config/spu/spu.c (spu_expand_prologue): Likewise.
8253 * explow.c: Include "params.h".
8254 (anti_adjust_stack_and_probe_stack_clash): New function.
8255 (get_stack_check_protect): Likewise.
8256 (compute_stack_clash_protection_loop_data): Likewise.
8257 (emit_stack_clash_protection_loop_start): Likewise.
8258 (emit_stack_clash_protection_loop_end): Likewise.
8259 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8260 Use anti_adjust_stack_and_probe_stack_clash.
8261 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8262 (emit_stack_clash_protection_loop_start): Likewise.
8263 (emit_stack_clash_protection_loop_end): Likewise.
8264 * rtl.h (get_stack_check_protect): Prototype.
8265 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8266 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8267 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8269 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8271 * doc/tm.texi: Rebuilt.
8272 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8273 get_stack_check_protect.
8274 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8275 * config/arm/arm.c (arm_expand_prologue): Likewise.
8276 (arm_frame_pointer_required): Likewise.
8277 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8278 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8279 * config/mips/mips.c (mips_expand_prologue): Likewise.
8280 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8281 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8282 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8283 (sparc_flat_expand_prologue): Likewise.
8285 * common.opt (-fstack-clash-protection): New option.
8286 * flag-types.h (enum stack_check_type): Note difference between
8287 -fstack-check= and -fstack-clash-protection.
8288 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8289 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8290 * toplev.c (process_options): Issue warnings/errors for cases
8291 not handled with -fstack-clash-protection.
8292 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8293 (-fstack-check): Note additional problem with -fstack-check=generic.
8294 Note that -fstack-check is primarily for Ada and refer users
8295 to -fstack-clash-protection for stack-clash-protection.
8296 Document new params for stack clash protection.
8298 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8300 * config/i386/i386.c (ix86_split_long_move): Do not handle
8301 address used for LEA in a special way.
8303 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8305 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8307 2017-09-19 Martin Sebor <msebor@redhat.com>
8310 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8311 of incompatible types.
8313 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8315 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8316 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8317 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8318 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8320 2017-09-19 Richard Biener <rguenther@suse.de>
8322 PR tree-optimization/82244
8323 * tree-vrp.c (remove_range_assertions): Do not propagate
8324 a constant to abnormals but replace the assert with a copy.
8326 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8328 PR rtl-optimization/57878
8329 PR rtl-optimization/68988
8330 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8331 avoidance test involving non_reload_pseudos. Move frequency test
8332 below the general fragmentation avoidance test.
8334 2017-09-19 Richard Biener <rguenther@suse.de>
8336 PR tree-optimization/69728
8337 * graphite-sese-to-poly.c (schedule_error): New global.
8338 (add_loop_schedule): Handle empty domain by failing the
8340 (build_original_schedule): Handle schedule_error.
8342 2017-09-19 Richard Biener <rguenther@suse.de>
8344 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8345 Do not iterate to sibling loops but only to siblings of inner
8348 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8351 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8352 that the registers are different.
8354 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8356 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8357 to processor_model and "amdfam17h" to arch_names_table.
8358 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8360 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8363 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8365 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8366 Alan Hayward <alan.hayward@arm.com>
8367 David Sherwood <david.sherwood@arm.com>
8369 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8371 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8373 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8374 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8375 vect_schedule_slp_instance.
8376 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8377 Update call to vect_slp_analyze_node_operations. Simplify return
8379 (vect_slp_analyze_bb_1): Update call accordingly.
8380 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8381 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8382 (vect_schedule_slp): Update call accordingly.
8384 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8385 Alan Hayward <alan.hayward@arm.com>
8386 David Sherwood <david.sherwood@arm.com>
8388 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8389 with types that aren't in fact scalar.
8391 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8393 * tree-vect-slp.c (vect_record_max_nunits): New function,
8395 (vect_build_slp_tree_1): ...here.
8396 (vect_build_slp_tree_2): Call it for phis too.
8398 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8400 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8401 to vect_get_vec_def_for_operand when getting the mask operand.
8403 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8404 Alan Hayward <alan.hayward@arm.com>
8405 David Sherwood <david.sherwood@arm.com>
8407 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8410 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8411 Alan Hayward <alan.hayward@arm.com>
8412 David Sherwood <david.sherwood@arm.com>
8414 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8415 calculation for vector booleans.
8417 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8418 Alan Hayward <alan.hayward@arm.com>
8419 David Sherwood <david.sherwood@arm.com>
8421 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8423 (vect_transform_stmt): ...here.
8424 (vect_analyze_stmt): Use it instead of calling
8425 vectorizable_live_operation directly.
8427 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8429 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8430 non-SIMT targets in acc vector loops.
8432 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8434 * configure.ac: Add arc and check if assembler supports gdwarf2.
8435 * configure: Regenerate.
8437 2017-09-18 Richard Biener <rguenther@suse.de>
8439 PR tree-optimization/82220
8440 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8441 epilogue niters from the min_profitable_iters compute.
8443 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8446 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8447 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8448 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8450 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8453 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8454 switching to a new text section.
8456 2017-09-18 Richard Biener <rguenther@suse.de>
8458 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8460 (build_alias_set): Reject aliases with no access function.
8462 2017-09-18 Richard Biener <rguenther@suse.de>
8464 PR tree-optimization/79622
8465 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8467 (build_cross_bb_scalars_use): Likewise.
8469 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8471 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8473 2017-09-18 Alan Modra <amodra@gmail.com>
8476 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8477 stack_pointer_rtx for count 0. Update comments. Break up
8478 large rtl expression.
8480 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8482 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8483 Increase to 20 bytes.
8484 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8485 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8486 or avx version of the stub.
8488 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8491 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8492 compute the minimum stack alignment. Also update preferred stack
8493 boundary for leaf functions.
8495 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8497 PR tree-optimization/82228
8498 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8501 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8503 * common/config/nds32/nds32-common.c
8504 (nds32_option_optimization_table): Refine formatting.
8505 (nds32_option_optimization_table): Use -fsched-pressure and
8506 -fomit-frame-pointer for specific optimization level.
8508 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8510 * config/nds32/nds32.c: Refine formatting and comments.
8511 * config/nds32/nds32.h: Likewise.
8512 * config/nds32/nds32.md: Likewise.
8513 * config/nds32/nds32-cost.c: Likewise.
8514 * config/nds32/nds32-isr.c: Likewise.
8515 * config/nds32/nds32-md-auxiliary.c: Likewise.
8516 * config/nds32/nds32-multiple.md: Likewise.
8517 * config/nds32/nds32-predicates.c: Likewise.
8519 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8520 Jakub Jelinek <jakub@redhat.com>
8522 Add support for -std=c++2a.
8523 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8525 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8527 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8530 * doc/extend.texi (Common Function Attributes): Add
8531 references to ARM, AArch64, and S/390 specific attributes.
8532 (Function Specific Option Pragmas): Add AArch64 and S/390
8533 to list of back ends that support the target pragma.
8535 2017-09-15 Nathan Sidwell <nathan@acm.org>
8537 * doc/standards.texi: Fix C++17 description. Update URLs for
8540 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8542 * common.opt (Wcast-align=strict): New warning option.
8543 * doc/invoke.texi: Document -Wcast-align=strict.
8545 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8547 * cgraph.h (cgraph_thunk_info): Add comments.
8548 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8549 assert for VIRTUAL_* arguments stricter.
8551 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8553 PR tree-optimization/71026
8554 * match.pd: Move RDIV patterns from fold-const.c
8555 * fold-const.c (distribute_real_division): Removed.
8556 (fold_binary_loc): Remove calls to distribute_real_divison.
8558 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8560 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8561 c++1z and gnu++1z as deprecated. Change other references to
8562 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8563 Change -Wc++1z-compat to -Wc++17-compat.
8564 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8565 * dwarf2out.c (highest_c_language): Handle C++17.
8566 (gen_compile_unit_die): Likewise.
8568 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8570 PR rtl-optimization/82192
8571 * combine.c (make_extraction): Don't look through non-paradoxical
8572 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8575 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8576 Alan Hayward <alan.hayward@arm.com>
8577 David Sherwood <david.sherwood@arm.com>
8579 * target.def (function_arg_offset): New hook.
8580 * targhooks.h (default_function_arg_offset): Declare.
8581 * targhooks.c (default_function_arg_offset): New function.
8582 * function.c (locate_and_pad_parm): Use
8583 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8584 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8585 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8586 * doc/tm.texi: Regenerate.
8587 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8588 * config/spu/spu.c (spu_function_arg_offset): New function.
8589 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8590 * system.h (FUNCTION_ARG_OFFSET): Poison.
8592 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8593 Alan Hayard <alan.hayward@arm.com>
8594 David Sherwood <david.sherwood@arm.com>
8596 * target.def (truly_noop_truncation): New hook.
8597 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8598 than TRULY_NOOP_TRUNCATION.
8599 * hooks.h (hook_bool_uint_uint_true): Declare.
8600 * hooks.c (hook_bool_uint_uint_true): New function.
8601 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8602 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8603 * doc/tm.texi: Regenerate.
8604 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8605 rather than TRULY_NOOP_TRUNCATION in comments.
8606 (simplify_comparison): Likewise.
8607 (record_truncated_value): Likewise.
8608 * expmed.c (extract_bit_field_1): Likewise.
8609 (extract_split_bit_field): Likewise.
8610 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8611 instead of TRULY_NOOP_TRUNCATION.
8612 * function.c (assign_parm_setup_block): Likewise.
8613 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8614 * rtlhooks.c: Include target.h.
8615 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8616 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8617 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8618 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8619 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8620 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8621 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8622 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8623 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8624 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8625 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8626 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8627 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8628 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8629 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8630 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8631 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8632 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8633 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8634 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8635 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8636 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8637 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8638 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8639 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8640 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8641 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8642 rather than TRULY_NOOP_TRUNCATION in comments.
8643 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8644 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8645 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8646 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8647 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8648 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8649 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8650 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8651 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8652 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8653 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8654 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8655 rather than TRULY_NOOP_TRUNCATION in comments.
8656 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8657 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8658 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8659 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8660 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8661 TRULY_NOOP_TRUNCATION condition.
8662 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8663 (TRULY_NOOP_TRUNCATION): Delete.
8664 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8665 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8666 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8667 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8668 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8669 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8670 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8671 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8672 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8673 rather than TRULY_NOOP_TRUNCATION in comments.
8674 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8675 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8676 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8677 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8678 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8679 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8681 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8684 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8685 (*cmp_ior): Likewise.
8686 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8687 (*ior_scc_scc_cmp): Likewise.
8688 (*and_scc_scc): Likewise.
8689 (*and_scc_scc_cmp): Likewise.
8691 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8692 Alan Hayard <alan.hayward@arm.com>
8693 David Sherwood <david.sherwood@arm.com>
8695 * target.def (can_change_mode_class): New hook.
8696 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8697 (hard_regno_nregs): Likewise.
8698 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8699 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8700 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8701 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8702 (LOAD_EXTEND_OP): Update accordingly.
8703 * doc/tm.texi: Regenerate.
8704 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8705 CANNOT_CHANGE_MODE_CLASS.
8706 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8707 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8708 * combine.c (simplify_set): Update accordingly.
8709 * emit-rtl.c (validate_subreg): Likewise.
8710 * recog.c (general_operand): Likewise.
8711 * regcprop.c (mode_change_ok): Likewise.
8712 * reload1.c (choose_reload_regs): Likewise.
8713 (inherit_piecemeal_p): Likewise.
8714 * rtlanal.c (simplify_subreg_regno): Likewise.
8715 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8716 instead of CANNOT_CHANGE_MODE_CLASS.
8717 (reload_cse_simplify_operands): Likewise.
8718 * reload.c (push_reload): Use targetm.can_change_mode_class
8719 instead of CANNOT_CHANGE_MODE_CLASS.
8720 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8721 REG_CANNOT_CHANGE_MODE_P.
8722 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8723 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8724 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8725 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8726 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8727 (arm_can_change_mode_class): New function.
8728 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8729 than CANNOT_CHANGE_MODE_CLASS in comments.
8730 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8731 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8732 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8733 (ix86_can_change_mode_class): ...this new function, inverting the
8734 sense of the return value.
8735 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8736 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8737 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8738 (ia64_can_change_mode_class): New function.
8739 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8740 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8741 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8742 (m32c_can_change_mode_class): ...this new function, inverting the
8743 sense of the return value.
8744 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8745 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8746 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8747 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8748 (mips_can_change_mode_class): ...this new function, inverting the
8749 sense of the return value.
8750 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8751 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8752 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8753 (msp430_can_change_mode_class): New function.
8754 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8755 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8756 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8757 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8758 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8759 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8760 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8761 (pa_cannot_change_mode_class): Replace with...
8762 (pa_can_change_mode_class): ...this new function, inverting the
8763 sense of the return value.
8764 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8765 than CANNOT_CHANGE_MODE_CLASS in comments.
8766 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8767 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8768 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8769 (pdp11_cannot_change_mode_class): Replace with...
8770 (pdp11_can_change_mode_class): ...this new function, inverting the
8771 sense of the return value.
8772 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8773 * config/powerpcspe/powerpcspe-protos.h
8774 (rs6000_cannot_change_mode_class_ptr): Delete.
8775 * config/powerpcspe/powerpcspe.c
8776 (rs6000_cannot_change_mode_class_ptr): Delete.
8777 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8778 (rs6000_option_override_internal): Assign to
8779 targetm.can_change_mode_class instead of
8780 rs6000_cannot_change_mode_class_ptr.
8781 (rs6000_cannot_change_mode_class): Replace with...
8782 (rs6000_can_change_mode_class): ...this new function, inverting the
8783 sense of the return value.
8784 (rs6000_debug_cannot_change_mode_class): Replace with...
8785 (rs6000_debug_can_change_mode_class): ...this new function.
8786 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8787 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8788 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8789 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8790 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8792 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8793 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8794 (rs6000_option_override_internal): Assign to
8795 targetm.can_change_mode_class instead of
8796 rs6000_cannot_change_mode_class_ptr.
8797 (rs6000_cannot_change_mode_class): Replace with...
8798 (rs6000_can_change_mode_class): ...this new function, inverting the
8799 sense of the return value.
8800 (rs6000_debug_cannot_change_mode_class): Replace with...
8801 (rs6000_debug_can_change_mode_class): ...this new function.
8802 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8803 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8804 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8805 (s390_can_change_mode_class): ...this new function, inverting the
8806 sense of the return value.
8807 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8808 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8809 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8810 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8811 (sh_cannot_change_mode_class): Replace with...
8812 (sh_can_change_mode_class): ...this new function, inverting the
8813 sense of the return value.
8814 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8815 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8816 (sparc_can_change_mode_class): New function.
8817 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8818 * config/spu/spu.c (spu_can_change_mode_class): New function.
8819 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8820 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8821 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8822 (visium_can_change_mode_class): New function.
8823 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8825 2017-09-15 Richard Biener <rguenther@suse.de>
8827 PR tree-optimization/82217
8828 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8829 but not undefined case.
8831 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8834 * postreload.c (reload_cse_simplify_operands): Skip
8835 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8837 2017-09-15 Richard Biener <rguenther@suse.de>
8839 PR tree-optimization/68823
8840 * graphite-scop-detection.c (build_alias_set): If we have a
8841 possible dependence check whether we can handle them by just
8842 looking at the DRs DR_ACCESS_FNs.
8843 (build_scops): If build_alias_set fails, fail the SCOP.
8845 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8847 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8848 to support float128 built-in functions that require the ISA 3.0
8850 (BU_FLOAT128_3_HW): Likewise.
8851 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8853 (FMAF128): Likewise.
8855 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8856 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8857 floating point instructions.
8858 (rs6000_invalid_builtin): Likewise.
8859 (rs6000_builtin_mask_names): Likewise.
8860 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8861 (RS6000_BTM_FLOAT128_HW): Likewise.
8862 (RS6000_BTM_COMMON): Likewise.
8863 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8865 * doc/extend.texi (RS/6000 built-in functions): Document the
8866 IEEE 128-bit floating point square root and fused multiply-add
8869 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8871 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8872 reg (r2) isn't in the set of registers defined in the prologue.
8874 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8875 Alan Hayward <alan.hayward@arm.com>
8876 David Sherwood <david.sherwood@arm.com>
8878 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8879 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8880 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8881 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8882 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8884 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8885 max_vectorization_factor.
8886 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8888 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8889 Alan Hayward <alan.hayward@arm.com>
8890 David Sherwood <david.sherwood@arm.com>
8892 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8893 (vect_worthwhile_without_simd_p): Declare.
8894 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8895 (vectorizable_reduction): Use it.
8896 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8897 (vectorizable_operation): Likewise.
8899 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8900 Alan Hayward <alan.hayward@arm.com>
8901 David Sherwood <david.sherwood@arm.com>
8903 * tree-vectorizer.h (vect_get_num_copies): New function.
8904 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8905 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8906 (vectorizable_induction): Likewise.
8907 (vectorizable_live_operation): Likewise.
8908 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8909 (vectorizable_bswap): Likewise.
8910 (vectorizable_call): Likewise.
8911 (vectorizable_conversion): Likewise.
8912 (vectorizable_assignment): Likewise.
8913 (vectorizable_shift): Likewise.
8914 (vectorizable_operation): Likewise.
8915 (vectorizable_store): Likewise.
8916 (vectorizable_load): Likewise.
8917 (vectorizable_condition): Likewise.
8918 (vectorizable_comparison): Likewise.
8919 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8921 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8922 Alan Hayward <alan.hayward@arm.com>
8923 David Sherwood <david.sherwood@arm.com>
8925 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8926 of vect_init_vector.
8928 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8929 Alan Hayward <alan.hayward@arm.com>
8930 David Sherwood <david.sherwood@arm.com>
8932 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8933 an inline wrapper that provides a location.
8934 (gimple_build_vector): Likewise.
8935 * gimple-fold.c (gimple_build_vector_from_val): New function.
8936 (gimple_build_vector): Likewise.
8937 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8938 functions to build the initial value. Always return a gimple value.
8939 (get_initial_defs_for_reduction): Likewise. Only compute
8941 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8942 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8943 (vectorizable_induction): Use gimple_build_vector rather than
8946 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8947 Alan Hayward <alan.hayward@arm.com>
8948 David Sherwood <david.sherwood@arm.com>
8950 * target.h (vec_perm_indices): New typedef.
8951 (auto_vec_perm_indices): Likewise.
8952 * optabs-query.h: Include target.h
8953 (can_vec_perm_p): Take a vec_perm_indices *.
8954 * optabs-query.c (can_vec_perm_p): Likewise.
8955 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8956 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8957 * tree-vect-generic.c (lower_vec_perm): Likewise.
8958 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8959 (vect_grouped_load_supported): Likewise.
8960 (vect_shift_permute_load_chain): Likewise.
8961 (vect_permute_store_chain): Use auto_vec_perm_indices.
8962 (vect_permute_load_chain): Likewise.
8963 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8964 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8965 Update uses of can_vec_perm_p.
8966 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8967 mode with a number of elements. Take a vec_perm_indices *.
8968 (vect_create_epilog_for_reduction): Update accordingly.
8969 Use auto_vec_perm_indices.
8970 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8971 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8972 (vect_transform_slp_perm_load): Likewise.
8973 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8974 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8975 (vect_gen_perm_mask_checked): Likewise.
8976 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8977 (vect_gen_perm_mask_checked): Likewise.
8978 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8979 (vectorizable_store): Likewise.
8980 (vectorizable_load): Likewise.
8981 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8982 (vectorizable_bswap): Likewise.
8984 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8985 Alan Hayward <alan.hayward@arm.com>
8986 David Sherwood <david.sherwood@arm.com>
8988 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8989 * tree.c (build_vector): Likewise.
8990 (build_vector_from_ctor): Update accordingly.
8991 (build_vector_from_val): Likewise.
8992 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8993 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8994 * tree-vect-generic.c (add_rshift): Likewise.
8995 (expand_vector_divmod): Likewise.
8996 (optimize_vector_constructor): Likewise.
8997 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8998 (vect_transform_slp_perm_load): Likewise.
8999 (vect_schedule_slp_instance): Likewise.
9000 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9001 (vectorizable_call): Likewise.
9002 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9003 * expmed.c (make_tree): Likewise.
9004 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9005 a vector passed to build_vector.
9006 (fold_convert_const): Likewise.
9007 (exact_inverse): Likewise.
9008 (fold_ternary_loc): Likewise.
9009 (fold_relational_const): Likewise.
9010 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9011 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9012 (const_unop): Likewise. Store the reduction accumulator in a
9013 variable rather than an array.
9014 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9015 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9016 the new vector, rather than constructing it after the input arrays.
9017 (native_interpret_vector): Use auto_vec<tree> when building
9018 a vector passed to build_vector. Add elements in order.
9019 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9020 auto_vec<tree> when building a vector passed to build_vector.
9021 (vect_create_epilog_for_reduction): Likewise.
9022 (vectorizable_induction): Likewise.
9023 (get_initial_def_for_reduction): Likewise. Fix indentation of
9025 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9027 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9028 passed to build_vector.
9030 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9031 Alan Hayward <alan.hayward@arm.com>
9032 David Sherwood <david.sherwood@arm.com>
9034 * tree-core.h (tree_base::u): Add an "nelts" field.
9035 (tree_vector): Use VECTOR_CST_NELTS as the length.
9036 * tree.c (tree_size): Likewise.
9037 (make_vector): Initialize VECTOR_CST_NELTS.
9038 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9039 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9040 TYPE_VECTOR_SUBPARTS.
9041 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9042 as the number of units, setting it from VECTOR_CST_NELTS.
9043 (const_vector_from_tree): Likewise.
9044 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9045 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9046 (fold_negate_expr_1): Likewise.
9047 (fold_convert_const): Likewise.
9048 (const_binop): Likewise. Differentiate the number of output and
9050 (const_unop): Likewise.
9051 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9052 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9053 in cases that did the opposite.
9055 2017-09-14 Richard Biener <rguenther@suse.de>
9057 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9060 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9062 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9064 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9067 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9068 if and where to split a bb, except for splitting before debug insn
9069 sequences followed by non-label real insn. Delete debug insns
9070 in between basic blocks.
9072 * combine.c (make_compound_operation_int): Formatting fixes.
9074 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9075 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9076 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9077 * config/sol2.h (LINK_EH_SPEC): Likewise.
9078 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9079 * config/s390/linux.h (LINK_SPEC): Likewise.
9080 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9081 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9082 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9083 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9084 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9085 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9086 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9087 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9089 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9091 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9093 (ENDFILE_LINUX_SPEC): Likewise.
9094 (LINK_EH_SPEC): Likewise.
9095 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9096 (LINK_OS_LINUX_SPEC32): Likewise.
9097 (LINK_OS_LINUX_SPEC64): Likewise.
9098 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9099 (LINK_OS_LINUX_SPEC): Likewise.
9101 2017-09-13 Martin Liska <mliska@suse.cz>
9104 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9105 CASE_HIGH is NULL_TREE.
9107 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9108 Alan Hayward <alan.hayward@arm.com>
9109 David Sherwood <david.sherwood@arm.com>
9111 * target.def (secondary_memory_needed): New hook.
9112 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9113 instead of SECONDARY_MEMORY_NEEDED.
9114 (secondary_memory_needed_mode): Likewise.
9115 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9116 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9117 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9118 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9119 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9120 * doc/tm.texi: Regenerate.
9121 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9122 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9123 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9124 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9125 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9126 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9127 mode argument first and change the reg_class arguments to reg_class_t.
9128 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9129 Make static. Update the call to inline_secondary_memory_needed.
9130 (ix86_register_move_cost): Update the call to
9131 inline_secondary_memory_needed.
9132 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9133 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9135 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9136 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9138 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9139 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9140 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9141 and match hook interface. Add comment from mips.h.
9142 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9143 * config/mmix/mmix.md (truncdfsf2): Refer to
9144 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9146 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9147 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9148 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9149 (pa_secondary_memory_needed): New function.
9150 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9151 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9152 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9153 (pdp11_secondary_memory_needed): Make static and match hook interface.
9154 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9155 * config/powerpcspe/powerpcspe-protos.h
9156 (rs6000_secondary_memory_needed_ptr): Delete.
9157 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9159 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9160 (rs6000_option_override_internal): Assign to
9161 targetm.secondary_memory_needed rather than
9162 rs6000_secondary_memory_needed_ptr.
9163 (rs6000_secondary_memory_needed): Match hook interface.
9164 (rs6000_debug_secondary_memory_needed): Likewise.
9165 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9166 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9167 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9168 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9169 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9170 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9172 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9173 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9174 (rs6000_option_override_internal): Assign to
9175 targetm.secondary_memory_needed rather than
9176 rs6000_secondary_memory_needed_ptr.
9177 (rs6000_secondary_memory_needed): Match hook interface.
9178 (rs6000_debug_secondary_memory_needed): Likewise.
9179 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9180 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9181 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9182 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9183 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9184 (sparc_secondary_memory_needed): New function.
9185 * lra-constraints.c (check_and_process_move): Refer to
9186 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9188 (curr_insn_transform): Likewise.
9189 (process_alt_operands): Use targetm.secondary_memory_needed
9190 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9191 (check_secondary_memory_needed_p): Likewise.
9192 (choose_split_class): Likewise.
9193 * reload.c: Unconditionally include code that was previously
9194 conditional on SECONDARY_MEMORY_NEEDED.
9195 (push_secondary_reload): Use targetm.secondary_memory_needed
9196 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9197 (push_reload): Likewise.
9198 * reload1.c: Unconditionally include code that was previously
9199 conditional on SECONDARY_MEMORY_NEEDED.
9200 (choose_reload_regs): Use targetm.secondary_memory_needed
9201 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9202 (gen_reload): Likewise.
9203 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9205 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9206 Alan Hayward <alan.hayward@arm.com>
9207 David Sherwood <david.sherwood@arm.com>
9209 * target.def (secondary_memory_needed_mode): New hook:
9210 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9211 * targhooks.h (default_secondary_memory_needed_mode): New function.
9212 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9213 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9214 * doc/tm.texi: Regenerate.
9215 * lra-constraints.c (check_and_process_move): Use
9216 targetm.secondary_memory_needed_mode instead of
9217 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9218 (curr_insn_transform): Likewise.
9219 * reload.c (get_secondary_mem): Likewise.
9220 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9221 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9223 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9224 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9225 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9226 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9227 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9229 * config/powerpcspe/powerpcspe-protos.h
9230 (rs6000_secondary_memory_needed_mode): Delete.
9231 * config/powerpcspe/powerpcspe.c
9232 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9233 (rs6000_secondary_memory_needed_mode): Make static.
9234 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9235 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9237 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9239 (rs6000_secondary_memory_needed_mode): Make static.
9240 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9241 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9242 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9243 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9244 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9246 (sparc_secondary_memory_needed_mode): New function.
9247 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9249 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9251 * config/aarch64/constraints.md (Umq): New constraint.
9252 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9254 (mov<mode>): Update condition.
9256 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9258 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9259 when bitposition is the same.
9261 2017-09-13 Richard Biener <rguenther@suse.de>
9263 * dwarf2out.c (output_die_symbol): Remove.
9264 (output_die): Do not output a DIEs symbol.
9266 2017-09-13 Richard Biener <rguenther@suse.de>
9269 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9270 default-def to avoid breaking iterator update with the weird
9271 interaction with cgraph_update_edges_for_call_stmt_node.
9273 2017-09-13 Richard Biener <rguenther@suse.de>
9275 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9276 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9277 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9278 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9280 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9282 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9283 Disable pc relative literal load irrespective of
9284 TARGET_FIX_ERR_A53_84341 for default.
9286 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9288 * config/sparc/sparc.c (output_return): Output the source location of
9289 the insn in the delay slot, if any.
9290 (output_sibcall): Likewise.
9292 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9295 * common.opt (-static-pie): New alias.
9296 (shared): Negate static-pie.
9297 (-no-pie): Update help text.
9299 (static-pie): New option.
9300 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9301 -static-pie support.
9302 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9303 (LINK_EH_SPEC): Likewise.
9304 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9305 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9306 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9307 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9308 (init_gcc_specs): Likewise.
9309 (init_spec): Likewise.
9310 (display_help): Update help message for -pie.
9311 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9314 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9316 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9317 (movdi_aarch64): Likewise.
9318 (movti_aarch64): Likewise.
9320 2017-09-12 Simon Wright <simon@pushface.org>
9323 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9324 calculation of the minor version, always output as 0.
9326 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9329 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9330 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9331 on it early, rather than manual conversion late. For
9332 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9333 instead of performing manual conversion.
9335 2017-09-12 Carl Love <cel@us.ibm.com>
9337 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9338 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9340 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9341 VMULOSW): Add definitions.
9342 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9343 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9344 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9345 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9346 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9348 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9350 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9352 (movti_aarch64): Likewise.
9353 (movdf_aarch64): Likewise.
9354 (movtf_aarch64): Likewise.
9355 (load_pairdi): Likewise.
9356 (store_pairdi): Likewise.
9357 (load_pairdf): Likewise.
9358 (store_pairdf): Likewise.
9359 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9360 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9361 (ldr_got_small_<mode>): Likewise.
9362 (ldr_got_small_28k_<mode>): Likewise.
9363 (ldr_got_tiny): Likewise.
9364 * config/aarch64/iterators.md (ldst_sz): New.
9365 (ldpstp_sz): Likewise.
9366 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9368 (thunderx_load): Split load_8 to load_16.
9369 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9371 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9372 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9373 (xgene1_store_pair): Split store_8 to store_16.
9374 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9375 (falkor_st_0_st_sd): Split store_8 to store_16.
9377 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9379 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9380 and store1/2/3/4 to store_4/8/12/16.
9381 * config/aarch64/aarch64.md: Update for rename.
9382 * config/arm/arm.md: Likewise.: Likewise.
9383 * config/arm/arm.c: Likewise.
9384 * config/arm/thumb1.md: Likewise.
9385 * config/arm/thumb2.md: Likewise.
9386 * config/arm/vfp.md: Likewise.
9387 * config/arm/arm-generic.md: Likewise.
9388 * config/arm/arm1020e.md: Likewise.
9389 * config/arm/arm1026ejs.md: Likewise.
9390 * config/arm/arm1136jfs.md: Likewise.
9391 * config/arm/arm926ejs.md: Likewise.
9392 * config/arm/cortex-a15.md: Likewise.
9393 * config/arm/cortex-a17.md: Likewise.
9394 * config/arm/cortex-a5.md: Likewise.
9395 * config/arm/cortex-a53.md: Likewise.
9396 * config/arm/cortex-a57.md: Likewise.
9397 * config/arm/cortex-a7.md: Likewise.
9398 * config/arm/cortex-a8.md: Likewise.
9399 * config/arm/cortex-a9.md: Likewise.
9400 * config/arm/cortex-m4.md: Likewise.
9401 * config/arm/cortex-m7.md: Likewise.
9402 * config/arm/cortex-r4.md: Likewise.
9403 * config/arm/exynos-m1.md: Likewise.
9404 * config/arm/fa526.md: Likewise.
9405 * config/arm/fa606te.md: Likewise.
9406 * config/arm/fa626te.md: Likewise.
9407 * config/arm/fa726te.md: Likewise.
9408 * config/arm/fmp626.md: Likewise.
9409 * config/arm/iwmmxt.md: Likewise.
9410 * config/arm/ldmstm.md: Likewise.
9411 * config/arm/marvell-pj4.md: Likewise.
9412 * config/arm/xgene1.md: Likewise.
9413 * config/aarch64/thunderx.md: Likewise.
9414 * config/aarch64/thunderx2t99.md: Likewise.
9415 * config/aarch64/falkor.md: Likewise.
9417 2017-09-12 Martin Liska <mliska@suse.cz>
9419 * attribs.c (private_lookup_attribute): New function.
9420 * attribs.h (private_lookup_attribute): Declared here.
9421 (lookup_attribute): Called from this place.
9423 2017-09-12 Richard Biener <rguenther@suse.de>
9425 PR tree-optimization/82157
9426 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9427 stmts with side-effects.
9429 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9430 Alan Hayward <alan.hayward@arm.com>
9431 David Sherwood <david.sherwood@arm.com>
9433 * target.def (hard_regno_nregs): New hook.
9434 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9435 * targhooks.h (default_hard_regno_nregs): Declare.
9436 * targhooks.c (default_hard_regno_nregs): New function.
9437 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9438 (TARGET_HARD_REGNO_NREGS): ...this hook.
9439 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9440 (CLASS_MAX_NREGS): Likewise.
9441 * doc/tm.texi: Regenerate.
9442 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9443 instead of HARD_REGNO_NREGS.
9444 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9445 HARD_REGNO_NREGS in the comment.
9446 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9447 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9448 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9449 Return an unsigned int.
9450 (TARGET_HARD_REGNO_NREGS): Redefine.
9451 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9452 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9453 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9454 (arc_hard_regno_nregs): New function.
9455 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9456 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9457 (arm_hard_regno_nregs): New function.
9458 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9459 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9460 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9461 (TARGET_HARD_REGNO_NREGS): Redefine.
9462 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9463 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9464 (HARD_REGNO_NREGS): Delete.
9465 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9466 (cr16_hard_regno_nregs): New function.
9467 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9468 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9469 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9470 (cris_hard_regno_nregs): New function.
9471 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9472 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9473 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9474 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9475 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9476 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9477 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9478 (frv_hard_regno_nregs): Make static. Take and return an
9480 (frv_class_max_nregs): Remove outdated copy of documentation.
9481 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9482 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9483 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9484 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9485 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9486 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9487 (TARGET_HARD_REGNO_NREGS): Redefine.
9488 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9489 (CLASS_MAX_NREGS): Update comment.
9490 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9491 (ia64_hard_regno_nregs): New function.
9492 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9493 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9494 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9495 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9496 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9498 (m32c_hard_regno_nregs): Likewise. Make static.
9499 (TARGET_HARD_REGNO_NREGS): Redefine.
9500 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9501 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9502 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9503 (m68k_hard_regno_nregs): New function.
9504 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9505 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9506 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9507 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9508 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9509 Take and return an unsigned int.
9510 (TARGET_HARD_REGNO_NREGS): Redefine.
9511 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9512 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9513 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9514 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9515 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9516 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9517 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9518 (msp430_hard_regno_nregs): Make static. Take and return an
9520 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9521 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9522 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9523 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9524 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9525 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9526 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9527 (TARGET_HARD_REGNO_NREGS): Redefine.
9528 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9529 (PA_HARD_REGNO_NREGS): ...this.
9530 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9531 (PA_HARD_REGNO_NREGS): ...this.
9532 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9533 (pa_hard_regno_nregs): New function.
9534 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9535 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9536 (pdp11_hard_regno_nregs): New function.
9537 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9538 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9539 (rs6000_hard_regno_nregs_hook): New function.
9540 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9541 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9542 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9543 Take and return an unsigned int. Move earlier in file.
9544 (TARGET_HARD_REGNO_NREGS): Redefine.
9545 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9546 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9547 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9548 (rl78_hard_regno_nregs): Make static. Take and return an
9550 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9551 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9552 (rs6000_hard_regno_nregs_hook): New function.
9553 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9554 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9555 (TARGET_HARD_REGNO_NREGS): Redefine.
9556 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9557 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9558 instead of HARD_REGNO_NREGS.
9559 (s390_hard_regno_nregs): New function.
9560 (s390_hard_regno_mode_ok): Add comment from s390.h.
9561 (TARGET_HARD_REGNO_NREGS): Redefine.
9562 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9563 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9564 (sh_hard_regno_nregs): New function.
9565 (sh_pass_in_reg_p): Use it.
9566 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9567 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9568 (sparc_hard_regno_nregs): New function.
9569 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9570 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9571 (spu_function_arg_advance): Use it, supplying a valid register number.
9572 (TARGET_HARD_REGNO_NREGS): Redefine.
9573 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9574 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9575 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9576 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9577 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9578 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9579 (CLASS_MAX_NREGS): Remove copy of old documentation.
9580 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9581 (visium_hard_regno_nregs): New function.
9582 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9583 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9584 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9585 xtensa_hard_regno_nregs): New function.
9586 * system.h (HARD_REGNO_NREGS): Poison.
9588 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9590 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9591 hard_regno_nregs instead of HARD_REGNO_NREGS.
9592 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9593 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9594 (c6x_expand_epilogue): Likewise.
9595 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9596 (frv_read_iacc_argument): Likewise.
9597 * config/sh/sh.c: Include regs.h.
9598 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9599 (regs_used): Likewise.
9600 (output_stack_adjust): Likewise.
9601 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9602 * expmed.c: Include regs.h.
9603 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9604 * ree.c: Include regs.h.
9605 (combine_reaching_defs): Use hard_regno_nregs instead of
9607 (add_removable_extension): Likewise.
9609 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9611 * regs.h (hard_regno_nregs): Turn into a function.
9612 (end_hard_regno): Update accordingly.
9613 * caller-save.c (setup_save_areas): Likewise.
9614 (save_call_clobbered_regs): Likewise.
9615 (replace_reg_with_saved_mem): Likewise.
9616 (insert_restore): Likewise.
9617 (insert_save): Likewise.
9618 * combine.c (can_change_dest_mode): Likewise.
9619 (move_deaths): Likewise.
9620 (distribute_notes): Likewise.
9621 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9622 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9623 (rs6000_split_multireg_move): Likewise.
9624 (rs6000_register_move_cost): Likewise.
9625 (rs6000_memory_move_cost): Likewise.
9626 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9627 (rs6000_split_multireg_move): Likewise.
9628 (rs6000_register_move_cost): Likewise.
9629 (rs6000_memory_move_cost): Likewise.
9630 * cselib.c (cselib_reset_table): Likewise.
9631 (cselib_lookup_1): Likewise.
9632 * emit-rtl.c (set_mode_and_regno): Likewise.
9633 * function.c (aggregate_value_p): Likewise.
9634 * ira-color.c (setup_profitable_hard_regs): Likewise.
9635 (check_hard_reg_p): Likewise.
9636 (calculate_saved_nregs): Likewise.
9637 (assign_hard_reg): Likewise.
9638 (improve_allocation): Likewise.
9639 (calculate_spill_cost): Likewise.
9640 * ira-emit.c (modify_move_list): Likewise.
9641 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9642 (ira_hard_reg_in_set_p): Likewise.
9643 * ira.c (setup_reg_mode_hard_regset): Likewise.
9644 (clarify_prohibited_class_mode_regs): Likewise.
9645 (check_allocation): Likewise.
9646 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9647 (lra_setup_reg_renumber): Likewise.
9648 (setup_try_hard_regno_pseudos): Likewise.
9649 (spill_for): Likewise.
9650 (assign_hard_regno): Likewise.
9651 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9652 * lra-constraints.c (in_class_p): Likewise.
9653 (lra_constraint_offset): Likewise.
9654 (simplify_operand_subreg): Likewise.
9655 (lra_constraints): Likewise.
9656 (split_reg): Likewise.
9657 (split_if_necessary): Likewise.
9658 (invariant_p): Likewise.
9659 (inherit_in_ebb): Likewise.
9660 * lra-lives.c (process_bb_lives): Likewise.
9661 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9662 (get_hard_regs): Likewise.
9663 (do_remat): Likewise.
9664 * lra-spills.c (assign_spill_hard_regs): Likewise.
9665 * mode-switching.c (create_pre_exit): Likewise.
9666 * postreload.c (reload_combine_recognize_pattern): Likewise.
9667 * recog.c (peep2_find_free_register): Likewise.
9668 * regcprop.c (kill_value_regno): Likewise.
9669 (set_value_regno): Likewise.
9670 (copy_value): Likewise.
9671 (maybe_mode_change): Likewise.
9672 (find_oldest_value_reg): Likewise.
9673 (copyprop_hardreg_forward_1): Likewise.
9674 * regrename.c (check_new_reg_p): Likewise.
9675 (regrename_do_replace): Likewise.
9676 * reload.c (push_reload): Likewise.
9677 (combine_reloads): Likewise.
9678 (find_dummy_reload): Likewise.
9679 (operands_match_p): Likewise.
9680 (find_reloads): Likewise.
9681 (find_equiv_reg): Likewise.
9682 (reload_adjust_reg_for_mode): Likewise.
9683 * reload1.c (count_pseudo): Likewise.
9684 (count_spilled_pseudo): Likewise.
9685 (find_reg): Likewise.
9686 (clear_reload_reg_in_use): Likewise.
9687 (free_for_value_p): Likewise.
9688 (allocate_reload_reg): Likewise.
9689 (choose_reload_regs): Likewise.
9690 (reload_adjust_reg_for_temp): Likewise.
9691 (emit_reload_insns): Likewise.
9692 (delete_output_reload): Likewise.
9693 * rtlanal.c (subreg_get_info): Likewise.
9694 * sched-deps.c (sched_analyze_reg): Likewise.
9695 * sel-sched.c (init_regs_for_mode): Likewise.
9696 (mark_unavailable_hard_regs): Likewise.
9697 (choose_best_reg_1): Likewise.
9698 (verify_target_availability): Likewise.
9699 * valtrack.c (dead_debug_insert_temp): Likewise.
9700 * var-tracking.c (track_loc_p): Likewise.
9701 (emit_note_insn_var_location): Likewise.
9702 * varasm.c (make_decl_rtl): Likewise.
9703 * reginfo.c (choose_hard_reg_mode): Likewise.
9704 (init_reg_modes_target): Refer directly to
9705 this_target_regs->x_hard_regno_nregs.
9707 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9709 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9710 instead of hard_regno_nregs.
9712 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9714 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9715 end_hard_regno instead of hard_regno_nregs.
9716 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9717 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9718 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9719 * ira-color.c (improve_allocation): Likewise.
9720 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9721 * lra-lives.c (mark_regno_live): Likewise.
9722 (mark_regno_dead): Likewise.
9723 * lra-remat.c (operand_to_remat): Likewise.
9724 * lra.c (collect_non_operand_hard_regs): Likewise.
9725 * postreload.c (reload_combine_note_store): Likewise.
9726 (move2add_valid_value_p): Likewise.
9727 * reload.c (regno_clobbered_p): Likewise.
9729 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9731 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9733 * config/v850/v850.c (v850_reorg): Likewise.
9734 * reload.c (refers_to_regno_for_reload_p): Likewise.
9735 (find_equiv_reg): Likewise.
9736 * reload1.c (reload_reg_reaches_end_p): Likewise.
9738 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9740 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9742 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9743 * config/arm/arm.c (output_move_neon): Likewise.
9744 (arm_attr_length_move_neon): Likewise.
9745 (neon_split_vcombine): Likewise.
9746 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9747 (c6x_mark_reg_written): Likewise.
9748 (c6x_dwarf_register_span): Likewise.
9749 * config/i386/i386.c (ix86_save_reg): Likewise.
9750 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9751 (rws_access_reg): Likewise.
9752 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9753 * mode-switching.c (create_pre_exit): Likewise.
9754 * ree.c (combine_reaching_defs): Likewise.
9755 (add_removable_extension): Likewise.
9756 * regcprop.c (find_oldest_value_reg): Likewise.
9757 (copyprop_hardreg_forward_1): Likewise.
9758 * reload.c (reload_inner_reg_of_subreg): Likewise.
9759 (push_reload): Likewise.
9760 (combine_reloads): Likewise.
9761 (find_dummy_reload): Likewise.
9762 (reload_adjust_reg_for_mode): Likewise.
9763 * reload1.c (find_reload_regs): Likewise.
9764 (forget_old_reloads_1): Likewise.
9765 (reload_reg_free_for_value_p): Likewise.
9766 (reload_adjust_reg_for_temp): Likewise.
9767 (emit_reload_insns): Likewise.
9768 (delete_output_reload): Likewise.
9769 * sel-sched.c (choose_best_reg_1): Likewise.
9770 (choose_best_pseudo_reg): Likewise.
9772 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9773 Alan Hayward <alan.hayward@arm.com>
9774 David Sherwood <david.sherwood@arm.com>
9776 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9777 * target.def (slow_unaligned_access): New hook.
9778 * targhooks.h (default_slow_unaligned_access): Declare.
9779 * targhooks.c (default_slow_unaligned_access): New function.
9780 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9781 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9782 * doc/tm.texi: Regenerate.
9783 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9784 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9785 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9787 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9788 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9790 (rs6000_slow_unaligned_access): New function.
9791 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9792 (expand_block_compare): Likewise.
9793 (expand_strn_compare): Likewise.
9794 (rs6000_rtx_costs): Likewise.
9795 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9796 (riscv_slow_unaligned_access): Likewise.
9797 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9798 (riscv_slow_unaligned_access_p): ...this and make static.
9799 (riscv_option_override): Update accordingly.
9800 (riscv_slow_unaligned_access): New function.
9801 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9802 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9803 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9804 (rs6000_slow_unaligned_access): New function.
9805 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9806 (rs6000_rtx_costs): Likewise.
9807 * config/rs6000/rs6000-string.c (expand_block_compare)
9808 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9809 of SLOW_UNALIGNED_ACCESS.
9810 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9811 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9812 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9813 of SLOW_UNALIGNED_ACCESS.
9814 * expmed.c (simple_mem_bitfield_p): Likewise.
9815 * expr.c (alignment_for_piecewise_move): Likewise.
9816 (emit_group_load_1): Likewise.
9817 (emit_group_store): Likewise.
9818 (copy_blkmode_from_reg): Likewise.
9819 (emit_push_insn): Likewise.
9820 (expand_assignment): Likewise.
9821 (store_field): Likewise.
9822 (expand_expr_real_1): Likewise.
9823 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9824 * lra-constraints.c (simplify_operand_subreg): Likewise.
9825 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9826 * gimple-ssa-store-merging.c: Likewise in block comment at start
9828 * tree-ssa-strlen.c: Include target.h.
9829 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9830 of SLOW_UNALIGNED_ACCESS.
9831 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9833 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9835 PR rtl-optimization/82185
9836 * expmed.c (emit_store_flag_int): Only test tem if it has been
9839 2017-09-12 Richard Biener <rguenther@suse.de>
9842 * match.pd ((FTYPE) N CMP CST): Fix typo.
9844 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9846 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9848 (mips_near_type_p): Add 'short_call' attribute as a synonym
9850 * doc/extend.texi (short_call): Document new function attribute.
9852 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9855 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9856 assertion check that in the condition.
9857 (get_atomic_generic_size): Likewise. Before testing if parameter
9858 has pointer type, if it has array type, call for C++
9859 default_conversion to perform array-to-pointer conversion.
9861 2017-09-12 Richard Biener <rguenther@suse.de>
9863 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9864 for operations we cannot scalarize.
9866 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9868 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9869 vectors heap vectors. Clean up comments.
9870 Make visited_bbs a reference.
9871 (profitable_jump_thread_path): Make GC
9872 vectors heap vectors. Clean up comments.
9874 (convert_and_register_jump_thread_path): Make GC vectors heap
9876 (check_subpath_and_update_thread_path): Same. Clean up comments.
9877 Make visited_bbs a reference.
9878 (handle_phi): Abstract common code to to
9879 register_jump_thread_path_if_profitable.
9880 Rename VAR_BB to DEF_BB.
9882 Make GC vectors heap vectors.
9883 Make visited_bbs a reference.
9884 (handle_assignment): Same.
9885 (register_jump_thread_path_if_profitable): New.
9886 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9888 Make GC vectors heap vectors. Clean up comments.
9889 Make visited_bbs a reference.
9890 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9891 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9894 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9897 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9898 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9900 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9902 Revert r251800 and r251799.
9904 2017-09-11 Martin Jambor <mjambor@suse.cz>
9907 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9908 arguments in advance.
9909 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9910 use it to find predecessor edges.
9911 (naive_outof_ssa): Collect vector of predecessors.
9913 2017-09-08 Jason Merrill <jason@redhat.com>
9915 PR c++/70029 - ICE with ref-qualifier and -flto
9916 * langhooks.h (struct lang_hooks_for_types): Add
9917 copy_lang_qualifiers.
9918 * attribs.c (build_type_attribute_qual_variant): Use it.
9919 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9921 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9922 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9924 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9927 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9928 (*mulsi3_sp64): New instruction.
9929 (mulsi3): New expander.
9931 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9933 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9935 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9937 * sancov.c: Include memmodel.h.
9939 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9942 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9945 2017-09-07 Carl Love <cel@us.ibm.com>
9947 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9948 the sldi instruction.
9950 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9952 * sancov.c: Include tm_p.h.
9954 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9957 * output.h (switch_to_other_text_partition): New declaration.
9958 * varasm.c (switch_to_other_text_partition): New function.
9959 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9960 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9961 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9962 to the other text partition before emitting LCL label and switch back
9963 after emitting the word after it.
9965 2017-09-07 Richard Biener <rguenther@suse.de>
9967 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9968 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9969 critical edges here, after loop init.
9970 (pass_data_pre): Remove PROP_no_crit_edges flags.
9971 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9972 for valueization of call args to avoid leaking VN_TOP.
9973 (visit_use): Assert we do not visit default defs.
9974 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9975 Use error_mark_node to more easily detect leaking VN_TOP.
9976 All default-defs are varying, not VN_TOP. Mark them visited.
9977 (run_scc_vn): Make code match comment.
9979 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9981 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9982 OPTION_MASK_FLOAT128_KEYWORD.
9983 (POWERPC_MASKS): Likewise.
9984 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9985 support for the -mfloat128-type option, and make -mfloat128
9986 default on PowerPC Linux systems. Define or undefine
9987 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9988 Define __float128 to be __ieee128 if IEEE 128-bit support is
9989 enabled, or undefine it.
9990 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9991 Delete defining __FLOAT128_TYPE__.
9992 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9993 -mfloat128-type option and make -mfloat128 default on PowerPC
9995 (TARGET_FLOAT128_TYPE): Likewise.
9996 (-mfloat128-type): Likewise.
9997 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9998 Delete the -mfloat128-type option and make -mfloat128 default on
9999 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10000 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10001 128-bit floating point is enabled. Change tests from using
10002 -mfloat128-type to -mfloat128.
10003 (rs6000_mangle_type): Use the correct mangling for the __float128
10004 type even if normal long double is restricted to 64-bits.
10005 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10007 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10008 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10009 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10011 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10012 documentation for -mfloat128.
10014 2017-09-06 Olivier Hainque <hainque@adacore.com>
10016 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10018 2017-09-06 Wish Wu <wishwu007@gmail.com>
10019 Jakub Jelinek <jakub@redhat.com>
10021 * asan.c (initialize_sanitizer_builtins): Add
10022 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10023 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10024 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10025 BT_FN_VOID_UINT64_PTR variables.
10026 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10027 (BT_FN_VOID_UINT16_UINT16): Likewise.
10028 (BT_FN_VOID_UINT32_UINT32): Likewise.
10029 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10030 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10031 (BT_FN_VOID_UINT64_PTR): Likewise.
10032 * common.opt (flag_sanitize_coverage): New variable.
10033 (fsanitize-coverage=trace-pc): Remove.
10034 (fsanitize-coverage=): Add.
10035 * flag-types.h (enum sanitize_coverage_code): New enum.
10036 * fold-const.c (fold_range_test): Disable non-short-circuit
10037 optimization if flag_sanitize_coverage.
10038 (fold_truth_andor): Likewise.
10039 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10040 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10041 (coverage_sanitizer_opts): New array.
10042 (get_closest_sanitizer_option): Add OPTS argument, handle also
10043 OPT_fsanitize_coverage_.
10044 (parse_sanitizer_options): Adjusted to also handle
10045 OPT_fsanitize_coverage_.
10046 (common_handle_option): Add OPT_fsanitize_coverage_.
10047 * sancov.c (instrument_comparison, instrument_switch): New function.
10048 (sancov_pass): Add trace-cmp support.
10049 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10050 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10051 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10052 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10053 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10054 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10055 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10056 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10057 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10058 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10060 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10062 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10063 error. Only quit immediately if parsing is complete.
10064 (BEGIN): Initialize fatal_err and parse_done.
10065 (begin fpu, end fpu): Check number of arguments.
10066 (begin arch, end arch): Likewise.
10067 (begin cpu, end cpu): Likewise.
10068 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10069 (optalias): Likewise.
10071 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10073 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10074 * config/arm/arm-isa.h: Delete. Move definitions to ...
10075 * arm-cpus.in: ... here. Use new feature and fgroup values.
10076 * config/arm/arm.c (arm_option_override): Use lower case for feature
10078 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10079 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10080 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10082 (print_isa_bits_for): New function.
10083 (gen_isa): New function.
10084 (gen_comm_data): Use print_isa_bits_for.
10085 (define feature): New keyword.
10086 (define fgroup): New keyword.
10087 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10088 (arm-isa.h): Add rule to generate file.
10089 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10090 case for feature bit names.
10092 2017-09-06 Richard Biener <rguenther@suse.de>
10094 * tree-ssa-pre.c (NECESSARY): Remove.
10095 (create_expression_by_pieces): Do not touch pass-local flags.
10096 (insert_into_preds_of_block): Likewise.
10097 (do_pre_regular_insertion): Likewise.
10098 (eliminate_insert): Likewise.
10099 (eliminate_dom_walker::before_dom_children): Likewise.
10100 (fini_eliminate): Do not look at inserted_exprs.
10101 (mark_operand_necessary): Remove.
10102 (remove_dead_inserted_code): Replace with simple work-list
10103 algorithm based on inserted_exprs and SSA uses.
10104 (pass_pre::execute): Re-order fini_eliminate and
10105 remove_dead_inserted_code.
10107 2017-09-06 Olivier Hainque <hainque@adacore.com>
10109 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10110 for VxWorks 7. Adjust surrounding comments.
10112 2017-09-06 Richard Biener <rguenther@suse.de>
10114 * gimple-ssa-strength-reduction.c
10115 (find_candidates_dom_walker::before_dom_children): Also allow
10118 2017-09-06 Richard Biener <rguenther@suse.de>
10120 PR tree-optimization/82108
10121 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10122 for gap in the non-permutation SLP case.
10124 2017-09-06 Martin Jambor <mjambor@suse.cz>
10126 PR tree-optimization/82078
10127 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10128 add_access_to_work_queue...
10129 (build_accesses_from_assign): ...here.
10130 (propagate_all_subaccesses): Make sure racc is the group
10131 representative, if there is one.
10133 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10135 PR middle-end/82095
10136 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10139 2017-09-06 Richard Biener <rguenther@suse.de>
10141 * gimple-ssa-strength-reduction.c
10142 (find_candidates_dom_walker::before_doom_children): Use a
10143 type and not a mode check.
10145 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10148 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10150 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10152 2017-09-05 Jeff Law <law@redhat.com>
10154 PR tree-optimization/64910
10155 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10156 cases where we have 3 or more operands.
10158 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10160 PR middle-end/81768
10161 * omp-low.c (lower_omp_for): Recompute tree invariant if
10162 gimple_omp_for_initial/final is ADDR_EXPR.
10164 PR middle-end/81768
10165 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10166 into gimple val before gimplification fo the COND_EXPR.
10168 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10170 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10171 REGION_COPY argument.
10172 (thread_through_all_blocks): Remove unused argument to
10173 duplicate_thread_path.
10175 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10176 Alan Hayward <alan.hayward@arm.com>
10177 David Sherwood <david.sherwood@arm.com>
10179 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10180 Take a scalar_mode rather than a machine_mode.
10181 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10182 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10183 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10184 (aarch64_gen_adjusted_ldpstp): Likewise.
10185 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10187 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10188 Alan Hayward <alan.hayward@arm.com>
10189 David Sherwood <david.sherwood@arm.com>
10191 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10192 Take a scalar_int_mode instead of a machine_mode.
10193 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10194 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10195 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10196 (aarch64_simd_attr_length_rglist): Delete.
10197 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10198 a scalar_int_mode instead of a machine_mode.
10199 (aarch64_add_offset): Likewise.
10200 (aarch64_internal_mov_immediate): Likewise
10201 (aarch64_add_constant_internal): Likewise.
10202 (aarch64_add_constant): Likewise.
10203 (aarch64_movw_imm): Likewise.
10204 (aarch64_rtx_arith_op_extract_p): Likewise.
10205 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10206 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10207 Remove assert that the mode isn't a vector.
10208 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10209 (aarch64_expand_mov_immediate): Update calls after above changes.
10210 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10211 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10212 (aarch64_move_imm): Likewise.
10213 (aarch64_can_const_movi_rtx_p): Likewise.
10214 (aarch64_strip_extend): Likewise.
10215 (aarch64_extr_rtx_p): Likewise.
10216 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10217 a CONST_INT when the mode parameter is VOIDmode.
10218 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10220 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10222 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10223 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10224 (bitwise_type_for_mode): Update accordingly.
10226 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10228 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10229 * stor-layout.c (mode_for_size_tree): Likewise.
10230 (mode_for_array): Update accordingly.
10231 (layout_decl): Likewise.
10232 (compute_record_mode): Likewise. Only set the mode once.
10234 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10236 * target.def (get_mask_mode): Change return type to opt_mode.
10238 * doc/tm.texi: Regenerate.
10239 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10240 * targhooks.c (default_get_mask_mode): Likewise.
10241 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10242 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10243 targetm.get_mask_mode.
10244 * tree.c (build_truth_vector_type): Likewise.
10246 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10248 * machmode.h (mode_for_vector): Return an opt_mode.
10249 * stor-layout.c (mode_for_vector): Likewise.
10250 (mode_for_int_vector): Update accordingly.
10251 (layout_type): Likewise.
10252 * config/i386/i386.c (emit_memmov): Likewise.
10253 (ix86_expand_set_or_movmem): Likewise.
10254 (ix86_expand_vector_init): Likewise.
10255 (ix86_get_mask_mode): Likewise.
10256 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10258 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10259 * expmed.c (extract_bit_field_1): Likewise.
10260 * expr.c (expand_expr_real_2): Likewise.
10261 * optabs-query.c (can_vec_perm_p): Likewise.
10262 (can_vec_mask_load_store_p): Likewise.
10263 * optabs.c (expand_vec_perm): Likewise.
10264 * targhooks.c (default_get_mask_mode): Likewise.
10265 * tree-vect-stmts.c (vectorizable_store): Likewise.
10266 (vectorizable_load): Likewise.
10267 (get_vectype_for_scalar_type_and_size): Likewise.
10269 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10271 * machmode.h (mode_for_int_vector): New function.
10272 * stor-layout.c (mode_for_int_vector): Likewise.
10273 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10274 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10275 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10276 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10277 (s390_expand_vcond): Likewise.
10279 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10281 * machmode.h (opt_machine_mode): New type.
10282 (opt_mode<T>): Allow construction from anything that can be
10284 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10285 (mode_for_size): Return an opt_machine_mode.
10286 * stor-layout.c (mode_for_size): Likewise.
10287 (mode_for_size_tree): Update call accordingly.
10288 (bitwise_mode_for_mode): Likewise.
10289 (make_fract_type): Likewise.
10290 (make_accum_type): Likewise.
10291 * caller-save.c (replace_reg_with_saved_mem): Update call
10293 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10294 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10295 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10296 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10297 * expmed.c (extract_bit_field_1): Likewise.
10298 * reload.c (get_secondary_mem): Likewise.
10299 * varasm.c (assemble_integer): Likewise.
10300 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10303 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10305 * machmode.h (decimal_float_mode_for_size): New function.
10306 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10307 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10308 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10309 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10310 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10311 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10313 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10315 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10316 (get_builtin_sync_mode): Likewise.
10317 (expand_ifn_atomic_compare_exchange): Likewise.
10318 (expand_builtin_atomic_clear): Likewise.
10319 (expand_builtin_atomic_test_and_set): Likewise.
10320 (fold_builtin_atomic_always_lock_free): Likewise.
10321 * calls.c (compute_argument_addresses): Likewise.
10322 (emit_library_call_value_1): Likewise.
10323 (store_one_arg): Likewise.
10324 * combine.c (combine_instructions): Likewise.
10325 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10326 * config/arm/arm.c (arm_function_value): Likewise.
10327 (aapcs_allocate_return_reg): Likewise.
10328 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10329 * config/i386/i386.c (construct_container): Likewise.
10330 (ix86_gimplify_va_arg): Likewise.
10331 (ix86_expand_sse_cmp): Likewise.
10332 (emit_memmov): Likewise.
10333 (emit_memset): Likewise.
10334 (expand_small_movmem_or_setmem): Likewise.
10335 (ix86_expand_pextr): Likewise.
10336 (ix86_expand_pinsr): Likewise.
10337 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10338 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10340 * config/mips/mips.c (mips_function_value_1) Likewise.
10341 (mips_block_move_straight): Likewise.
10342 (mips_expand_ins_as_unaligned_store): Likewise.
10343 * config/powerpcspe/powerpcspe.c
10344 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10345 (rs6000_darwin64_record_arg_flush): Likewise.
10346 * config/rs6000/rs6000.c
10347 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10348 (rs6000_darwin64_record_arg_flush): Likewise.
10349 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10350 (sparc_function_value_1): Likewise.
10351 * config/spu/spu.c (adjust_operand): Likewise.
10352 (spu_emit_branch_or_set): Likewise.
10353 (arith_immediate_p): Likewise.
10354 * emit-rtl.c (gen_lowpart_common): Likewise.
10355 * expr.c (expand_expr_real_1): Likewise.
10356 * function.c (assign_parm_setup_block): Likewise.
10357 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10358 * reload1.c (alter_reg): Likewise.
10359 * stor-layout.c (mode_for_vector): Likewise.
10360 (layout_type): Likewise.
10362 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10364 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10365 (spu_convert_move): Likewise.
10366 * lower-subreg.c (resolve_simple_move): Likewise.
10368 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10371 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10372 define_insn to a define_expand.
10373 (altivec_vsum2sws_direct): New define_insn.
10374 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10376 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10378 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10381 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10386 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10387 realign stack if stack alignment needed is less than incoming
10390 2017-09-05 Marek Polacek <polacek@redhat.com>
10393 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10396 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10398 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10400 2017-09-05 Richard Biener <rguenther@suse.de>
10402 PR tree-optimization/82084
10403 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10405 2017-09-05 Richard Biener <rguenther@suse.de>
10407 PR tree-optimization/82102
10408 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10410 2017-09-05 Martin Liska <mliska@suse.cz>
10412 PR tree-optimization/82032
10413 * tree-cfg.c (generate_range_test): New function.
10414 * tree-cfg.h (generate_range_test): Declared here.
10415 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10416 (cleanup_control_expr_graph): Use it.
10417 * tree-switch-conversion.c (try_switch_expansion): Remove
10419 (emit_case_nodes): Use generate_range_test.
10421 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10424 * config/i386/i386.md (*<btsc><mode>_mask): Add
10425 TARGET_USE_BT to insn constraint.
10426 (*btr<mode>_mask): Ditto.
10428 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10430 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10431 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10433 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10436 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10437 TARGET_NEON and TARGET_IWMMXT.
10438 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10439 TARGET_NEON and TARGET_IWMMXT.
10440 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10442 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10444 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10445 (ix86_rewrite_tls_address): Ditto.
10446 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10447 (ix86_rewrite_tls_address_1): Ditto.
10448 (ix86_rewrite_tls_address): Ditto.
10449 * config/i386/predicates.md (tls_address_pattern): New predicate.
10450 * config/i386/i386.md (TLS address splitter): New splitter.
10452 2017-09-04 Richard Biener <rguenther@suse.de>
10454 PR tree-optimization/82084
10455 * fold-const.h (can_native_encode_string_p): Declare.
10456 * fold-const.c (can_native_encode_string_p): Factor out from ...
10457 (native_encode_string): ... here.
10458 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10459 vectorizing stores from constants we later cannot handle.
10461 2017-09-04 Marek Polacek <polacek@redhat.com>
10464 * doc/invoke.texi: Update -Wtautological-compare documentation.
10466 2017-09-04 Jeff Law <law@redhat.com>
10468 PR tree-optimization/64910
10469 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10470 swap the first and last operand if the last is a constant.
10472 2017-09-04 Marek Polacek <polacek@redhat.com>
10475 * convert.c (do_narrow): When sanitizing signed integer overflows,
10476 bail out for signed types.
10477 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10479 2017-09-04 Richard Biener <rguenther@suse.de>
10481 PR tree-optimization/82060
10482 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10483 Move devirtualization after stmt folding and before EH/AB/noreturn
10484 cleanup to get the stmt refs canonicalized. Use a bool instead
10485 of gimple_modified_p since that doesn't work for NOPs. Schedule
10486 NOPs generated by folding for removal.
10488 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10489 Alan Hayward <alan.hayward@arm.com>
10490 David Sherwood <david.sherwood@arm.com>
10492 * coretypes.h (pad_direction): New enum.
10493 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10494 (FUNCTION_ARG_PADDING): Likewise.
10495 * target.def (function_arg_padding): New hook.
10496 * targhooks.h (default_function_arg_padding): Declare.
10497 * targhooks.c (default_function_arg_padding): New function.
10498 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10499 (TARGET_FUNCTION_ARG_PADDING): ...this.
10500 * doc/tm.texi: Regenerate.
10501 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10502 instead of direction.
10503 (compute_argument_addresses): Likewise.
10504 (load_register_parameters): Likewise.
10505 (emit_library_call_value_1): Likewise.
10506 (store_one_arg): Use targetm.calls.function_arg_padding instead
10507 of FUNCTION_ARG_PADDING.
10508 (must_pass_in_stack_var_size_or_pad): Likewise.
10509 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10510 (emit_group_store): Likewise.
10511 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10512 instead of FUNCTION_ARG_PADDING.
10513 (emit_push_insn): Likewise, and propagate enum change throughout
10515 * function.h (direction): Delete.
10516 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10518 * function.c (assign_parm_find_stack_rtl): Likewise.
10519 (assign_parm_setup_block_p): Likewise.
10520 (assign_parm_setup_block): Likewise.
10521 (gimplify_parameters): Likewise.
10522 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10523 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10525 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10526 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10527 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10528 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10529 (aarch64_function_arg_padding): ...this new function.
10530 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10531 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10532 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10533 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10534 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10535 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10536 (arm_pad_arg_upward): Replace with...
10537 (arm_function_arg_padding): ...this new function.
10538 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10540 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10541 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10542 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10543 (ia64_hpux_function_arg_padding): Replace with...
10544 (ia64_function_arg_padding): ...this new function. Use pad_direction
10545 instead of direction. Check for TARGET_HPUX.
10546 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10547 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10548 (iq2000_function_arg_padding): New function.
10549 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10550 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10551 (mips_function_arg_padding): ...this new function.
10552 (mips_pad_reg_upward): Update accordingly.
10553 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10554 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10555 targetm.calls.function_arg_padding.
10556 (FUNCTION_ARG_PADDING): Delete.
10557 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10558 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10559 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10560 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10561 (nios2_block_reg_padding): Return pad_direction instead of direction.
10562 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10563 instead of direction.
10564 (nios2_function_arg_padding): Likewise. Make static.
10565 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10566 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10567 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10568 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10569 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10570 (pa_function_arg_padding): Make static. Return pad_direction instead
10572 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10573 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10574 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10575 instead of direction. Use targetm.calls.function_arg_padding.
10576 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10577 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10578 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10579 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10580 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10582 (function_arg_padding): Rename to...
10583 (rs6000_function_arg_padding): ...this. Make static. Return
10584 pad_direction instead of direction.
10585 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10586 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10587 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10588 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10589 instead of direction. Use targetm.calls.function_arg_padding.
10590 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10591 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10592 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10593 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10594 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10595 (function_arg_padding): Rename to...
10596 (rs6000_function_arg_padding): ...this. Make static. Return
10597 pad_direction instead of direction.
10598 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10599 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10600 * config/s390/s390.c (s390_function_arg_padding): New function.
10601 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10602 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10603 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10604 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10605 (function_arg_padding): Rename to...
10606 (sparc_function_arg_padding): ...this. Make static. Return
10607 pad_direction instead of direction.
10608 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10609 * config/spu/spu.c (spu_function_arg_padding): New function.
10610 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10611 * system.h (FUNCTION_ARG_PADDING): Poison.
10613 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10614 Alan Hayward <alan.hayward@arm.com>
10615 David Sherwood <david.sherwood@arm.com>
10617 * target.def (modes_tieable_p): New hook.
10618 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10619 (TARGET_MODES_TIEABLE_P): ...this.
10620 * doc/tm.texi.in: Regenerate.
10621 * hooks.h (hook_bool_mode_mode_true): Declare.
10622 * hooks.c (hook_bool_mode_mode_true): New function.
10623 * combine.c (subst): Use targetm.modes_tieable_p instead of
10625 * dse.c (find_shift_sequence): Likewise.
10626 * expmed.c (extract_low_bits): Likewise.
10627 * lower-subreg.c: Include target.h.
10628 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10630 * rtlanal.c (rtx_cost): Likewise.
10631 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10632 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10633 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10634 (TARGET_MODES_TIEABLE_P): Redefine.
10635 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10636 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10637 (TARGET_MODES_TIEABLE_P): Redefine.
10638 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10639 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10640 (arc_modes_tieable_p): New function.
10641 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10642 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10643 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10644 (arm_modes_tieable_p): Make static.
10645 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10646 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10647 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10648 (TARGET_MODES_TIEABLE_P): Redefine.
10649 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10650 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10651 (TARGET_MODES_TIEABLE_P): Redefine.
10652 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10653 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10654 (cr16_modes_tieable_p): New function.
10655 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10656 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10657 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10658 (TRULY_NOOP_TRUNCATION): Update comment.
10659 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10660 (TRULY_NOOP_TRUNCATION): Update comment.
10661 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10662 (frv_modes_tieable_p): New function.
10663 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10664 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10665 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10666 (TARGET_MODES_TIEABLE_P): Redefine.
10667 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10668 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10669 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10670 (TARGET_MODES_TIEABLE_P): Redefine.
10671 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10672 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10673 (ia64_modes_tieable_p): New function.
10674 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10675 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10676 (iq2000_modes_tieable_p): New function.
10677 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10678 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10679 (lm32_modes_tieable_p): New function.
10680 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10681 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10682 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10683 (TARGET_MODES_TIEABLE_P): Redefine.
10684 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10685 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10686 (m32r_modes_tieable_p): New function.
10687 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10688 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10689 (m68k_modes_tieable_p): New function.
10690 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10691 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10692 (mcore_modes_tieable_p): New function.
10693 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10694 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10696 (TARGET_MODES_TIEABLE_P): Redefine.
10697 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10698 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10699 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10700 (TARGET_MODES_TIEABLE_P): Redefine.
10701 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10702 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10703 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10704 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10705 (mn10300_modes_tieable_p): ...this and make static.
10706 (TARGET_MODES_TIEABLE_P): Redefine.
10707 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10708 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10709 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10710 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10711 (msp430_modes_tieable_p): Make static.
10712 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10713 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10714 (TARGET_MODES_TIEABLE_P): Redefine.
10715 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10716 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10717 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10718 (TARGET_MODES_TIEABLE_P): Redefine.
10719 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10720 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10721 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10722 (TARGET_MODES_TIEABLE_P): Redefine.
10723 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10724 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10725 (pdp11_modes_tieable_p): New function.
10726 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10727 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10728 (rs6000_modes_tieable_p): New function.
10729 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10730 * config/powerpcspe/powerpcspe.md: Update comment.
10731 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10732 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10733 (TARGET_MODES_TIEABLE_P): Redefine.
10734 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10735 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10736 (rl78_modes_tieable_p): New function.
10737 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10738 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10739 (rs6000_modes_tieable_p): New function.
10740 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10741 * config/rs6000/rs6000.md: Update comment.
10742 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10743 * config/rx/rx.c (rx_modes_tieable_p): New function.
10744 (TARGET_MODES_TIEABLE_P): Redefine.
10745 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10746 * config/s390/s390.c (s390_modes_tieable_p): New function.
10747 (TARGET_MODES_TIEABLE_P): Redefine.
10748 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10749 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10750 (sh_modes_tieable_p): New function.
10751 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10752 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10753 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10754 (sparc_modes_tieable_p): Make static.
10755 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10756 * config/spu/spu.c (spu_modes_tieable_p): New function.
10757 (TARGET_MODES_TIEABLE_P): Redefine.
10758 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10759 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10760 (TARGET_MODES_TIEABLE_P): Redefine.
10761 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10762 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10763 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10764 * config/v850/v850.c (v850_modes_tieable_p): New function.
10765 (TARGET_MODES_TIEABLE_P): Redefine.
10766 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10767 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10768 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10769 (visium_modes_tieable_p): New function.
10770 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10771 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10772 (xtensa_modes_tieable_p): New function.
10773 * system.h (MODES_TIEABLE_P): Poison.
10775 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10776 Alan Hayward <alan.hayward@arm.com>
10777 David Sherwood <david.sherwood@arm.com>
10779 * target.def (hard_regno_mode_ok): New hook.
10780 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10781 (TARGET_HARD_REGNO_MODE_OK): ...this.
10782 * doc/tm.texi.in: Regenerate.
10783 * hooks.h (hook_bool_uint_mode_true): Declare.
10784 * hooks.c (hook_bool_uint_mode_true): New function.
10785 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10786 HARD_REGNO_MODE_OK.
10787 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10788 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10789 instead of HARD_REGNO_MODE_OK.
10790 * caller-save.c: Include target.h.
10791 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10792 HARD_REGNO_MODE_OK.
10793 * combine.c (can_combine_p): Likewise.
10794 (combinable_i3pat): Likewise.
10795 (can_change_dest_mode): Likewise.
10796 * expr.c (init_expr_target): Likewise.
10797 (convert_move): Likewise.
10798 (convert_modes): Likewise.
10799 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10800 (setup_prohibited_mode_move_regs): Likewise.
10801 * ira.h (target_ira): Likewise.
10802 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10803 * lra-constraints.c (process_alt_operands): Likewise.
10804 (split_reg): Likewise.
10805 * recog.c (peep2_find_free_register): Likewise.
10806 * ree.c (combine_reaching_defs): Likewise.
10807 * regcprop.c (maybe_mode_change): Likewise.
10808 * reginfo.c (init_reg_sets_1): Likewise.
10809 (choose_hard_reg_mode): Likewise.
10810 (simplifiable_subregs): Likewise.
10811 * regrename.c (check_new_reg_p): Likewise.
10812 * reload.c (find_valid_class): Likewise.
10813 (find_valid_class_1): Likewise.
10814 (reload_inner_reg_of_subreg): Likewise.
10815 (push_reload): Likewise.
10816 (combine_reloads): Likewise.
10817 (find_dummy_reload): Likewise.
10818 (find_reloads): Likewise.
10819 * reload1.c (find_reg): Likewise.
10820 (set_reload_reg): Likewise.
10821 (allocate_reload_reg): Likewise.
10822 (choose_reload_regs): Likewise.
10823 (reload_adjust_reg_for_temp): Likewise.
10824 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10825 (simplify_subreg_regno): Likewise.
10826 * sel-sched.c (init_regs_for_mode): Likewise.
10827 * varasm.c (make_decl_rtl): Likewise.
10828 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10829 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10830 HARD_REGNO_MODE_OK.
10831 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10832 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10833 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10834 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10835 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10837 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10838 (arc_mode_class): Delete.
10839 (HARD_REGNO_MODE_OK): Delete.
10840 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10841 (arc_hard_regno_mode_ok): Rename old array to...
10842 (arc_hard_regno_mode_ok_modes): ...this.
10843 (arc_conditional_register_usage): Update accordingly.
10844 (arc_mode_class): Make static.
10845 (arc_hard_regno_mode_ok): New function.
10846 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10847 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10848 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10849 (arm_hard_regno_mode_ok): Make static.
10850 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10851 HARD_REGNO_MODE_OK.
10852 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10853 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10854 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10856 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10857 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10858 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10859 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10860 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10861 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10862 * config/bfin/predicates.md (valid_reg_operand): Use
10863 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10864 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10865 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10866 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10867 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10868 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10869 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10870 (cr16_hard_regno_mode_ok): Make static and return a bool.
10871 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10872 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10873 (cris_hard_regno_mode_ok): New function.
10874 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10875 (epiphany_mode_class): Delete.
10876 (HARD_REGNO_MODE_OK): Delete.
10877 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10878 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10879 (hard_regno_mode_ok): Rename to...
10880 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10881 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10882 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10883 HARD_REGNO_MODE_OK.
10884 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10885 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10886 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10887 (frv_hard_regno_mode_ok): Make static and return a bool.
10888 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10889 HARD_REGNO_MODE_OK.
10890 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10891 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10892 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10893 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10895 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10896 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10897 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10898 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10901 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10902 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10903 (ia64_hard_regno_mode_ok): New function.
10904 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10905 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10906 (iq2000_hard_regno_mode_ok): New function.
10907 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10908 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10909 (lm32_hard_regno_mode_ok): New function.
10910 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10911 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10912 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10913 instead of HARD_REGNO_MODE_OK.
10914 (m32c_hard_regno_ok): Rename to...
10915 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10916 (m32c_cannot_change_mode_class): Update accordingly.
10917 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10918 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10919 (m32r_mode_class): Delete.
10920 (HARD_REGNO_MODE_OK): Delete.
10921 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10922 (m32r_hard_regno_mode_ok): Rename to...
10923 (m32r_hard_regno_modes): ...this.
10924 (m32r_mode_class): Make static.
10925 (m32r_hard_regno_mode_ok): New function.
10926 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10927 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10928 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10929 (m68k_hard_regno_mode_ok): Make static.
10930 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10931 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10932 (mcore_hard_regno_mode_ok): New function.
10933 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10934 (HARD_REGNO_MODE_OK): Delete.
10935 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10937 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10938 (microblaze_hard_regno_mode_ok): New function.
10939 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10940 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10941 (mips_hard_regno_mode_ok): Delete.
10942 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10943 (mips_hard_regno_mode_ok_p): ...this and make static.
10944 (mips_hard_regno_mode_ok_p): Rename to...
10945 (mips_hard_regno_mode_ok_uncached): ...this.
10946 (mips_hard_regno_mode_ok): New function.
10947 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10948 of HARD_REGNO_MODE_OK.
10949 (mips_option_override): Update after above name changes.
10950 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10951 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10952 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10953 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10954 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10955 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10956 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10957 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10958 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10959 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10960 (msp430_hard_regno_mode_ok): Make static and return a bool.
10961 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10962 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10963 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10965 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10966 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10967 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10968 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10969 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10970 (PA_HARD_REGNO_MODE_OK): ...this
10971 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10972 (PA_HARD_REGNO_MODE_OK): ...this.
10973 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10974 (pa_hard_regno_mode_ok): New function.
10975 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10976 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10977 (pdp11_hard_regno_mode_ok): New function.
10978 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10979 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10981 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10983 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10984 (rs6000_hard_regno_mode_ok): Rename to...
10985 (rs6000_hard_regno_mode_ok_uncached): ...this.
10986 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10987 (rs6000_hard_regno_mode_ok): New function.
10988 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10989 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10990 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10991 (riscv_hard_regno_mode_ok): ...this and make static.
10992 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10993 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10994 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10995 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10996 (rl78_hard_regno_mode_ok): Make static and return bool.
10997 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10998 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11000 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11002 (rs6000_hard_regno_mode_ok): Rename to...
11003 (rs6000_hard_regno_mode_ok_uncached): ...this.
11004 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11005 (rs6000_hard_regno_mode_ok): New function.
11006 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11007 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11008 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11009 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11010 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11011 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11012 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11013 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11014 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11015 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11016 (sh_hard_regno_mode_ok): Make static.
11017 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11018 instead of HARD_REGNO_MODE_OK.
11019 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11020 (sparc_mode_class): Delete.
11021 (HARD_REGNO_MODE_OK): Delete.
11022 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11023 (hard_regno_mode_classes): Make static.
11024 (sparc_mode_class): Likewise.
11025 (sparc_hard_regno_mode_ok): New function.
11026 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11027 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11028 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11030 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11031 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11032 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11033 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11034 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11035 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11036 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11037 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11038 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11039 (visium_hard_regno_mode_ok): New function.
11040 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11041 instead of HARD_REGNO_MODE_OK.
11042 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11043 (HARD_REGNO_MODE_OK): Delete.
11044 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11045 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11046 (xtensa_option_override): Update accordingly.
11047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11048 (xtensa_hard_regno_mode_ok): New function.
11049 * system.h (HARD_REGNO_MODE_OK): Poison.
11051 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11052 Alan Hayward <alan.hayward@arm.com>
11053 David Sherwood <david.sherwood@arm.com>
11055 * target.def (hard_regno_call_part_clobbered): New hook.
11056 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11057 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11058 * doc/tm.texi: Regenerate.
11059 * hooks.h (hook_bool_uint_mode_false): Declare.
11060 * hooks.c (hook_bool_uint_mode_false): New function.
11061 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11062 * cselib.c (cselib_process_insn): Use
11063 targetm.hard_regno_call_part_clobbered instead of
11064 HARD_REGNO_CALL_PART_CLOBBERED.
11065 * ira-conflicts.c (ira_build_conflicts): Likewise.
11066 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11067 * lra-constraints.c (need_for_call_save_p): Likewise.
11068 * lra-lives.c: Include target.h.
11069 (check_pseudos_live_through_calls): Use
11070 targetm.hard_regno_call_part_clobbered instead of
11071 HARD_REGNO_CALL_PART_CLOBBERED.
11072 * regcprop.c: Include target.h.
11073 (copyprop_hardreg_forward_1): Use
11074 targetm.hard_regno_call_part_clobbered instead of
11075 HARD_REGNO_CALL_PART_CLOBBERED.
11076 * reginfo.c (choose_hard_reg_mode): Likewise.
11077 * regrename.c (check_new_reg_p): Likewise.
11078 * reload.c (find_equiv_reg): Likewise.
11079 * reload1.c (emit_reload_insns): Likewise.
11080 * sched-deps.c (deps_analyze_insn): Likewise.
11081 * sel-sched.c (init_regs_for_mode): Likewise.
11082 (mark_unavailable_hard_regs): Likewise.
11083 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11084 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11085 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11087 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11088 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11089 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11091 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11093 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11094 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11095 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11097 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11098 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11099 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11101 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11102 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11104 * config/powerpcspe/powerpcspe.c
11105 (rs6000_hard_regno_call_part_clobbered): New function.
11106 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11107 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11108 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11110 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11111 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11112 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11114 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11115 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11116 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11118 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11119 Alan Hayward <alan.hayward@arm.com>
11120 David Sherwood <david.sherwood@arm.com>
11122 * rtl.h (subreg_memory_offset): Declare.
11123 * emit-rtl.c (subreg_memory_offset): New function.
11124 * expmed.c (store_bit_field_1): Use it.
11125 * expr.c (undefined_operand_subword_p): Likewise.
11126 * simplify-rtx.c (simplify_subreg): Likewise.
11128 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11130 PR rtl-optimization/57448
11133 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11134 using atomic_load pattern.
11135 (expand_atomic_store): Likewise.
11137 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11140 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11141 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11142 handling. Use replace_call_with_value with NULL instead of
11143 gsi_replace, unlink_stmt_vdef and release_defs.
11145 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11148 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11150 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11153 * rtl.h (emit_library_call_value_1): Declare.
11154 (emit_library_call): Replace declaration with a series of overloads.
11155 Remove the parameter count argument.
11156 (emit_library_call_value): Likewise.
11157 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11158 with an "rtx_mode_t *".
11159 (emit_library_call_value): Delete.
11160 (emit_library_call): Likewise.
11161 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11162 (asan_emit_allocas_unpoison): Likewise.
11163 * builtins.c (expand_builtin_powi): Likewise.
11164 (expand_asan_emit_allocas_unpoison): Likewise.
11165 * cfgexpand.c (expand_main_function): Likewise.
11166 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11167 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11168 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11169 * config/arm/arm.c (arm_trampoline_init): Likewise.
11170 (arm_call_tls_get_addr): Likewise.
11171 (arm_expand_divmod_libfunc): Likewise.
11172 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11173 (smulsi3_highpart): Likewise.
11174 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11175 (c6x_expand_compare): Likewise.
11176 (c6x_expand_movmem): Likewise.
11177 * config/frv/frv.c (frv_trampoline_init): Likewise.
11178 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11179 (ix86_expand_divmod_libfunc): Likewise.
11180 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11181 (ia64_expand_compare): Likewise.
11182 (ia64_profile_hook): Likewise.
11183 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11184 (nonlocal_goto): Likewise.
11185 (restore_stack_nonlocal): Likewise.
11186 * config/m32r/m32r.c (block_move_call): Likewise.
11187 (m32r_trampoline_init): Likewise.
11188 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11189 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11190 (m68k_call_m68k_read_tp): Likewise.
11191 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11192 (microblaze_expand_divide): Likewise.
11193 * config/mips/mips.h (mips_args): Likewise.
11194 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11195 (MIPS_ICACHE_SYNC): Likewise.
11196 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11197 (nios2_trampoline_init): Likewise.
11198 * config/pa/pa.c (hppa_tls_call): Likewise.
11199 (pa_trampoline_init): Likewise.
11200 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11201 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11202 (expand_strn_compare): Likewise.
11203 (rs6000_generate_compare): Likewise.
11204 (rs6000_expand_float128_convert): Likewise.
11205 (output_profile_hook): Likewise.
11206 (rs6000_trampoline_init): Likewise.
11207 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11208 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11209 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11210 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11211 (rs6000_generate_compare): Likewise.
11212 (rs6000_expand_float128_convert): Likewise.
11213 (output_profile_hook): Likewise.
11214 (rs6000_trampoline_init): Likewise.
11215 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11216 * config/sh/sh.c (sh_trampoline_init): Likewise.
11217 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11218 (sparc_emit_float_lib_cmp): Likewise.
11219 (sparc32_initialize_trampoline): Likewise.
11220 (sparc64_initialize_trampoline): Likewise.
11221 (sparc_profile_hook): Likewise.
11222 * config/spu/spu.c (ea_load_store): Likewise.
11223 * config/spu/spu.md (floatunssidf2): Likewise.
11224 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11225 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11226 * config/visium/visium.c (expand_block_move_4): Likewise.
11227 (expand_block_move_2): Likewise.
11228 (expand_block_move_1): Likewise.
11229 (expand_block_set_4): Likewise.
11230 (expand_block_set_2): Likewise.
11231 (expand_block_set_1): Likewise.
11232 (visium_trampoline_init): Likewise.
11233 (visium_profile_hook): Likewise.
11234 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11235 (xtensa_setup_frame_addresses): Likewise.
11236 (xtensa_trampoline_init): Likewise.
11237 * except.c (sjlj_emit_function_enter): Likewise.
11238 (sjlj_emit_function_exit): Likewise.
11239 * explow.c (allocate_dynamic_stack_space): Likewise.
11240 (probe_stack_range): Likewise.
11241 * expr.c (convert_mode_scalar): Likewise.
11242 * optabs.c (expand_binop): Likewise.
11243 (expand_twoval_binop_libfunc): Likewise.
11244 (expand_unop): Likewise.
11245 (prepare_cmp_insn): Likewise.
11246 (prepare_float_lib_cmp): Likewise.
11247 (expand_float): Likewise.
11248 (expand_fix): Likewise.
11249 (expand_fixed_convert): Likewise.
11250 (maybe_emit_sync_lock_test_and_set): Likewise.
11251 (expand_atomic_compare_and_swap): Likewise.
11252 (expand_mem_thread_fence): Likewise.
11253 (expand_atomic_fetch_op): Likewise.
11255 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11257 * doc/generic.texi (OpenACC): Adjust URL.
11258 * doc/invoke.texi (C Dialect Options): Ditto.
11260 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11262 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11263 predicate for operand 1. Add (m,<S>) constraint.
11264 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11265 Prevent memory operand 1 with register operand 2.
11267 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11269 PR rtl-optimization/82024
11270 * combine.c (try_combine): If the combination result is a PARALLEL,
11271 and we only need to retain the SET in there that would be placed
11272 at I2, check that we can place that at I3 instead, before doing so.
11274 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11277 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11279 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11280 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11283 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11284 Jeff Law <law@redhat.com>
11286 * varasm.c (bss_initializer_p): Do not put constants into .bss
11287 (categorize_decl_for_section): Handle bss_initializer_p returning
11288 false when DECL_INITIAL is NULL.
11290 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11293 * config/s390/s390.c (s390_can_inline_p): New function.
11295 2017-09-01 Jeff Law <law@redhat.com>
11297 PR tree-optimization/82052
11298 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11299 Always initialize the returned slot after a hash table miss
11300 when INSERT is true.
11302 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11304 * config/s390/s390.md (mem_signal_fence): Remove.
11305 * doc/md.texi (mem_signal_fence): Remove.
11306 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11308 * target-insns.def (mem_signal_fence): Remove.
11310 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11313 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11316 * asan.c (create_odr_indicator): Strip name encoding from assembler
11317 name before appending it after __odr_asan_.
11319 2017-09-01 Martin Liska <mliska@suse.cz>
11321 PR tree-optimization/82059
11322 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11323 frequency only when an edge is redirected.
11325 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11327 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11328 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11329 (arc_conditional_register_usage): Remove ARC600 lp_count
11331 (arc_file_start): Emit Tag_ARC_CPU_variation.
11332 (arc_can_use_doloop_p): New conditions to use ZOLs.
11333 (hwloop_fail): New function.
11334 (hwloop_optimize): Likewise.
11335 (hwloop_pattern_reg): Likewise.
11336 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11337 (arc_reorg_loops): New function, calls reorg_loops.
11338 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11339 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11341 (arc_loop_hazard): Remove function, functionality moved into
11343 (arc_hazard): Remove arc_loop_hazard call.
11344 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11345 into hwloop_optimize.
11346 (arc_label_align): Remove ZOL handling.
11347 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11348 * config/arc/arc.md (doloop_begin): Remove pattern.
11349 (doloop_begin_i): Likewise.
11350 (doloop_end_i): Likewise.
11351 (doloop_fallback): Likewise.
11352 (doloop_fallback_m): Likewise.
11353 (doloop_end): Reimplement expand.
11354 (arc_lp): New pattern for LP instruction.
11355 (loop_end): New pattern.
11356 (loop_fail): Likewise.
11357 (decrement_and_branch_until_zero): Likewise.
11358 * config/arc/arc.opt (mlpc-width): New option.
11359 * doc/invoke.texi (mlpc-width): Document option.
11361 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11363 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11364 (arc_ccfsm_advance): Fix checking for delay slots.
11365 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11367 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11369 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11371 (movhi_insn): Update store instruction constraint which are saving
11372 6-bit short immediates.
11373 (movsi_insn): Consider also short scaled load operations.
11374 (zero_extendhisi2_i): Use Usd constraint instead of T.
11375 (extendhisi2_i): Add q constraint.
11376 (arc_clzsi2): Add type and length attributes.
11377 (arc_ctzsi2): Likewise.
11378 * config/arc/constraints.md (Usc): Update constraint, the
11379 assembler can parse two relocations for a single instruction.
11381 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11383 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11384 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11386 2017-08-31 Olivier Hainque <hainque@adacore.com>
11388 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11389 match as powerpc-wrs-vxworks*.
11391 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11393 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11394 register constraint for by-element operand.
11395 (aarch64_mls_elt_merge<mode>): Likewise.
11397 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11399 * config/arc/arc.c (arc_can_follow_jump): Check for short
11402 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11404 * config.gcc: Use g.opt for arc.
11405 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11406 functionality moved to ...
11407 (legitimate_scaled_address_p): New function, ...here.
11408 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11409 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11410 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11412 (arc_override_options): Handle G option.
11413 (arc_output_pic_addr_const): Correct function definition.
11414 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11415 (arc_decl_anon_ns_mem_p): Delete.
11416 (arc_in_small_data_p): Overhaul this function to take into
11417 consideration the value given via G option.
11418 (arc_rewrite_small_data_1): Renamed and corrected old
11419 arc_rewrite_small_data function.
11420 (arc_rewrite_small_data): New function.
11421 (small_data_pattern): Don't use pic_offset_table_rtx.
11422 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11423 * config/arc/simdext.md (movmisalignv2hi): Use
11424 prepare_move_operands function.
11426 (movmisalign*): Likewise.
11427 * doc/invoke.texi (ARC options): Document -G option.
11429 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11431 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11433 * config/arc/arc.c (arc_print_operand): Output scalled address for
11434 sdata whenever is possible.
11435 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11436 load/stores are available.
11437 (compact_sda_memory_operand): Check for the alignment required by
11438 code density instructions.
11439 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11441 * config/arc/constraints.md (Usd): Update constraint.
11442 (Us0): New constraint.
11443 (Usc): Update constraint.
11445 2017-08-31 Richard Biener <rguenther@suse.de>
11447 PR middle-end/82054
11448 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11449 function only once.
11451 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11453 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11454 Resize type_signature.
11456 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11457 Alan Hayward <alan.hayward@arm.com>
11458 David Sherwood <david.sherwood@arm.com>
11460 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11461 subregs whose inner modes can be stored in GPRs.
11462 (aarch64_classify_index): Likewise.
11464 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11465 Alan Hayward <alan.hayward@arm.com>
11466 David Sherwood <david.sherwood@arm.com>
11468 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11469 (V_INT_EQUIV): ...this.
11470 (v_cmp_result): Rename to...
11471 (v_int_equiv): ...this.
11472 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11473 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11474 (copysign<mode>3): Likewise.
11475 (aarch64_simd_bsl<mode>_internal): Likewise.
11476 (aarch64_simd_bsl<mode>): Likewise.
11477 (vec_cmp<mode><mode>): Likewise.
11478 (vcond<mode><mode>): Likewise.
11479 (vcond<v_cmp_mixed><mode>): Likewise.
11480 (vcondu<mode><v_cmp_mixed>): Likewise.
11481 (aarch64_cm<optab><mode>): Likewise.
11482 (aarch64_cmtst<mode>): Likewise.
11483 (aarch64_fac<optab><mode>): Likewise.
11484 (vec_perm_const<mode>): Likewise.
11485 (vcond_mask_<mode><v_cmp_result>): Rename to...
11486 (vcond_mask_<mode><v_int_equiv>): ...this.
11487 (vec_cmp<mode><v_cmp_result>): Rename to...
11488 (vec_cmp<mode><v_int_equiv>): ...this.
11490 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11491 Alan Hayward <alan.hayward@arm.com>
11492 David Sherwood <david.sherwood@arm.com>
11494 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11496 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11497 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11498 (UNSPEC_LD4_DREG): New unspecs.
11499 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11500 (aarch64_ld2<mode>_dreg_be): Replace with...
11501 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11503 (aarch64_ld3<mode>_dreg_le)
11504 (aarch64_ld3<mode>_dreg_be): Replace with...
11505 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11507 (aarch64_ld4<mode>_dreg_le)
11508 (aarch64_ld4<mode>_dreg_be): Replace with...
11509 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11512 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11514 PR tree-optimization/81987
11515 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11516 insert an initializer in a location not dominated by the stride
11519 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11521 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11522 on the entire header of the finally block in the fallthru case.
11524 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11526 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11528 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11530 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11531 rs6000_emit_move_from_cr and call renamed function.
11532 (rs6000_emit_prologue): Call renamed functions.
11533 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11534 movesi_from_cr, remove volatile CRs.
11536 2017-08-30 Jon Beniston <jon@beniston.com>
11537 Richard Biener <rguenther@suse.de>
11539 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11540 of VECTOR_MODE_P check.
11541 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11542 element vector types.
11544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11546 * df.h (df_read_modify_subreg_p): Remove in favor of...
11547 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11548 const_rtx instead of an rtx.
11549 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11550 * df-problems.c (df_word_lr_mark_ref): Likewise.
11551 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11552 (mark_pseudo_reg_dead): Likewise.
11553 (mark_ref_dead): Likewise.
11554 * reginfo.c (init_subregs_of_mode): Likewise.
11555 * sched-deps.c (sched_analyze_1): Likewise.
11556 * df-scan.c (df_def_record_1): Likewise.
11557 (df_uses_record): Likewise.
11558 (df_read_modify_subreg_p): Remove in favor of...
11559 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11560 const_rtx instead of an rtx.
11562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11563 Alan Hayward <alan.hayward@arm.com>
11564 David Sherwood <david.sherwood@arm.com>
11566 * rtl.h (partial_subreg_p): New function.
11567 * caller-save.c (save_call_clobbered_regs): Use it.
11568 * calls.c (expand_call): Likewise.
11569 * combine.c (combinable_i3pat): Likewise.
11570 (simplify_set): Likewise.
11571 (make_extraction): Likewise.
11572 (make_compound_operation_int): Likewise.
11573 (gen_lowpart_or_truncate): Likewise.
11574 (force_to_mode): Likewise.
11575 (make_field_assignment): Likewise.
11576 (reg_truncated_to_mode): Likewise.
11577 (record_truncated_value): Likewise.
11578 (move_deaths): Likewise.
11579 * cse.c (record_jump_cond): Likewise.
11580 (cse_insn): Likewise.
11581 * cselib.c (cselib_lookup_1): Likewise.
11582 * expmed.c (extract_bit_field_using_extv): Likewise.
11583 * function.c (assign_parm_setup_reg): Likewise.
11584 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11585 * ira-build.c (create_insn_allocnos): Likewise.
11586 * lra-coalesce.c (merge_pseudos): Likewise.
11587 * lra-constraints.c (match_reload): Likewise.
11588 (simplify_operand_subreg): Likewise.
11589 (curr_insn_transform): Likewise.
11590 * lra-lives.c (process_bb_lives): Likewise.
11591 * lra.c (new_insn_reg): Likewise.
11592 (lra_substitute_pseudo): Likewise.
11593 * regcprop.c (mode_change_ok): Likewise.
11594 (maybe_mode_change): Likewise.
11595 (copyprop_hardreg_forward_1): Likewise.
11596 * reload.c (push_reload): Likewise.
11597 (find_reloads): Likewise.
11598 (find_reloads_subreg_address): Likewise.
11599 * reload1.c (alter_reg): Likewise.
11600 (eliminate_regs_1): Likewise.
11601 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11603 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11605 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11606 back to if statements, including unpack.
11608 2017-08-30 Martin Liska <mliska@suse.cz>
11610 PR inline-asm/82001
11611 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11613 (func_checker::compare_asm_inputs_outputs): ... this function.
11614 (func_checker::compare_gimple_asm): Use the function to compare
11615 also ASM constrains.
11616 * ipa-icf-gimple.h: Rename the function.
11618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11619 Alan Hayward <alan.hayward@arm.com>
11620 David Sherwood <david.sherwood@arm.com>
11622 * coretypes.h (complex_mode): New type.
11623 * gdbhooks.py (build_pretty_printer): Handle it.
11624 * machmode.h (complex_mode): New class.
11625 (complex_mode::includes_p): New function.
11626 (is_complex_int_mode): Likewise.
11627 (is_complex_float_mode): Likewise.
11628 * genmodes.c (get_mode_class): Handle complex mode classes.
11629 * function.c (expand_function_end): Use is_complex_int_mode.
11631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11632 Alan Hayward <alan.hayward@arm.com>
11633 David Sherwood <david.sherwood@arm.com>
11635 * coretypes.h (scalar_mode_pod): New typedef.
11636 * gdbhooks.py (build_pretty_printer): Handle it.
11637 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11638 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11639 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11640 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11641 as_a <scalar_mode>.
11643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11644 Alan Hayward <alan.hayward@arm.com>
11645 David Sherwood <david.sherwood@arm.com>
11647 * machmode.h (mode_for_vector): Take a scalar_mode instead
11649 * stor-layout.c (mode_for_vector): Likewise.
11650 * explow.c (promote_mode): Use as_a <scalar_mode>.
11651 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11654 Alan Hayward <alan.hayward@arm.com>
11655 David Sherwood <david.sherwood@arm.com>
11657 * target.def (preferred_simd_mode): Take a scalar_mode
11658 instead of a machine_mode.
11659 * targhooks.h (default_preferred_simd_mode): Likewise.
11660 * targhooks.c (default_preferred_simd_mode): Likewise.
11661 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11662 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11663 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11664 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11665 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11666 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11667 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11668 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11670 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11671 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11672 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11673 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11674 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11675 * doc/tm.texi: Regenerate.
11676 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11680 Alan Hayward <alan.hayward@arm.com>
11681 David Sherwood <david.sherwood@arm.com>
11683 * target.def (scalar_mode_supported_p): Take a scalar_mode
11684 instead of a machine_mode.
11685 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11686 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11687 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11688 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11689 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11690 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11691 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11692 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11693 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11694 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11695 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11696 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11697 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11698 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11700 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11701 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11702 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11703 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11704 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11706 * doc/tm.texi: Regenerate.
11708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11709 Alan Hayward <alan.hayward@arm.com>
11710 David Sherwood <david.sherwood@arm.com>
11712 * coretypes.h (opt_scalar_mode): New typedef.
11713 * gdbhooks.py (build_pretty_printers): Handle it.
11714 * machmode.h (mode_iterator::get_2xwider): Add overload for
11716 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11718 * expr.c (convert_mode_scalar): Likewise.
11719 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11720 * optabs.c (expand_float): Likewise.
11721 (expand_fix): Likewise.
11722 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11725 Alan Hayward <alan.hayward@arm.com>
11726 David Sherwood <david.sherwood@arm.com>
11728 * optabs.c (expand_float): Explicitly check for scalars before
11729 using a branching expansion.
11730 (expand_fix): Likewise.
11732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11733 Alan Hayward <alan.hayward@arm.com>
11734 David Sherwood <david.sherwood@arm.com>
11736 * expr.c (convert_mode): Split scalar handling out into...
11737 (convert_mode_scalar): ...this new function. Treat the modes
11740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11741 Alan Hayward <alan.hayward@arm.com>
11742 David Sherwood <david.sherwood@arm.com>
11744 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11746 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11749 Alan Hayward <alan.hayward@arm.com>
11750 David Sherwood <david.sherwood@arm.com>
11752 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11753 rather than a machine_mode.
11754 (fixed_from_string): Likewise.
11755 (fixed_convert): Likewise.
11756 (fixed_convert_from_int): Likewise.
11757 (fixed_convert_from_real): Likewise.
11758 (real_convert_from_fixed): Likewise.
11759 * fixed-value.c (fixed_from_double_int): Likewise.
11760 (fixed_from_string): Likewise.
11761 (fixed_convert): Likewise.
11762 (fixed_convert_from_int): Likewise.
11763 (fixed_convert_from_real): Likewise.
11764 (real_convert_from_fixed): Likewise.
11765 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11768 Alan Hayward <alan.hayward@arm.com>
11769 David Sherwood <david.sherwood@arm.com>
11771 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11772 of separate mode class checks. Do not allow vector modes here.
11773 (immed_wide_int_const): Use as_a <scalar_mode>.
11774 * explow.c (trunc_int_for_mode): Likewise.
11775 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11776 (wi::shwi): Likewise.
11777 (wi::min_value): Likewise.
11778 (wi::max_value): Likewise.
11779 * dwarf2out.c (loc_descriptor): Likewise.
11780 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11781 for CONST_WIDE_INT.
11783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11784 Alan Hayward <alan.hayward@arm.com>
11785 David Sherwood <david.sherwood@arm.com>
11787 * tree.h (SCALAR_TYPE_MODE): New macro.
11788 * expr.c (expand_expr_addr_expr_1): Use it.
11789 (expand_expr_real_2): Likewise.
11790 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11791 (fold_convert_const_fixed_from_int): Likewise.
11792 (fold_convert_const_fixed_from_real): Likewise.
11793 (native_encode_fixed): Likewise
11794 (native_encode_complex): Likewise
11795 (native_encode_vector): Likewise.
11796 (native_interpret_fixed): Likewise.
11797 (native_interpret_real): Likewise.
11798 (native_interpret_complex): Likewise.
11799 (native_interpret_vector): Likewise.
11800 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11801 (simd_clone_adjust_argument_types): Likewise.
11802 (simd_clone_init_simd_arrays): Likewise.
11803 (simd_clone_adjust): Likewise.
11804 * stor-layout.c (layout_type): Likewise.
11805 * tree.c (build_minus_one_cst): Likewise.
11806 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11807 * tree-inline.c (estimate_move_cost): Likewise.
11808 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11809 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11810 (vectorizable_reduction): Likewise.
11811 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11812 (vect_recog_mixed_size_cond_pattern): Likewise.
11813 (check_bool_pattern): Likewise.
11814 (adjust_bool_pattern): Likewise.
11815 (search_type_for_mask_1): Likewise.
11816 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11817 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11818 (vectorizable_load): Likewise.
11819 (vectorizable_store): Likewise.
11820 * ubsan.c (ubsan_encode_value): Likewise.
11821 * varasm.c (output_constant): Likewise.
11823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11824 Alan Hayward <alan.hayward@arm.com>
11825 David Sherwood <david.sherwood@arm.com>
11827 * coretypes.h (scalar_mode): New class.
11828 * machmode.h (scalar_mode): Likewise.
11829 (scalar_mode::includes_p): New function.
11830 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11831 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11832 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11833 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11834 * expmed.c (store_bit_field_1): Likewise.
11835 (extract_bit_field_1): Likewise.
11836 * expr.c (write_complex_part): Likewise.
11837 (read_complex_part): Likewise.
11838 (emit_move_complex_push): Likewise.
11839 (expand_expr_real_2): Likewise.
11840 * function.c (assign_parm_setup_reg): Likewise.
11841 (assign_parms_unsplit_complex): Likewise.
11842 * optabs.c (expand_binop): Likewise.
11843 * rtlanal.c (subreg_get_info): Likewise.
11844 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11845 * varasm.c (output_constant_pool_2): Likewise.
11847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11848 Alan Hayward <alan.hayward@arm.com>
11849 David Sherwood <david.sherwood@arm.com>
11851 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11853 (expmed_mult_highpart_optab): Likewise.
11854 (expmed_mult_highpart): Likewise.
11856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11857 Alan Hayward <alan.hayward@arm.com>
11858 David Sherwood <david.sherwood@arm.com>
11860 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11861 instead of a machine_mode.
11862 (builtin_memset_read_str): Likewise.
11863 * builtins.c (c_readstr): Likewise.
11864 (builtin_memcpy_read_str): Likewise.
11865 (builtin_strncpy_read_str): Likewise.
11866 (builtin_memset_read_str): Likewise.
11867 (builtin_memset_gen_str): Likewise.
11868 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11869 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11870 instead of a machine_mode.
11871 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11873 (make_extraction): Likewise.
11874 (try_widen_shift_mode): Take and return scalar_int_modes instead
11876 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11877 a scalar_int_mode instead of a machine_mode.
11878 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11879 (avr_addr_space_pointer_mode): Likewise.
11880 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11881 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11882 (msp430_unwind_word_mode): Likewise.
11883 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11884 (spu_addr_space_pointer_mode): Likewise.
11885 (spu_addr_space_address_mode): Likewise.
11886 (spu_libgcc_cmp_return_mode): Likewise.
11887 (spu_libgcc_shift_count_mode): Likewise.
11888 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11889 (rl78_addr_space_pointer_mode): Likewise.
11890 (fl78_unwind_word_mode): Likewise.
11891 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11893 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11894 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11895 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11896 (mips_valid_pointer_mode): Likewise.
11897 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11898 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11899 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11901 (ft32_addr_space_address_mode): Likewise.
11902 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11903 scalar_int_mode instead of a machine_mode.
11904 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11906 (m32c_addr_space_address_mode): Likewise.
11907 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11908 (rs6000_eh_return_filter_mode): Likewise.
11909 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11910 (rs6000_eh_return_filter_mode): Likewise.
11911 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11912 (s390_libgcc_shift_count_mode): Likewise.
11913 (s390_unwind_word_mode): Likewise.
11914 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11916 * target.def (mode_rep_extended): Likewise.
11917 (valid_pointer_mode): Likewise.
11918 (addr_space.valid_pointer_mode): Likewise.
11919 (eh_return_filter_mode): Return a scalar_int_mode rather than
11921 (libgcc_cmp_return_mode): Likewise.
11922 (libgcc_shift_count_mode): Likewise.
11923 (unwind_word_mode): Likewise.
11924 (addr_space.pointer_mode): Likewise.
11925 (addr_space.address_mode): Likewise.
11926 * doc/tm.texi: Regenerate.
11927 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11929 (do_jump): Use scalar_int_mode for local variables.
11930 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11931 rather than a machine_mode.
11932 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11933 (scompare_loc_descriptor_wide): Likewise.
11934 (scompare_loc_descriptor_narrow): Likewise.
11935 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11937 * except.c (sjlj_emit_dispatch_table): Likewise.
11938 (expand_builtin_eh_copy_values): Likewise.
11939 * explow.c (convert_memory_address_addr_space_1): Likewise.
11940 Take a scalar_int_mode rather than a machine_mode.
11941 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11942 than a machine_mode.
11943 (memory_address_addr_space): Use scalar_int_mode for local variables.
11944 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11945 rather than a machine_mode.
11946 * expmed.c (mask_rtx): Likewise.
11947 (init_expmed_one_conv): Likewise.
11948 (expand_mult_highpart_adjust): Likewise.
11949 (extract_high_half): Likewise.
11950 (expmed_mult_highpart_optab): Likewise.
11951 (expmed_mult_highpart): Likewise.
11952 (expand_smod_pow2): Likewise.
11953 (expand_sdiv_pow2): Likewise.
11954 (emit_store_flag_int): Likewise.
11955 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11957 (extract_low_bits): Likewise.
11958 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11960 * expr.c (pieces_addr::adjust): Likewise.
11961 (can_store_by_pieces): Likewise.
11962 (store_by_pieces): Likewise.
11963 (clear_by_pieces_1): Likewise.
11964 (expand_expr_addr_expr_1): Likewise.
11965 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11966 (expand_expr_real_1): Likewise.
11967 (try_casesi): Likewise.
11968 * final.c (shorten_branches): Likewise.
11969 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11970 type of "mode" to machine_mode.
11971 * internal-fn.c (expand_arith_overflow_result_store): Take a
11972 scalar_int_mode rather than a machine_mode.
11973 (expand_mul_overflow): Use scalar_int_mode for local variables.
11974 * loop-doloop.c (doloop_modify): Likewise.
11975 (doloop_optimize): Likewise.
11976 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11977 than a machine_mode.
11978 (expand_doubleword_shift_condmove): Likewise.
11979 (expand_doubleword_shift): Likewise.
11980 (expand_doubleword_clz): Likewise.
11981 (expand_doubleword_popcount): Likewise.
11982 (expand_doubleword_parity): Likewise.
11983 (expand_absneg_bit): Use scalar_int_mode for local variables.
11984 (prepare_float_lib_cmp): Likewise.
11985 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11986 rather than a machine_mode.
11987 (convert_memory_address_addr_space): Likewise.
11988 (get_mode_bounds): Likewise.
11989 (get_address_mode): Return a scalar_int_mode rather than a
11991 * rtlanal.c (get_address_mode): Likewise.
11992 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11993 than a machine_mode.
11994 * targhooks.c (default_mode_rep_extended): Likewise.
11995 (default_valid_pointer_mode): Likewise.
11996 (default_addr_space_valid_pointer_mode): Likewise.
11997 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11998 than a machine_mode.
11999 (default_libgcc_cmp_return_mode): Likewise.
12000 (default_libgcc_shift_count_mode): Likewise.
12001 (default_unwind_word_mode): Likewise.
12002 (default_addr_space_pointer_mode): Likewise.
12003 (default_addr_space_address_mode): Likewise.
12004 * targhooks.h (default_eh_return_filter_mode): Likewise.
12005 (default_libgcc_cmp_return_mode): Likewise.
12006 (default_libgcc_shift_count_mode): Likewise.
12007 (default_unwind_word_mode): Likewise.
12008 (default_addr_space_pointer_mode): Likewise.
12009 (default_addr_space_address_mode): Likewise.
12010 (default_mode_rep_extended): Take a scalar_int_mode rather than
12012 (default_valid_pointer_mode): Likewise.
12013 (default_addr_space_valid_pointer_mode): Likewise.
12014 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12016 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12017 rather than a machine_mode.
12018 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12019 for local variables.
12020 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12021 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12022 than a machine_mode.
12024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12025 Alan Hayward <alan.hayward@arm.com>
12026 David Sherwood <david.sherwood@arm.com>
12028 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12029 the mode argument to scalar_int_mode.
12030 (do_jump_by_parts_zero_rtx): Likewise.
12031 (do_jump_by_parts_equality_rtx): Likewise.
12032 (do_jump_by_parts_greater): Take a mode argument.
12033 (do_jump_by_parts_equality): Likewise.
12034 (do_jump_1): Update calls accordingly.
12036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12037 Alan Hayward <alan.hayward@arm.com>
12038 David Sherwood <david.sherwood@arm.com>
12040 * is-a.h (safe_dyn_cast): New function.
12041 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12042 (jump_table_for_label): Likewise.
12043 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12044 instead of an rtx_insn *.
12045 (shorten_branches): Use dyn_cast instead of LABEL_P and
12046 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12047 rtx_jump_table_data::get_data_mode.
12048 (final_scan_insn): Likewise.
12050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12051 Alan Hayward <alan.hayward@arm.com>
12052 David Sherwood <david.sherwood@arm.com>
12054 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12055 trying to combine a full-register integer set with a subreg
12058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12059 Alan Hayward <alan.hayward@arm.com>
12060 David Sherwood <david.sherwood@arm.com>
12062 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12063 that is always either address_mode or pointer_mode.
12065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12066 Alan Hayward <alan.hayward@arm.com>
12067 David Sherwood <david.sherwood@arm.com>
12069 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12070 when the two are known to be equal.
12072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12073 Alan Hayward <alan.hayward@arm.com>
12074 David Sherwood <david.sherwood@arm.com>
12076 * simplify-rtx.c (simplify_const_unary_operation): Use
12077 is_a <scalar_int_mode> instead of checking for a nonzero
12078 precision. Forcibly convert op_mode to a scalar_int_mode
12079 in that case. More clearly differentiate the operand and
12080 result modes and use the former when deciding what the value
12081 of a count-bits operation should be. Use is_int_mode instead
12082 of checking for a MODE_INT. Remove redundant check for whether
12083 this mode has a zero precision.
12085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12086 Alan Hayward <alan.hayward@arm.com>
12087 David Sherwood <david.sherwood@arm.com>
12089 * optabs.c (widen_leading): Change the type of the mode argument
12090 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12091 (widen_bswap): Likewise.
12092 (expand_parity): Likewise.
12093 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12094 (expand_ffs): Likewise.
12095 (epand_unop): Check for scalar integer modes before calling the
12098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12099 Alan Hayward <alan.hayward@arm.com>
12100 David Sherwood <david.sherwood@arm.com>
12102 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12104 (expand_expr_real_1): Update call accordingly.
12106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12107 Alan Hayward <alan.hayward@arm.com>
12108 David Sherwood <david.sherwood@arm.com>
12110 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12111 value_mode arguments. Use scalar_int_mode internally.
12112 (store_bit_field_1): Rename the new integer mode from imode
12113 to op0_mode and use it instead of GET_MODE (op0). Update calls
12114 to store_split_bit_field, store_bit_field_using_insv and
12115 store_fixed_bit_field.
12116 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12117 Use scalar_int_mode internally. Use a bit count rather than a mode
12118 when calculating the largest bit size for get_best_mode.
12119 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12120 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12121 Remove assertion that OP0 has a scalar integer mode.
12122 (store_split_bit_field): Add op0_mode and value_mode arguments.
12123 Update calls to extract_fixed_bit_field.
12124 (extract_bit_field_using_extv): Add an op0_mode argument.
12125 Use scalar_int_mode internally.
12126 (extract_bit_field_1): Rename the new integer mode from imode to
12127 op0_mode and use it instead of GET_MODE (op0). Update calls to
12128 extract_split_bit_field, extract_bit_field_using_extv and
12129 extract_fixed_bit_field.
12130 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12131 to extract_split_bit_field and extract_fixed_bit_field_1.
12132 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12133 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12134 on the target mode.
12135 (extract_split_bit_field): Add an op0_mode argument. Update call
12136 to extract_fixed_bit_field.
12138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12139 Alan Hayward <alan.hayward@arm.com>
12140 David Sherwood <david.sherwood@arm.com>
12142 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12143 * explow.c (hard_function_value): Likewise.
12144 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12145 convert_to_mode call outside the loop.
12146 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12147 for the mode iterator. Require the mode specified by max_pieces
12149 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12151 (copy_blkmode_to_reg): Likewise.
12152 (set_storage_via_setmem): Likewise.
12153 * optabs.c (prepare_cmp_insn): Likewise.
12154 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12155 * stor-layout.c (finish_bitfield_representative): Likewise.
12157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12158 Alan Hayward <alan.hayward@arm.com>
12159 David Sherwood <david.sherwood@arm.com>
12161 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12162 * expr.c (convert_move): Use them.
12163 (convert_modes): Likewise.
12164 (store_expr_with_bounds): Likewise.
12166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12167 Alan Hayward <alan.hayward@arm.com>
12168 David Sherwood <david.sherwood@arm.com>
12170 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12171 parameter for the mode of "x". Remove the "known_x", "known_mode"
12172 and "known_ret" arguments. Change the type of the mode argument
12173 to scalar_int_mode.
12174 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12175 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12176 (reg_num_sign_bit_copies_for_combine): Likewise.
12177 * rtlanal.c (nonzero_bits1): Likewise.
12178 (num_sign_bit_copies1): Likewise.
12179 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12180 (reg_num_sign_bit_copies_general): Likewise.
12181 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12182 (reg_nonzero_bits_general): Likewise.
12184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12185 Alan Hayward <alan.hayward@arm.com>
12186 David Sherwood <david.sherwood@arm.com>
12188 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12189 than in subroutines. Return 1 for non-integer modes.
12190 (cached_num_sign_bit_copies): Change the type of the mode parameter
12191 to scalar_int_mode.
12192 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12193 classes. Handle CONST_INT_P first and then check whether X also
12194 has a scalar integer mode. Check the same thing for inner registers
12195 of a SUBREG and for values that are being extended or truncated.
12197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12198 Alan Hayward <alan.hayward@arm.com>
12199 David Sherwood <david.sherwood@arm.com>
12201 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12202 in subroutines. Return the mode mask for non-integer modes.
12203 (cached_nonzero_bits): Change the type of the mode parameter
12204 to scalar_int_mode.
12205 (nonzero_bits1): Likewise. Remove early exit for other mode
12206 classes. Handle CONST_INT_P first and then check whether X
12207 also has a scalar integer mode.
12209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12210 Alan Hayward <alan.hayward@arm.com>
12211 David Sherwood <david.sherwood@arm.com>
12213 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12214 Return a scalar_int_mode and assert that the size is greater than
12216 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12217 (op_by_pieces_d::op_by_pieces_d): Likewise.
12218 (op_by_pieces_d::run): Likewise.
12219 (can_store_by_pieces): Likewise.
12221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12222 Alan Hayward <alan.hayward@arm.com>
12223 David Sherwood <david.sherwood@arm.com>
12225 * combine.c (extract_left_shift): Add a mode argument and update
12227 (make_compound_operation_int): Change the type of the mode parameter
12228 to scalar_int_mode and update the call to extract_left_shift.
12230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12231 Alan Hayward <alan.hayward@arm.com>
12232 David Sherwood <david.sherwood@arm.com>
12234 * combine.c (simplify_and_const_int): Change the type of the mode
12235 parameter to scalar_int_mode.
12236 (simplify_and_const_int_1): Likewise. Update recursive call.
12238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12239 Alan Hayward <alan.hayward@arm.com>
12240 David Sherwood <david.sherwood@arm.com>
12242 * combine.c (simplify_compare_const): Check that the mode is a
12243 scalar_int_mode (rather than VOIDmode) before testing its
12245 (simplify_comparison): Move COMPARISON_P handling out of the
12246 loop and restrict the latter part of the loop to scalar_int_modes.
12247 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12248 and when considering SUBREG_REGs. Use is_int_mode instead of
12249 checking GET_MODE_CLASS against MODE_INT.
12251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12252 Alan Hayward <alan.hayward@arm.com>
12253 David Sherwood <david.sherwood@arm.com>
12255 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12256 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12257 shift_unit_mode and for modes involved in scalar shifts.
12259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12260 Alan Hayward <alan.hayward@arm.com>
12261 David Sherwood <david.sherwood@arm.com>
12263 * combine.c (force_int_to_mode): New function, split out from...
12264 (force_to_mode): ...here. Keep xmode up-to-date and use it
12265 instead of GET_MODE (x).
12267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12268 Alan Hayward <alan.hayward@arm.com>
12269 David Sherwood <david.sherwood@arm.com>
12271 * optabs-query.h (extraction_insn::struct_mode): Change type to
12272 opt_scalar_int_mode and update comment.
12273 (extraction_insn::field_mode): Change type to scalar_int_mode.
12274 (extraction_insn::pos_mode): Likewise.
12275 * combine.c (make_extraction): Update accordingly.
12276 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12277 (get_optab_extraction_insn): Likewise.
12278 * recog.c (simplify_while_replacing): Likewise.
12279 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12280 parameter to opt_scalar_int_mode.
12282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12283 Alan Hayward <alan.hayward@arm.com>
12284 David Sherwood <david.sherwood@arm.com>
12286 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12287 to a scalar_int_mode instead of a machine_mode.
12288 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12289 (get_best_mode): Return a boolean and use a pointer argument to store
12290 the selected mode. Replace the limit mode parameter with a bit limit.
12291 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12292 for the values returned by bit_field_mode_iterator::next_mode.
12293 (store_bit_field): Update call to get_best_mode.
12294 (store_fixed_bit_field): Likewise.
12295 (extract_fixed_bit_field): Likewise.
12296 * expr.c (optimize_bitfield_assignment_op): Likewise.
12297 * fold-const.c (optimize_bit_field_compare): Likewise.
12298 (fold_truth_andor_1): Likewise.
12299 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12300 Update for new type of m_mode.
12301 (get_best_mode): As above.
12303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12304 Alan Hayward <alan.hayward@arm.com>
12305 David Sherwood <david.sherwood@arm.com>
12307 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12308 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12309 (store_bit_field): Check is_a <scalar_int_mode> before calling
12310 strict_volatile_bitfield_p.
12311 (extract_bit_field): Likewise.
12313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12314 Alan Hayward <alan.hayward@arm.com>
12315 David Sherwood <david.sherwood@arm.com>
12317 * target.def (cstore_mode): Return a scalar_int_mode.
12318 * doc/tm.texi: Regenerate.
12319 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12320 * targhooks.h (default_cstore_mode): Likewise.
12321 * targhooks.c (default_cstore_mode): Likewise, using a forced
12323 * expmed.c (emit_cstore): Expect the target of the cstore to be
12326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12327 Alan Hayward <alan.hayward@arm.com>
12328 David Sherwood <david.sherwood@arm.com>
12330 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12332 (niter_desc): Likewise mode.
12333 (iv_analyze): Add a mode parameter.
12335 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12336 and change its type to scalar_int_mode.
12337 * loop-iv.c: Update commentary at head of file.
12338 (iv_constant): Pass the mode paraeter before the rtx it describes
12339 and change its type to scalar_int_mode. Remove VOIDmode handling.
12340 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12341 (iv_extend): Likewise.
12342 (shorten_into_mode): Likewise.
12343 (iv_add): Use scalar_int_mode.
12344 (iv_mult): Likewise.
12345 (iv_shift): Likewise.
12346 (canonicalize_iv_subregs): Likewise.
12347 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12348 it describes and change its mode to scalar_int_mode. Also change
12349 the type of the returned inner_mode to scalar_int_mode.
12350 (get_biv_step): Likewise, turning outer_mode from a pointer
12351 into a direct parameter. Update call to get_biv_step_1.
12352 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12353 iv_constant and get_biv_step.
12354 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12355 and change its type to scalar_int_mode. Don't initialise iv->mode
12356 to VOIDmode and remove later checks for its still being VOIDmode.
12357 Update calls to iv_analyze_op and iv_analyze_expr. Check
12358 is_a <scalar_int_mode> when changing the mode under consideration.
12359 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12360 Update call to iv_analyze_expr.
12361 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12362 inner register is not also a scalar_int_mode. Update call to
12364 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12365 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12366 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12367 separate mode class checks. Update calls to iv_analyze. Remove
12368 fix-up of VOIDmodes after iv_analyze_biv.
12369 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12370 don't have a scalar_int_mode. Update call to biv_p.
12372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12373 Alan Hayward <alan.hayward@arm.com>
12374 David Sherwood <david.sherwood@arm.com>
12376 * cfgexpand.c (convert_debug_memory_address): Use
12377 as_a <scalar_int_mode>.
12378 * combine.c (expand_compound_operation): Likewise.
12379 (make_extraction): Likewise.
12380 (change_zero_ext): Likewise.
12381 (simplify_comparison): Likewise.
12382 * cse.c (cse_insn): Likewise.
12383 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12384 (mem_loc_descriptor): Likewise.
12385 (loc_descriptor): Likewise.
12386 * expmed.c (init_expmed_one_mode): Likewise.
12387 (synth_mult): Likewise.
12388 (emit_store_flag_1): Likewise.
12389 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12390 of a comparison with size.
12391 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12392 (reduce_to_bit_field_precision): Likewise.
12393 * function.c (expand_function_end): Likewise.
12394 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12395 * loop-doloop.c (doloop_modify): Likewise.
12396 * optabs.c (expand_binop): Likewise.
12397 (expand_unop): Likewise.
12398 (expand_copysign_absneg): Likewise.
12399 (prepare_cmp_insn): Likewise.
12400 (maybe_legitimize_operand): Likewise.
12401 * recog.c (const_scalar_int_operand): Likewise.
12402 * rtlanal.c (get_address_mode): Likewise.
12403 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12404 (simplify_cond_clz_ctz): Likewise.
12405 * tree-nested.c (get_nl_goto_field): Likewise.
12406 * tree.c (build_vector_type_for_mode): Likewise.
12407 * var-tracking.c (use_narrower_mode): Likewise.
12409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12410 Alan Hayward <alan.hayward@arm.com>
12411 David Sherwood <david.sherwood@arm.com>
12413 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12414 * builtins.c (expand_builtin_signbit): Use it.
12415 * cfgexpand.c (expand_debug_expr): Likewise.
12416 * dojump.c (do_jump): Likewise.
12417 (do_compare_and_jump): Likewise.
12418 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12419 * expmed.c (make_tree): Likewise.
12420 * expr.c (expand_expr_real_2): Likewise.
12421 (expand_expr_real_1): Likewise.
12422 (try_casesi): Likewise.
12423 * fold-const-call.c (fold_const_call_ss): Likewise.
12424 * fold-const.c (unextend): Likewise.
12425 (extract_muldiv_1): Likewise.
12426 (fold_single_bit_test): Likewise.
12427 (native_encode_int): Likewise.
12428 (native_encode_string): Likewise.
12429 (native_interpret_int): Likewise.
12430 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12431 * internal-fn.c (expand_addsub_overflow): Likewise.
12432 (expand_neg_overflow): Likewise.
12433 (expand_mul_overflow): Likewise.
12434 (expand_arith_overflow): Likewise.
12435 * match.pd: Likewise.
12436 * stor-layout.c (layout_type): Likewise.
12437 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12438 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12439 * tree-ssanames.c (get_range_info): Likewise.
12440 * tree-switch-conversion.c (array_value_type) Likewise.
12441 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12442 (vect_recog_divmod_pattern): Likewise.
12443 (vect_recog_mixed_size_cond_pattern): Likewise.
12444 * tree-vrp.c (extract_range_basic): Likewise.
12445 (simplify_float_conversion_using_ranges): Likewise.
12446 * tree.c (int_fits_type_p): Likewise.
12447 * ubsan.c (instrument_bool_enum_load): Likewise.
12448 * varasm.c (mergeable_string_section): Likewise.
12449 (narrowing_initializer_constant_valid_p): Likewise.
12450 (output_constant): Likewise.
12452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12453 Alan Hayward <alan.hayward@arm.com>
12454 David Sherwood <david.sherwood@arm.com>
12456 * machmode.h (NARROWEST_INT_MODE): New macro.
12457 * expr.c (alignment_for_piecewise_move): Use it instead of
12458 GET_CLASS_NARROWEST_MODE (MODE_INT).
12459 (push_block): Likewise.
12460 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12462 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12465 Alan Hayward <alan.hayward@arm.com>
12466 David Sherwood <david.sherwood@arm.com>
12468 * postreload.c (move2add_valid_value_p): Change the type of the
12469 mode parameter to scalar_int_mode.
12470 (move2add_use_add2_insn): Add a mode parameter and use it instead
12472 (move2add_use_add3_insn): Likewise.
12473 (reload_cse_move2add): Update accordingly.
12475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12476 Alan Hayward <alan.hayward@arm.com>
12477 David Sherwood <david.sherwood@arm.com>
12479 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12481 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12482 * optabs.c (expand_unop): Likewise.
12484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12485 Alan Hayward <alan.hayward@arm.com>
12486 David Sherwood <david.sherwood@arm.com>
12488 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12489 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12490 (popcount_loc_descriptor): Likewise.
12491 (bswap_loc_descriptor): Likewise.
12492 (rotate_loc_descriptor): Likewise.
12493 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12494 calling the functions above.
12496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12497 Alan Hayward <alan.hayward@arm.com>
12498 David Sherwood <david.sherwood@arm.com>
12500 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12502 (try_combine): Likewise.
12503 (simplify_if_then_else): Likewise.
12504 * cse.c (cse_insn): Likewise.
12505 * dwarf2out.c (mem_loc_descriptor): Likewise.
12506 * emit-rtl.c (gen_lowpart_common): Likewise.
12507 * simplify-rtx.c (simplify_truncation): Likewise.
12508 (simplify_binary_operation_1): Likewise.
12509 (simplify_const_relational_operation): Likewise.
12510 (simplify_ternary_operation): Likewise.
12511 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12514 Alan Hayward <alan.hayward@arm.com>
12515 David Sherwood <david.sherwood@arm.com>
12517 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12518 * reload.c (push_reload): Likewise.
12519 (find_reloads): Likewise.
12521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12522 Alan Hayward <alan.hayward@arm.com>
12523 David Sherwood <david.sherwood@arm.com>
12525 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12526 (make_compound_operation_int): Likewise.
12527 (change_zero_ext): Likewise.
12528 * expr.c (convert_move): Likewise.
12529 (convert_modes): Likewise.
12530 * fwprop.c (forward_propagate_subreg): Likewise.
12531 * loop-iv.c (get_biv_step_1): Likewise.
12532 * optabs.c (widen_operand): Likewise.
12533 * postreload.c (move2add_valid_value_p): Likewise.
12534 * recog.c (simplify_while_replacing): Likewise.
12535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12536 (simplify_binary_operation_1): Likewise. Remove redundant
12537 mode equality check.
12539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12540 Alan Hayward <alan.hayward@arm.com>
12541 David Sherwood <david.sherwood@arm.com>
12543 * combine.c (combine_simplify_rtx): Add checks for
12544 is_a <scalar_int_mode>.
12545 (simplify_if_then_else): Likewise.
12546 (make_field_assignment): Likewise.
12547 (simplify_comparison): Likewise.
12548 * ifcvt.c (noce_try_bitop): Likewise.
12549 * loop-invariant.c (canonicalize_address_mult): Likewise.
12550 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12553 Alan Hayward <alan.hayward@arm.com>
12554 David Sherwood <david.sherwood@arm.com>
12556 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12557 is_a <scalar_int_mode> instead of != BLKmode.
12559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12560 Alan Hayward <alan.hayward@arm.com>
12561 David Sherwood <david.sherwood@arm.com>
12563 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12564 instead of != VOIDmode.
12565 * combine.c (if_then_else_cond): Likewise.
12566 (change_zero_ext): Likewise.
12567 * dwarf2out.c (mem_loc_descriptor): Likewise.
12568 (loc_descriptor): Likewise.
12569 * rtlanal.c (canonicalize_condition): Likewise.
12570 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12573 Alan Hayward <alan.hayward@arm.com>
12574 David Sherwood <david.sherwood@arm.com>
12576 * simplify-rtx.c (simplify_binary_operation_1): Use
12577 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12580 Alan Hayward <alan.hayward@arm.com>
12581 David Sherwood <david.sherwood@arm.com>
12583 * wide-int.h (int_traits<unsigned char>) New class.
12584 (int_traits<unsigned short>) Likewise.
12585 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12586 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12588 * combine.c (set_nonzero_bits_and_sign_copies): Use
12589 is_a <scalar_int_mode>.
12590 (find_split_point): Likewise.
12591 (combine_simplify_rtx): Likewise.
12592 (simplify_logical): Likewise.
12593 (expand_compound_operation): Likewise.
12594 (expand_field_assignment): Likewise.
12595 (make_compound_operation): Likewise.
12596 (extended_count): Likewise.
12597 (change_zero_ext): Likewise.
12598 (simplify_comparison): Likewise.
12599 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12600 (ucompare_loc_descriptor): Likewise.
12601 (minmax_loc_descriptor): Likewise.
12602 (mem_loc_descriptor): Likewise.
12603 (loc_descriptor): Likewise.
12604 * expmed.c (init_expmed_one_mode): Likewise.
12605 * lra-constraints.c (lra_constraint_offset): Likewise.
12606 * optabs.c (prepare_libcall_arg): Likewise.
12607 * postreload.c (move2add_note_store): Likewise.
12608 * reload.c (operands_match_p): Likewise.
12609 * rtl.h (load_extend_op): Likewise.
12610 * rtlhooks.c (gen_lowpart_general): Likewise.
12611 * simplify-rtx.c (simplify_truncation): Likewise.
12612 (simplify_unary_operation_1): Likewise.
12613 (simplify_binary_operation_1): Likewise.
12614 (simplify_const_binary_operation): Likewise.
12615 (simplify_const_relational_operation): Likewise.
12616 (simplify_subreg): Likewise.
12617 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12618 * var-tracking.c (adjust_mems): Likewise.
12619 (prepare_call_arguments): Likewise.
12621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12622 Alan Hayward <alan.hayward@arm.com>
12623 David Sherwood <david.sherwood@arm.com>
12625 * machmode.h (is_int_mode): New fuction.
12626 * combine.c (find_split_point): Use it.
12627 (combine_simplify_rtx): Likewise.
12628 (simplify_if_then_else): Likewise.
12629 (simplify_set): Likewise.
12630 (simplify_shift_const_1): Likewise.
12631 (simplify_comparison): Likewise.
12632 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12633 * cse.c (notreg_cost): Likewise.
12634 (cse_insn): Likewise.
12635 * cselib.c (cselib_lookup_1): Likewise.
12636 * dojump.c (do_jump_1): Likewise.
12637 (do_compare_rtx_and_jump): Likewise.
12638 * dse.c (get_call_args): Likewise.
12639 * dwarf2out.c (rtl_for_decl_init): Likewise.
12640 (native_encode_initializer): Likewise.
12641 * expmed.c (emit_store_flag_1): Likewise.
12642 (emit_store_flag): Likewise.
12643 * expr.c (convert_modes): Likewise.
12644 (store_field): Likewise.
12645 (expand_expr_real_1): Likewise.
12646 * fold-const.c (fold_read_from_constant_string): Likewise.
12647 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12648 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12649 * optabs.c (expand_binop): Likewise.
12650 (expand_unop): Likewise.
12651 (expand_abs_nojump): Likewise.
12652 (expand_one_cmpl_abs_nojump): Likewise.
12653 * simplify-rtx.c (mode_signbit_p): Likewise.
12654 (val_signbit_p): Likewise.
12655 (val_signbit_known_set_p): Likewise.
12656 (val_signbit_known_clear_p): Likewise.
12657 (simplify_relational_operation_1): Likewise.
12658 * tree.c (vector_type_mode): Likewise.
12660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12661 Alan Hayward <alan.hayward@arm.com>
12662 David Sherwood <david.sherwood@arm.com>
12664 * machmode.h (smallest_mode_for_size): Fix formatting.
12665 (smallest_int_mode_for_size): New function.
12666 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12667 instead of smallest_mode_for_size.
12668 * combine.c (make_extraction): Likewise.
12669 * config/arc/arc.c (arc_expand_movmem): Likewise.
12670 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12671 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12672 * config/s390/s390.c (s390_expand_insv): Likewise.
12673 * config/sparc/sparc.c (assign_int_registers): Likewise.
12674 * config/spu/spu.c (spu_function_value): Likewise.
12675 (spu_function_arg): Likewise.
12676 * coverage.c (get_gcov_type): Likewise.
12677 (get_gcov_unsigned_t): Likewise.
12678 * dse.c (find_shift_sequence): Likewise.
12679 * expmed.c (store_bit_field_1): Likewise.
12680 * expr.c (convert_move): Likewise.
12681 (store_field): Likewise.
12682 * internal-fn.c (expand_arith_overflow): Likewise.
12683 * optabs-query.c (get_best_extraction_insn): Likewise.
12684 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12685 * stor-layout.c (layout_type): Likewise.
12686 (initialize_sizetypes): Likewise.
12687 * targhooks.c (default_get_mask_mode): Likewise.
12688 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12691 Alan Hayward <alan.hayward@arm.com>
12692 David Sherwood <david.sherwood@arm.com>
12694 * machmode.h (opt_mode::else_blk): New function.
12695 (int_mode_for_mode): Declare.
12696 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12697 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12699 * cfgexpand.c (expand_debug_expr): Likewise.
12700 * combine.c (gen_lowpart_or_truncate): Likewise.
12701 (gen_lowpart_for_combine): Likewise.
12702 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12703 * config/avr/avr.c (avr_to_int_mode): Likewise.
12704 (avr_out_plus_1): Likewise.
12705 (avr_out_plus): Likewise.
12706 (avr_out_round): Likewise.
12707 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12708 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12709 (s390_expand_vcond): Likewise.
12710 * config/spu/spu.c (spu_split_immediate): Likewise.
12711 (spu_expand_mov): Likewise.
12712 * dse.c (get_stored_val): Likewise.
12713 * expmed.c (store_bit_field_1): Likewise.
12714 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12716 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12717 (extract_low_bits): Likewise.
12718 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12719 handling rather than repeating the check.
12720 (emit_group_store): Likewise.
12721 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12722 * optabs.c (expand_absneg_bit): Likewise.
12723 (expand_copysign_absneg): Likewise.
12724 (expand_copysign_bit): Likewise.
12725 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12726 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12727 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12728 * var-tracking.c (prepare_call_arguments): Likewise.
12729 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12730 int_mode_for_mode instead of mode_for_size.
12731 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12734 Alan Hayward <alan.hayward@arm.com>
12735 David Sherwood <david.sherwood@arm.com>
12737 * machmode.h (int_mode_for_size): New function.
12738 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12739 instead of mode_for_size.
12740 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12742 * combine.c (expand_field_assignment): Use int_mode_for_size
12743 instead of mode_for_size.
12744 (make_extraction): Likewise.
12745 (simplify_shift_const_1): Likewise.
12746 (simplify_comparison): Likewise.
12747 * dojump.c (do_jump): Likewise.
12748 * dwarf2out.c (mem_loc_descriptor): Likewise.
12749 * emit-rtl.c (init_derived_machine_modes): Likewise.
12750 * expmed.c (flip_storage_order): Likewise.
12751 (convert_extracted_bit_field): Likewise.
12752 * expr.c (copy_blkmode_from_reg): Likewise.
12753 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12754 * internal-fn.c (expand_mul_overflow): Likewise.
12755 * lower-subreg.c (simple_move): Likewise.
12756 * optabs-libfuncs.c (init_optabs): Likewise.
12757 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12758 * tree.c (vector_type_mode): Likewise.
12759 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12760 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12761 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12762 * tree-vect-stmts.c (vectorizable_load): Likewise.
12763 (vectorizable_store): Likewise.
12765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12766 Alan Hayward <alan.hayward@arm.com>
12767 David Sherwood <david.sherwood@arm.com>
12769 * coretypes.h (pod_mode): New type.
12770 (scalar_int_mode_pod): New typedef.
12771 * machmode.h (pod_mode): New class.
12772 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12773 * genmodes.c (emit_mode_int_n): Update accordingly.
12774 * lower-subreg.h (target_lower_subreg): Change type to
12775 scalar_int_mode_pod.
12776 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12777 scalar_int_mode_pod.
12779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12780 Alan Hayward <alan.hayward@arm.com>
12781 David Sherwood <david.sherwood@arm.com>
12783 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12784 machine_mode to scalar_int_mode.
12785 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12786 (rs6000_option_override_internal): Remove cast to int.
12787 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12788 machine_mode to scalar_int_mode.
12789 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12790 (rs6000_option_override_internal): Remove cast to int.
12791 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12792 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12794 * config/s390/s390.c (s390_expand_builtin): Likewise.
12795 * coretypes.h (scalar_int_mode): New type.
12796 (opt_scalar_int_mode): New typedef.
12797 * machmode.h (scalar_int_mode): New class.
12798 (scalar_int_mode::includes_p): New function.
12799 (byte_mode): Change type to scalar_int_mode.
12800 (word_mode): Likewise.
12801 (ptr_mode): Likewise.
12802 * emit-rtl.c (byte_mode): Likewise.
12803 (word_mode): Likewise.
12804 (ptr_mode): Likewise.
12805 (init_derived_machine_modes): Update accordingly.
12806 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12807 and MODE_PARTIAL_INT.
12808 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12809 opt_scalar_int_mode.
12811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12812 Alan Hayward <alan.hayward@arm.com>
12813 David Sherwood <david.sherwood@arm.com>
12815 * target.def (libgcc_floating_mode_supported_p): Take a
12817 * doc/tm.texi: Regenerate.
12818 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12820 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12821 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12825 Alan Hayward <alan.hayward@arm.com>
12826 David Sherwood <david.sherwood@arm.com>
12828 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12829 * doc/tm.texi: Regenerate.
12830 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12831 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12832 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12833 * targhooks.h (default_floatn_mode): Likewise.
12834 * targhooks.c (default_floatn_mode): Likewise.
12835 * tree.c (build_common_tree_nodes): Update accordingly.
12837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12838 Alan Hayward <alan.hayward@arm.com>
12839 David Sherwood <david.sherwood@arm.com>
12841 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12842 (mode_iterator::iterate_p): Likewise.
12843 (mode_iterator::get_wider): Likewise.
12844 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12847 Alan Hayward <alan.hayward@arm.com>
12848 David Sherwood <david.sherwood@arm.com>
12850 * coretypes.h (opt_scalar_float_mode): New typedef.
12851 * machmode.h (float_mode_for_size): New function.
12852 * emit-rtl.c (double_mode): Delete.
12853 (init_emit_once): Use float_mode_for_size.
12854 * stor-layout.c (layout_type): Likewise.
12855 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12858 Alan Hayward <alan.hayward@arm.com>
12859 David Sherwood <david.sherwood@arm.com>
12861 * output.h (assemble_real): Take a scalar_float_mode.
12862 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12863 * config/arm/arm.md (consttable_4): Likewise.
12864 (consttable_8): Likewise.
12865 (consttable_16): Likewise.
12866 * config/mips/mips.md (consttable_float): Likewise.
12867 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12868 * varasm.c (assemble_real): Take a scalar_float_mode.
12869 (output_constant_pool_2): Update accordingly.
12870 (output_constant): Likewise.
12872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12873 Alan Hayward <alan.hayward@arm.com>
12874 David Sherwood <david.sherwood@arm.com>
12876 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12877 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12878 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12879 (native_encode_real): Likewise.
12880 (native_interpret_real): Likewise.
12881 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12882 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12885 Alan Hayward <alan.hayward@arm.com>
12886 David Sherwood <david.sherwood@arm.com>
12888 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12889 <scalar_float_mode>. Simplify.
12890 (gen_extend_conv_libfunc): Likewise.
12892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12893 Alan Hayward <alan.hayward@arm.com>
12894 David Sherwood <david.sherwood@arm.com>
12896 * coretypes.h (scalar_float_mode): New type.
12897 * machmode.h (mode_traits::from_int): Use machine_mode if
12898 USE_ENUM_MODES is defined.
12899 (is_a): New function.
12901 (dyn_cast): Likewise.
12902 (scalar_float_mode): New class.
12903 (scalar_float_mode::includes_p): New function.
12904 (is_float_mode): Likewise.
12905 * gdbhooks.py (MachineModePrinter): New class.
12906 (build_pretty_printer): Use it for scalar_float_mode.
12907 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12908 (format_helper::format_helper): Turn into a template.
12909 * genmodes.c (get_mode_class): New function.
12910 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12911 or machine_mode if none.
12912 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12913 as_a <scalar_float_mode>.
12914 * dwarf2out.c (mem_loc_descriptor): Likewise.
12915 (insert_float): Likewise.
12916 (add_const_value_attribute): Likewise.
12917 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12918 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12919 (expand_unop): Update accordingly.
12920 (expand_abs_nojump): Likewise.
12921 (expand_copysign_absneg): Take a scalar_float_mode.
12922 (expand_copysign_bit): Likewise.
12923 (expand_copysign): Update accordingly.
12925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12926 Alan Hayward <alan.hayward@arm.com>
12927 David Sherwood <david.sherwood@arm.com>
12929 * coretypes.h (opt_mode): New class.
12930 * machmode.h (opt_mode): Likewise.
12931 (opt_mode::else_void): New function.
12932 (opt_mode::require): Likewise.
12933 (opt_mode::exists): Likewise.
12934 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12935 (GET_MODE_2XWIDER_MODE): Likewise.
12936 (mode_iterator::get_wider): Update accordingly.
12937 (mode_iterator::get_2xwider): Likewise.
12938 (mode_iterator::get_known_wider): Likewise, turning into a template.
12939 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12940 forcing a wider mode to exist.
12941 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12942 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12943 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12944 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12945 * lower-subreg.c (init_lower_subreg): Likewise.
12946 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12947 on the final iteration.
12948 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12949 a wider mode exists before asking for a move pattern.
12950 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12951 forcing a wider mode to exist.
12952 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12953 returning false if no such mode exists.
12954 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12955 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12956 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12957 Avoid checking for a MODE_INT if we already know the mode is not a
12959 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12960 forcing a wider mode to exist.
12961 (expmed_mult_highpart_optab): Likewise.
12962 (expmed_mult_highpart): Likewise.
12963 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12965 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12966 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12967 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12968 * internal-fn.c (expand_mul_overflow): Update use of
12969 GET_MODE_2XWIDER_MODE.
12970 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12971 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12972 GET_MODE_WIDER_MODE.
12973 (convert_plusminus_to_widen): Likewise.
12974 * tree-switch-conversion.c (array_value_type): Likewise.
12975 * var-tracking.c (emit_note_insn_var_location): Likewise.
12976 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12977 Return false inside rather than outside the loop if no wider mode
12979 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12980 and GET_MODE_2XWIDER_MODE
12981 (can_compare_p): Use else_void.
12982 * gdbhooks.py (OptMachineModePrinter): New class.
12983 (build_pretty_printer): Use it for opt_mode.
12985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12986 Alan Hayward <alan.hayward@arm.com>
12987 David Sherwood <david.sherwood@arm.com>
12989 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12990 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12993 Alan Hayward <alan.hayward@arm.com>
12994 David Sherwood <david.sherwood@arm.com>
12996 * machmode.h (mode_traits): New structure.
12997 (get_narrowest_mode): New function.
12998 (mode_iterator::start): Likewise.
12999 (mode_iterator::iterate_p): Likewise.
13000 (mode_iterator::get_wider): Likewise.
13001 (mode_iterator::get_known_wider): Likewise.
13002 (mode_iterator::get_2xwider): Likewise.
13003 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13004 (FOR_EACH_MODE): Likewise.
13005 (FOR_EACH_MODE_FROM): Likewise.
13006 (FOR_EACH_MODE_UNTIL): Likewise.
13007 (FOR_EACH_WIDER_MODE): Likewise.
13008 (FOR_EACH_2XWIDER_MODE): Likewise.
13009 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13010 * combine.c (simplify_comparison): Likewise
13011 * config/i386/i386.c (type_natural_mode): Likewise.
13012 * cse.c (cse_insn): Likewise.
13013 * dse.c (find_shift_sequence): Likewise.
13014 * emit-rtl.c (init_derived_machine_modes): Likewise.
13015 (init_emit_once): Likewise.
13016 * explow.c (hard_function_value): Likewise.
13017 * expmed.c (extract_fixed_bit_field_1): Likewise.
13018 (extract_bit_field_1): Likewise.
13019 (expand_divmod): Likewise.
13020 (emit_store_flag_1): Likewise.
13021 * expr.c (init_expr_target): Likewise.
13022 (convert_move): Likewise.
13023 (alignment_for_piecewise_move): Likewise.
13024 (widest_int_mode_for_size): Likewise.
13025 (emit_block_move_via_movmem): Likewise.
13026 (copy_blkmode_to_reg): Likewise.
13027 (set_storage_via_setmem): Likewise.
13028 (compress_float_constant): Likewise.
13029 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13030 * optabs-query.c (get_best_extraction_insn): Likewise.
13031 * optabs.c (expand_binop): Likewise.
13032 (expand_twoval_unop): Likewise.
13033 (expand_twoval_binop): Likewise.
13034 (widen_leading): Likewise.
13035 (widen_bswap): Likewise.
13036 (expand_parity): Likewise.
13037 (expand_unop): Likewise.
13038 (prepare_cmp_insn): Likewise.
13039 (prepare_float_lib_cmp): Likewise.
13040 (expand_float): Likewise.
13041 (expand_fix): Likewise.
13042 (expand_sfix_optab): Likewise.
13043 * postreload.c (move2add_use_add2_insn): Likewise.
13044 * reg-stack.c (reg_to_stack): Likewise.
13045 * reginfo.c (choose_hard_reg_mode): Likewise.
13046 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13047 * stor-layout.c (mode_for_size): Likewise.
13048 (smallest_mode_for_size): Likewise.
13049 (mode_for_vector): Likewise.
13050 (finish_bitfield_representative): Likewise.
13051 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13052 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13053 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13054 * var-tracking.c (prepare_call_arguments): Likewise.
13056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13057 Alan Hayward <alan.hayward@arm.com>
13058 David Sherwood <david.sherwood@arm.com>
13060 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13061 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13062 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13063 * machmode.h (mode_size): Move earlier in file.
13064 (mode_precision): Likewise.
13065 (mode_inner): Likewise.
13066 (mode_nunits): Likewise.
13067 (mode_unit_size): Likewise.
13068 (unit_unit_precision): Likewise.
13069 (mode_wider): Likewise.
13070 (mode_2xwider): Likewise.
13071 (machine_mode): New class.
13072 (mode_to_bytes): New function.
13073 (mode_to_bits): Likewise.
13074 (mode_to_precision): Likewise.
13075 (mode_to_inner): Likewise.
13076 (mode_to_unit_size): Likewise.
13077 (mode_to_unit_precision): Likewise.
13078 (mode_to_nunits): Likewise.
13079 (GET_MODE_SIZE): Use mode_to_bytes.
13080 (GET_MODE_BITSIZE): Use mode_to_bits.
13081 (GET_MODE_PRECISION): Use mode_to_precision.
13082 (GET_MODE_INNER): Use mode_to_inner.
13083 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13084 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13085 (GET_MODE_NUNITS): Use mode_to_nunits.
13086 * system.h (ALWAYS_INLINE): New macro.
13087 * config/powerpcspe/powerpcspe-c.c
13088 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13089 int for arg1_mode and arg2_mode.
13091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13092 Alan Hayward <alan.hayward@arm.com>
13093 David Sherwood <david.sherwood@arm.com>
13095 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13096 Prefix mode names with E_ in case statements.
13097 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13098 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13099 (aarch64_split_simd_move): Likewise.
13100 (aarch64_gen_storewb_pair): Likewise.
13101 (aarch64_gen_loadwb_pair): Likewise.
13102 (aarch64_gen_store_pair): Likewise.
13103 (aarch64_gen_load_pair): Likewise.
13104 (aarch64_get_condition_code_1): Likewise.
13105 (aarch64_constant_pool_reload_icode): Likewise.
13106 (get_rsqrte_type): Likewise.
13107 (get_rsqrts_type): Likewise.
13108 (get_recpe_type): Likewise.
13109 (get_recps_type): Likewise.
13110 (aarch64_gimplify_va_arg_expr): Likewise.
13111 (aarch64_simd_container_mode): Likewise.
13112 (aarch64_emit_load_exclusive): Likewise.
13113 (aarch64_emit_store_exclusive): Likewise.
13114 (aarch64_expand_compare_and_swap): Likewise.
13115 (aarch64_gen_atomic_cas): Likewise.
13116 (aarch64_emit_bic): Likewise.
13117 (aarch64_emit_atomic_swap): Likewise.
13118 (aarch64_emit_atomic_load_op): Likewise.
13119 (aarch64_evpc_trn): Likewise.
13120 (aarch64_evpc_uzp): Likewise.
13121 (aarch64_evpc_zip): Likewise.
13122 (aarch64_evpc_ext): Likewise.
13123 (aarch64_evpc_rev): Likewise.
13124 (aarch64_evpc_dup): Likewise.
13125 (aarch64_gen_ccmp_first): Likewise.
13126 (aarch64_gen_ccmp_next): Likewise.
13127 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13128 (alpha_emit_xfloating_libcall): Likewise.
13129 (emit_insxl): Likewise.
13130 (alpha_arg_type): Likewise.
13131 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13132 (arc_preferred_simd_mode): Likewise.
13133 (arc_secondary_reload): Likewise.
13134 (get_arc_condition_code): Likewise.
13135 (arc_print_operand): Likewise.
13136 (arc_legitimate_constant_p): Likewise.
13137 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13138 * config/arc/arc.md (casesi_load): Likewise.
13139 (casesi_compact_jump): Likewise.
13140 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13141 (cc_use_register): Likewise.
13142 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13143 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13144 (arm_init_iwmmxt_builtins): Likewise.
13145 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13146 (neon_expand_vector_init): Likewise.
13147 (arm_attr_length_move_neon): Likewise.
13148 (maybe_get_arm_condition_code): Likewise.
13149 (arm_emit_vector_const): Likewise.
13150 (arm_preferred_simd_mode): Likewise.
13151 (arm_output_iwmmxt_tinsr): Likewise.
13152 (thumb1_output_casesi): Likewise.
13153 (thumb2_output_casesi): Likewise.
13154 (arm_emit_load_exclusive): Likewise.
13155 (arm_emit_store_exclusive): Likewise.
13156 (arm_expand_compare_and_swap): Likewise.
13157 (arm_evpc_neon_vuzp): Likewise.
13158 (arm_evpc_neon_vzip): Likewise.
13159 (arm_evpc_neon_vrev): Likewise.
13160 (arm_evpc_neon_vtrn): Likewise.
13161 (arm_evpc_neon_vext): Likewise.
13162 (arm_validize_comparison): Likewise.
13163 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13164 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13165 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13166 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13167 (c6x_preferred_simd_mode): Likewise.
13168 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13169 (epiphany_rtx_costs): Likewise.
13170 * config/epiphany/predicates.md (proper_comparison_operator):
13172 * config/frv/frv.c (condexec_memory_operand): Likewise.
13173 (frv_emit_move): Likewise.
13174 (output_move_single): Likewise.
13175 (output_condmove_single): Likewise.
13176 (frv_hard_regno_mode_ok): Likewise.
13177 (frv_matching_accg_mode): Likewise.
13178 * config/h8300/h8300.c (split_adds_subs): Likewise.
13179 (h8300_rtx_costs): Likewise.
13180 (h8300_print_operand): Likewise.
13181 (compute_mov_length): Likewise.
13182 (output_logical_op): Likewise.
13183 (compute_logical_op_length): Likewise.
13184 (compute_logical_op_cc): Likewise.
13185 (h8300_shift_needs_scratch_p): Likewise.
13186 (output_a_shift): Likewise.
13187 (compute_a_shift_length): Likewise.
13188 (compute_a_shift_cc): Likewise.
13189 (expand_a_rotate): Likewise.
13190 (output_a_rotate): Likewise.
13191 * config/i386/i386.c (classify_argument): Likewise.
13192 (function_arg_advance_32): Likewise.
13193 (function_arg_32): Likewise.
13194 (function_arg_64): Likewise.
13195 (function_value_64): Likewise.
13196 (ix86_gimplify_va_arg): Likewise.
13197 (ix86_legitimate_constant_p): Likewise.
13198 (put_condition_code): Likewise.
13199 (split_double_mode): Likewise.
13200 (ix86_avx256_split_vector_move_misalign): Likewise.
13201 (ix86_expand_vector_logical_operator): Likewise.
13202 (ix86_split_idivmod): Likewise.
13203 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13204 (ix86_build_const_vector): Likewise.
13205 (ix86_build_signbit_mask): Likewise.
13206 (ix86_match_ccmode): Likewise.
13207 (ix86_cc_modes_compatible): Likewise.
13208 (ix86_expand_branch): Likewise.
13209 (ix86_expand_sse_cmp): Likewise.
13210 (ix86_expand_sse_movcc): Likewise.
13211 (ix86_expand_int_sse_cmp): Likewise.
13212 (ix86_expand_vec_perm_vpermi2): Likewise.
13213 (ix86_expand_vec_perm): Likewise.
13214 (ix86_expand_sse_unpack): Likewise.
13215 (ix86_expand_int_addcc): Likewise.
13216 (ix86_split_to_parts): Likewise.
13217 (ix86_vectorize_builtin_gather): Likewise.
13218 (ix86_vectorize_builtin_scatter): Likewise.
13219 (avx_vpermilp_parallel): Likewise.
13220 (inline_memory_move_cost): Likewise.
13221 (ix86_tieable_integer_mode_p): Likewise.
13222 (x86_maybe_negate_const_int): Likewise.
13223 (ix86_expand_vector_init_duplicate): Likewise.
13224 (ix86_expand_vector_init_one_nonzero): Likewise.
13225 (ix86_expand_vector_init_one_var): Likewise.
13226 (ix86_expand_vector_init_concat): Likewise.
13227 (ix86_expand_vector_init_interleave): Likewise.
13228 (ix86_expand_vector_init_general): Likewise.
13229 (ix86_expand_vector_set): Likewise.
13230 (ix86_expand_vector_extract): Likewise.
13231 (emit_reduc_half): Likewise.
13232 (ix86_emit_i387_round): Likewise.
13233 (ix86_mangle_type): Likewise.
13234 (ix86_expand_round_sse4): Likewise.
13235 (expand_vec_perm_blend): Likewise.
13236 (canonicalize_vector_int_perm): Likewise.
13237 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13238 (expand_vec_perm_1): Likewise.
13239 (expand_vec_perm_interleave3): Likewise.
13240 (expand_vec_perm_even_odd_pack): Likewise.
13241 (expand_vec_perm_even_odd_1): Likewise.
13242 (expand_vec_perm_broadcast_1): Likewise.
13243 (ix86_vectorize_vec_perm_const_ok): Likewise.
13244 (ix86_expand_vecop_qihi): Likewise.
13245 (ix86_expand_mul_widen_hilo): Likewise.
13246 (ix86_expand_sse2_abs): Likewise.
13247 (ix86_expand_pextr): Likewise.
13248 (ix86_expand_pinsr): Likewise.
13249 (ix86_preferred_simd_mode): Likewise.
13250 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13251 * config/i386/sse.md (*andnot<mode>3): Likewise.
13252 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13253 (*<code><mode>3): Likewise.
13254 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13255 (ia64_expand_atomic_op): Likewise.
13256 (ia64_arg_type): Likewise.
13257 (ia64_mode_to_int): Likewise.
13258 (ia64_scalar_mode_supported_p): Likewise.
13259 (ia64_vector_mode_supported_p): Likewise.
13260 (expand_vec_perm_broadcast): Likewise.
13261 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13262 (iq2000_function_arg_advance): Likewise.
13263 (iq2000_function_arg): Likewise.
13264 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13265 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13266 (m68k_libcall_value): Likewise.
13267 (m68k_function_value): Likewise.
13268 (sched_attr_op_type): Likewise.
13269 * config/mcore/mcore.c (mcore_output_move): Likewise.
13270 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13272 (microblaze_function_arg): Likewise.
13273 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13274 (mips_print_operand): Likewise.
13275 (mips_mode_ok_for_mov_fmt_p): Likewise.
13276 (mips_vector_mode_supported_p): Likewise.
13277 (mips_preferred_simd_mode): Likewise.
13278 (mips_expand_vpc_loongson_even_odd): Likewise.
13279 (mips_expand_vec_unpack): Likewise.
13280 (mips_expand_vi_broadcast): Likewise.
13281 (mips_expand_vector_init): Likewise.
13282 (mips_expand_vec_reduc): Likewise.
13283 (mips_expand_msa_cmp): Likewise.
13284 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13285 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13286 (cc_flags_for_mode): Likewise.
13287 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13288 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13289 (nds32_output_casesi_pc_relative): Likewise.
13290 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13291 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13292 (nvptx_gen_unpack): Likewise.
13293 (nvptx_gen_pack): Likewise.
13294 (nvptx_gen_shuffle): Likewise.
13295 (nvptx_gen_wcast): Likewise.
13296 (nvptx_preferred_simd_mode): Likewise.
13297 * config/pa/pa.c (pa_secondary_reload): Likewise.
13298 * config/pa/predicates.md (base14_operand): Likewise.
13299 * config/powerpcspe/powerpcspe-c.c
13300 (altivec_resolve_overloaded_builtin): Likewise.
13301 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13303 (rs6000_preferred_simd_mode): Likewise.
13304 (output_vec_const_move): Likewise.
13305 (rs6000_expand_vector_extract): Likewise.
13306 (rs6000_split_vec_extract_var): Likewise.
13307 (reg_offset_addressing_ok_p): Likewise.
13308 (rs6000_legitimate_offset_address_p): Likewise.
13309 (rs6000_legitimize_address): Likewise.
13310 (rs6000_emit_set_const): Likewise.
13311 (rs6000_const_vec): Likewise.
13312 (rs6000_emit_move): Likewise.
13313 (spe_build_register_parallel): Likewise.
13314 (rs6000_darwin64_record_arg_recurse): Likewise.
13315 (swap_selector_for_mode): Likewise.
13316 (spe_init_builtins): Likewise.
13317 (paired_init_builtins): Likewise.
13318 (altivec_init_builtins): Likewise.
13319 (do_load_for_compare): Likewise.
13320 (rs6000_generate_compare): Likewise.
13321 (rs6000_expand_float128_convert): Likewise.
13322 (emit_load_locked): Likewise.
13323 (emit_store_conditional): Likewise.
13324 (rs6000_output_function_epilogue): Likewise.
13325 (rs6000_handle_altivec_attribute): Likewise.
13326 (rs6000_function_value): Likewise.
13327 (emit_fusion_gpr_load): Likewise.
13328 (emit_fusion_p9_load): Likewise.
13329 (emit_fusion_p9_store): Likewise.
13330 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13331 (fusion_gpr_mem_load): Likewise.
13332 (fusion_addis_mem_combo_load): Likewise.
13333 (fusion_addis_mem_combo_store): Likewise.
13334 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13335 (fusion_gpr_mem_load): Likewise.
13336 (fusion_addis_mem_combo_load): Likewise.
13337 (fusion_addis_mem_combo_store): Likewise.
13338 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13340 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13341 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13342 (rs6000_preferred_simd_mode): Likewise.
13343 (output_vec_const_move): Likewise.
13344 (rs6000_expand_vector_extract): Likewise.
13345 (rs6000_split_vec_extract_var): Likewise.
13346 (reg_offset_addressing_ok_p): Likewise.
13347 (rs6000_legitimate_offset_address_p): Likewise.
13348 (rs6000_legitimize_address): Likewise.
13349 (rs6000_emit_set_const): Likewise.
13350 (rs6000_const_vec): Likewise.
13351 (rs6000_emit_move): Likewise.
13352 (rs6000_darwin64_record_arg_recurse): Likewise.
13353 (swap_selector_for_mode): Likewise.
13354 (paired_init_builtins): Likewise.
13355 (altivec_init_builtins): Likewise.
13356 (rs6000_expand_float128_convert): Likewise.
13357 (emit_load_locked): Likewise.
13358 (emit_store_conditional): Likewise.
13359 (rs6000_output_function_epilogue): Likewise.
13360 (rs6000_handle_altivec_attribute): Likewise.
13361 (rs6000_function_value): Likewise.
13362 (emit_fusion_gpr_load): Likewise.
13363 (emit_fusion_p9_load): Likewise.
13364 (emit_fusion_p9_store): Likewise.
13365 * config/rx/rx.c (rx_gen_move_template): Likewise.
13366 (flags_from_mode): Likewise.
13367 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13368 (s390_slb_comparison): Likewise.
13369 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13370 (s390_vector_mode_supported_p): Likewise.
13371 (s390_cc_modes_compatible): Likewise.
13372 (s390_match_ccmode_set): Likewise.
13373 (s390_canonicalize_comparison): Likewise.
13374 (s390_emit_compare_and_swap): Likewise.
13375 (s390_branch_condition_mask): Likewise.
13376 (s390_rtx_costs): Likewise.
13377 (s390_secondary_reload): Likewise.
13378 (__SECONDARY_RELOAD_CASE): Likewise.
13379 (s390_expand_cs): Likewise.
13380 (s390_preferred_simd_mode): Likewise.
13381 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13382 * config/sh/sh.c (sh_print_operand): Likewise.
13383 (dump_table): Likewise.
13384 (sh_secondary_reload): Likewise.
13385 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13386 * config/sh/sh.md (casesi_worker_1): Likewise.
13387 (casesi_worker_2): Likewise.
13388 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13389 (fcc_comparison_operator): Likewise.
13390 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13391 (emit_soft_tfmode_cvt): Likewise.
13392 (sparc_preferred_simd_mode): Likewise.
13393 (output_cbranch): Likewise.
13394 (sparc_print_operand): Likewise.
13395 (sparc_expand_vec_perm_bmask): Likewise.
13396 (vector_init_bshuffle): Likewise.
13397 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13398 (spu_vector_mode_supported_p): Likewise.
13399 (spu_expand_insv): Likewise.
13400 (spu_emit_branch_or_set): Likewise.
13401 (spu_handle_vector_attribute): Likewise.
13402 (spu_builtin_splats): Likewise.
13403 (spu_builtin_extract): Likewise.
13404 (spu_builtin_promote): Likewise.
13405 (spu_expand_sign_extend): Likewise.
13406 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13407 (tilegx_simd_int): Likewise.
13408 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13409 (tilepro_simd_int): Likewise.
13410 * config/v850/v850.c (const_double_split): Likewise.
13411 (v850_print_operand): Likewise.
13412 (ep_memory_offset): Likewise.
13413 * config/vax/vax.c (vax_rtx_costs): Likewise.
13414 (vax_output_int_move): Likewise.
13415 (vax_output_int_add): Likewise.
13416 (vax_output_int_subtract): Likewise.
13417 * config/visium/predicates.md (visium_branch_operator): Likewise.
13418 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13419 (visium_print_operand_address): Likewise.
13420 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13421 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13422 (xtensa_expand_conditional_branch): Likewise.
13423 (xtensa_copy_incoming_a7): Likewise.
13424 (xtensa_output_literal): Likewise.
13425 * dfp.c (decimal_real_maxval): Likewise.
13426 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13429 Alan Hayward <alan.hayward@arm.com>
13430 David Sherwood <david.sherwood@arm.com>
13432 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13433 (mode_nunits_inline): Likewise.
13434 (mode_inner_inline): Likewise.
13435 (mode_unit_size_inline): Likewise.
13436 (mode_unit_precision_inline): Likewise.
13437 (emit_insn_modes_h): Likewise. Also emit a #define of the
13439 (emit_mode_wider): Add an E_ prefix to mode names.
13440 (emit_mode_complex): Likewise.
13441 (emit_mode_inner): Likewise.
13442 (emit_mode_adjustments): Likewise.
13443 (emit_mode_int_n): Likewise.
13444 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13445 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13446 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13447 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13448 (CRC32_BUILTIN, ENTRY): Likewise.
13449 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13450 (aarch64_pop_regs): Likewise.
13451 (aarch64_process_components): Likewise.
13452 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13453 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13454 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13455 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13456 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13457 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13459 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13460 (pcmpistr): Likewise.
13461 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13462 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13463 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13465 * config/rl78/rl78.c (mduc_regs): Likewise.
13466 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13467 (htm_expand_builtin): Likewise.
13468 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13469 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13470 * config/xtensa/xtensa.c (print_operand): Likewise.
13471 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13472 (NUM_MODE_VECTOR_INT): Likewise.
13473 * genoutput.c (null_operand): Likewise.
13474 (output_operand_data): Likewise.
13475 * genrecog.c (print_parameter_value): Likewise.
13476 * lra.c (debug_operand_data): Likewise.
13478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13480 * dwarf2out.c (scompare_loc_descriptor_wide)
13481 (scompare_loc_descriptor_narrow): New functions, split out from...
13482 (scompare_loc_descriptor): ...here.
13483 * expmed.c (emit_store_flag_int): New function, split out from...
13484 (emit_store_flag): ...here.
13486 2017-08-30 Richard Biener <rguenther@suse.de>
13488 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13489 (dwarf2out_early_finish): Move setting of AT_pubnames from
13490 early debug output to early finish.
13492 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13494 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13495 and -mdata-region to the assembler.
13497 2017-08-30 Richard Biener <rguenther@suse.de>
13499 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13501 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13502 (dwarf2out_early_global_decl): Only generate a DIE for the
13503 abstract origin if it doesn't already exist or is a declaration DIE.
13504 (resolve_addr): Do not add the linkage name twice when
13505 generating a stub DIE for the DW_TAG_GNU_call_site target.
13507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13509 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13510 Use machine_mode rather than int for arg1_mode.
13512 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13515 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13516 that the second argument of the built-in functions to unpack
13517 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13518 switch statement instead a lot of if statements.
13519 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13520 Allow 64-bit values to be in Altivec registers as well as
13521 traditional floating point registers.
13522 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13524 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13526 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13527 MAX_REGS_PER_ADDRESS == 1.
13529 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13531 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13532 * config/i386/i386.c (ix86_option_override_internal): Simplify
13533 setting of opts->x_flag_entry.
13535 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13536 Jakub Jelinek <jakub@redhat.com>
13537 Richard Biener <rguenther@suse.de>
13539 PR tree-optimization/81503
13540 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13541 folded constant fits in the target type; reorder tests for clarity.
13543 2017-08-29 Martin Liska <mliska@suse.cz>
13545 * passes.def: Include pass_lower_switch.
13546 * stmt.c (dump_case_nodes): Remove and move to
13547 tree-switch-conversion.
13548 (case_values_threshold): Likewise.
13549 (expand_switch_as_decision_tree_p): Likewise.
13550 (emit_case_decision_tree): Likewise.
13551 (expand_case): Likewise.
13552 (balance_case_nodes): Likewise.
13553 (node_has_low_bound): Likewise.
13554 (node_has_high_bound): Likewise.
13555 (node_is_bounded): Likewise.
13556 (emit_case_nodes): Likewise.
13557 (struct simple_case_node): New struct.
13558 (add_case_node): Remove.
13559 (emit_case_dispatch_table): Use vector instead of case_list.
13560 (reset_out_edges_aux): Remove.
13561 (compute_cases_per_edge): Likewise.
13562 (expand_case): Build list of simple_case_node.
13563 (expand_sjlj_dispatch_table): Use it.
13564 * tree-switch-conversion.c (struct case_node): Moved from
13565 stmt.c and adjusted.
13566 (emit_case_nodes): Likewise.
13567 (node_has_low_bound): Likewise.
13568 (node_has_high_bound): Likewise.
13569 (node_is_bounded): Likewise.
13570 (case_values_threshold): Likewise.
13571 (reset_out_edges_aux): Likewise.
13572 (compute_cases_per_edge): Likewise.
13573 (add_case_node): Likewise.
13574 (dump_case_nodes): Likewise.
13575 (balance_case_nodes): Likewise.
13576 (expand_switch_as_decision_tree_p): Likewise.
13577 (emit_jump): Likewise.
13578 (emit_case_decision_tree): Likewise.
13579 (try_switch_expansion): Likewise.
13580 (do_jump_if_equal): Likewise.
13581 (emit_cmp_and_jump_insns): Likewise.
13582 (fix_phi_operands_for_edge): New function.
13583 (record_phi_operand_mapping): Likewise.
13584 (class pass_lower_switch): New pass.
13585 (pass_lower_switch::execute): New function.
13586 (make_pass_lower_switch): Likewise.
13587 (conditional_probability):
13588 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13589 * tree-pass.h: Add make_pass_lower_switch.
13591 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13594 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13597 2017-08-29 Richard Biener <rguenther@suse.de>
13599 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13600 we do not add a DW_AT_inline attribute twice.
13601 (gen_subprogram_die): Remove code setting DW_AT_inline on
13602 DECL_ABSTRACT_P nodes.
13604 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13606 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13607 calls to internal functions.
13608 (gimplify_modify_expr): Likewise.
13609 * tree-call-cdce.c (use_internal_fn): Likewise.
13610 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13611 (convert_to_divmod): Set the nothrow flag.
13612 * tree-if-conv.c (predicate_mem_writes): Likewise.
13613 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13614 (vectorizable_call): Likewise.
13615 (vectorizable_store): Likewise.
13616 (vectorizable_load): Likewise.
13617 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13618 (vect_recog_mask_conversion_pattern): Likewise.
13620 2017-08-29 Martin Liska <mliska@suse.cz>
13623 * gcc.c (driver_handle_option): Add new argument.
13624 * opts-common.c (handle_option): Pass
13625 target_option_override_hook.
13626 * opts-global.c (lang_handle_option): Add new option.
13627 (set_default_handlers): Add new argument.
13628 (decode_options): Likewise.
13629 * opts.c (target_handle_option): Likewise.
13630 (common_handle_option): Call target_option_override_hook.
13631 * opts.h (struct cl_option_handler_func): Add hook for
13632 target option override.
13633 (struct cl_option_handlers): Likewise.
13634 (set_default_handlers): Add new argument.
13635 (decode_options): Likewise.
13636 (common_handle_option): Likewise.
13637 (target_handle_option): Likewise.
13638 * toplev.c (toplev::main): Pass targetm.target_option.override
13641 2017-08-29 Richard Biener <rguenther@suse.de>
13642 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13644 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13645 life to the active subtree.
13647 2017-08-28 Jeff Law <law@redhat.com>
13649 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13650 derive_equivalences.
13651 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13652 Code moved into....
13653 (edge_info::derive_equivalences): New private member function
13655 * tree-ssa-dom.c (class edge_info): Changed from a struct
13656 to a class. Add ctor/dtor, methods and data members.
13657 (edge_info::edge_info): Renamed from allocate_edge_info.
13658 Initialize additional members.
13659 (edge_info::~edge_info): New.
13660 (free_dom_edge_info): Delete the edge info.
13661 (record_edge_info): Use new class & associated member functions.
13662 Tighten forms for testing for edge equivalences.
13663 (record_temporary_equivalences): Iterate over the simple
13664 equivalences rather than assuming there's only one per edge.
13665 (cprop_into_successor_phis): Iterate over the simple
13666 equivalences rather than assuming there's only one per edge.
13667 (optimize_stmt): Use operand_equal_p rather than pointer
13668 equality for mini-DSE code.
13670 2017-08-28 Nathan Sidwell <nathan@acm.org>
13672 * gcc.c (execute): Fold SIGPIPE handling into switch
13673 statement. Adjust internal error message.
13675 2017-08-28 Richard Biener <rguenther@suse.de>
13678 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13679 Do nothing for removed DIEs.
13681 2017-08-28 Richard Biener <rguenther@suse.de>
13683 PR tree-optimization/81977
13684 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13687 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13690 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13691 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13692 using targetm.gen_mem_thread_fence.
13694 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13697 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13698 predicate to register_operand. Reorder operands.
13699 (*btr<mode>): Ditto.
13700 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13701 (*btr<mode>_mask): Ditto.
13703 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13705 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13706 * config/rs6000/xmmintrin.h: New file.
13707 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13709 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13712 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13713 parameter and_insn and return it.
13714 (recombine_lvx_pattern): Insert a copy to ensure availability of
13715 the base register of the copied masking operation at the point of
13716 the instruction replacement.
13717 (recombine_stvx_pattern): Likewise.
13719 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13721 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13722 undocumented switches.
13723 (-mpower9-dform-vector): Likewise.
13724 (-mpower9-dform): Likewise.
13725 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13726 comments to delete references to -mpower9-dform* switches.
13727 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13728 Delete reference to -mpower9-dform* switches, test for
13729 -mpower9-vector instead.
13730 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13731 (OTHER_P9_VECTOR_MASKS): Likewise.
13732 (POWERPC_MASKS): Likewise.
13733 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13734 tests against -mpower9-dform* to -mpower9-vector. Delete code
13735 that checked for -mpower9-dform* consistancy with other options.
13736 Add test for -mpower9-misc to enable other power9 switches.
13737 (rs6000_init_hard_regno_mode_ok): Likewise.
13738 (rs6000_option_override_internal): Likewise.
13739 (rs6000_emit_prologue): Likewise.
13740 (rs6000_emit_epilogue): Likewise.
13741 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13742 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13743 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13745 (emit_fusion_p9_store): Likewise.
13746 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13747 resetting these macros if the assembler does not support ISA 3.0
13749 (TARGET_P9_DFORM_VECTOR): Likewise.
13750 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13751 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13753 2017-08-25 Alan Modra <amodra@gmail.com>
13755 PR rtl-optimization/81747
13756 * cse.c (cse_extended_basic_block): Don't attempt to record
13757 equivalences for degenerate conditional branches that branch
13758 to their fall-through.
13760 2017-08-24 Martin Sebor <msebor@redhat.com>
13762 PR middle-end/81908
13763 * gimple-fold.c (size_must_be_zero_p): New function.
13764 (gimple_fold_builtin_memory_op): Call it.
13766 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13768 * config/rs6000/mm_malloc.h: New file.
13770 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13772 PR tree-optimization/81913
13773 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13774 analysis when either IVs in condition can wrap.
13776 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13778 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13779 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13781 2017-08-24 Richard Biener <rguenther@suse.de>
13784 * targhooks.c (default_target_can_inline_p): Properly
13785 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13786 is present and always compare.
13787 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13788 infer -mfpmath=sse from TARGET_SSE_P.
13789 (ix86_can_inline_p): Properly use target_option_default_node when
13790 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13792 2017-08-24 Richard Biener <rguenther@suse.de>
13795 * dwarf2out.c (output_die): Handle flag_generate_offload like
13797 (output_comp_unit): Likewise.
13798 (gen_array_type_die): Likewise.
13799 (dwarf2out_early_finish): Likewise.
13800 (note_variable_value_in_expr): Likewise.
13801 (dwarf2out_finish): Likewise. Adjust assert.
13802 * cgraphunit.c (symbol_table::compile): Move setting of
13803 flag_generate_offload earlier ...
13804 (symbol_table::finalize_compilation_unit): ... here, before
13805 early debug finalization.
13807 2017-08-24 Richard Biener <rguenther@suse.de>
13809 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13810 and ipa-fnsummary.h.
13811 (ix86_can_inline_p): When ix86_fpmath flags do not match
13812 check whether the callee uses FP math at all.
13814 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13816 PR middle-end/81931
13817 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13818 instead of TYPE_PRECISION.
13820 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13821 Alan Hayward <alan.hayward@arm.com>
13822 David Sherwood <david.sherwood@arm.com>
13824 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13826 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13828 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13830 (ix86_frame::stack_realign_allocate): New field.
13831 (struct machine_frame_state): Modify comments.
13832 (machine_frame_state::sp_realigned_fp_end): New field.
13833 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13834 layout calculation.
13835 (sp_valid_at): Add assertion to assure no attempt to access invalid
13836 offset of a realigned stack.
13837 (fp_valid_at): Likewise.
13838 (choose_baseaddr): Modify comments.
13839 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13840 ix86_expand_prologue.
13841 (ix86_expand_prologue): Modify stack realignment and allocation.
13842 (ix86_expand_epilogue): Modify comments.
13843 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13844 avx2_runtime, avx512f, and avx512f_runtime.
13846 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13848 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13849 (mstackrealign): Do not init to -1.
13850 * config/i386/i386.c (ix86_option_override_internal):
13851 Check opts_set, not opts when setting default value of
13852 opts->x_ix86_force_align_arg_pointer.
13854 2017-08-23 Richard Biener <rguenther@suse.de>
13856 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13857 lang_hooks.decl_printable_name.
13858 * print-rtl-function.c (print_rtx_function): Likewise.
13859 * tree-pretty-print.c (dump_function_header): Likewise.
13861 2017-08-23 Richard Biener <rguenther@suse.de>
13864 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13865 -g0 at compile-time.
13867 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13869 PR middle-end/19706
13870 * doc/sourcebuild.texi (Other hardware attributes):
13873 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13875 PR middle-end/19706
13876 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13877 Fix single-use check.
13879 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13881 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13883 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13885 * doc/install.texi: Modify to add more details on running selected
13888 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13890 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13891 is combined with -mabi=ms.
13892 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13895 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13897 PR tree-optimization/81488
13898 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13899 and cached_basis fields.
13900 (MAX_SPREAD): New constant.
13901 (alloc_cand_and_find_basis): Initialize new fields.
13902 (clear_visited): New function.
13903 (create_phi_basis_1): Rename from create_phi_basis, set visited
13904 and cached_basis fields.
13905 (create_phi_basis): New wrapper function.
13906 (phi_add_costs_1): Rename from phi_add_costs, add spread
13907 parameter, set visited field, short-circuit when limits reached.
13908 (phi_add_costs): New wrapper function.
13909 (record_phi_increments_1): Rename from record_phi_increments, set
13911 (record_phi_increments): New wrapper function.
13912 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13913 (phi_incr_cost): New wrapper function.
13914 (all_phi_incrs_profitable_1): Rename from
13915 all_phi_incrs_profitable, set visited field.
13916 (all_phi_incrs_profitable): New wrapper function.
13918 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13919 Alan Hayward <alan.hayward@arm.com>
13920 David Sherwood <david.sherwood@arm.com>
13922 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13923 that takes the outer and inner modes.
13924 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13925 comparison as the canonical test for a paradoxical subreg.
13926 * combine.c (simplify_set): Use paradoxical_subreg_p.
13927 (make_extraction): Likewise.
13928 (force_to_mode): Likewise.
13929 (rtx_equal_for_field_assignment_p): Likewise.
13930 (gen_lowpart_for_combine): Likewise.
13931 (simplify_comparison): Likewise.
13932 * cse.c (equiv_constant): Likewise.
13933 * expmed.c (store_bit_field_1): Likewise.
13934 * final.c (alter_subreg): Likewise.
13935 * fwprop.c (propagate_rtx): Likewise.
13936 (forward_propagate_subreg): Likewise.
13937 * ira-conflicts.c (ira_build_conflicts): Likewise.
13938 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13939 * lra-constraints.c (curr_insn_transform): Likewise.
13940 (split_reg): Likewise.
13941 * lra-eliminations.c (move_plus_up): Likewise.
13942 (lra_eliminate_regs_1): Likewise.
13943 * recog.c (general_operand): Likewise.
13944 * ree.c (combine_reaching_defs): Likewise.
13945 * reload.c (push_reload): Likewise.
13946 (find_reloads): Likewise.
13947 * reload1.c (elimination_effects): Likewise.
13948 (compute_reload_subreg_offset): Likewise.
13949 (choose_reload_regs): Likewise.
13950 * rtlanal.c (subreg_lsb_1): Likewise.
13951 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13952 (simplify_subreg): Likewise.
13953 * var-tracking.c (track_loc_p): Likewise.
13954 * emit-rtl.c (byte_lowpart_offset): Likewise.
13955 (paradoxical_subreg_p): Delete out-of-line definition.
13957 2017-08-22 Jeff Law <law@redhat.com>
13959 PR tree-optimization/81741
13960 PR tree-optimization/71947
13961 * tree-ssa-dom.c: Include tree-inline.h.
13962 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13963 equivalences if one is more expensive to compute than the other.
13964 * tree-ssa-scopedtables.h (class const_or_copies): Make
13965 record_const_or_copy_raw method private.
13966 (class avail_exprs_stack): New method simplify_binary_operation.
13967 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13968 avail_exprs_stack::simplify_binary_operation as needed.
13969 (avail_exprs_stack::simplify_binary_operation): New function.
13971 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13973 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13974 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13975 (DOT_SYMBOLS): Likewise.
13976 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13977 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13978 (RS6000_ABI_NAME): Likewise.
13979 (TARGET_CMODEL): Likewise.
13980 (TOC_SECTION_ASM_OP): Likewise.
13981 (SET_CMODEL): New macro.
13982 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13984 2017-08-22 Richard Biener <rguenther@suse.de>
13986 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13987 to allow for free-lang-data replacements similar to verify_type_variant.
13989 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13990 Alan Hayward <alan.hayward@arm.com>
13991 David Sherwood <david.sherwood@arm.com>
13993 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13994 VOIDmode for the LABEL_REF.
13996 2017-08-22 Richard Biener <rguenther@suse.de>
13998 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14000 2017-08-22 Richard Biener <rguenther@suse.de>
14002 * common.opt (feliminate-dwarf2-dups): Ignore.
14003 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14004 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14005 same_die_p_wrap, compute_section_prefix,
14006 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14007 (comdat_symbol_id, comdat_symbol_number): Likewise.
14008 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14010 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14011 (output_die): Mark unreachable path unreachable.
14012 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14013 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14014 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14015 (dwarf2out_early_finish): Likewise.
14017 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14019 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14021 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14024 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14025 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14026 (avr_attribute_table) <io, io_low, address>: Initialize
14027 .decl_required with true.
14029 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14031 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14032 undocumented debugging options.
14033 (-mvsx-scalar-double): Likewise.
14034 (-mallow-df-permute): Likewise.
14035 (-mvectorize-builtins): Likewise.
14036 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14037 (rs6000_builtin_vectorized_function): Likewise.
14038 (rs6000_builtin_md_vectorized_function): Likewise.
14039 (rs6000_opt_vars): Likewise.
14041 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14044 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14045 (*btrq_imm): Rename from *btrq.
14046 (*btcq_imm): Rename from *btcq.
14047 (btsc): New code attribute.
14048 (*<btsc><mode>): New insn pattern.
14049 (*btr<mode>): Ditto.
14050 (*<btsc><mode>_mask): New insn_and_split pattern.
14051 (*btr<mode>_mask): Ditto.
14053 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14054 Alan Hayward <alan.hayward@arm.com>
14055 David Sherwood <david.sherwood@arm.com>
14057 * function.c (pad_below): Simplify padding calculation.
14059 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14060 Alan Hayward <alan.hayward@arm.com>
14061 David Sherwood <david.sherwood@arm.com>
14063 * target.def (function_prologue): Remove frame size argument.
14064 (function_epilogue): Likewise.
14065 * doc/tm.texi: Regenerate.
14066 * final.c (final_start_function): Update call to function_prologue.
14067 (final_end_function): Update call to function_epilogue.
14068 (default_function_pro_epilogue): Remove frame size argument.
14069 * output.h (default_function_pro_epilogue): Likewise.
14070 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14071 (arm_output_function_prologue): Likewise.
14072 * config/frv/frv.c (frv_function_prologue): Likewise.
14073 (frv_function_epilogue): Likewise.
14074 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14075 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14076 (ia64_output_function_epilogue): Likewise.
14077 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14078 (m32r_output_function_epilogue): Likewise.
14079 * config/microblaze/microblaze.c (microblaze_function_prologue)
14080 (microblaze_function_epilogue): Likewise.
14081 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14082 (mips_output_function_epilogue): Likewise.
14083 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14084 (mmix_target_asm_function_epilogue): Likewise.
14085 * config/msp430/msp430.c (msp430_start_function): Likewise.
14086 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14087 (nds32_asm_function_epilogue): Likewise.
14088 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14089 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14090 (pa_output_function_epilogue): Likewise.
14091 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14092 (rs6000_output_function_epilogue): Likewise.
14093 * config/rl78/rl78.c (rl78_start_function): Likewise.
14094 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14095 (rs6000_output_function_epilogue): Likewise.
14096 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14097 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14098 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14099 (sparc_asm_function_epilogue): Likewise.
14101 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14103 * tree.h (type_has_mode_precision_p): New function.
14104 * convert.c (convert_to_integer_1): Use it.
14105 * expr.c (expand_expr_real_2): Likewise.
14106 (expand_expr_real_1): Likewise.
14107 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14108 * match.pd: Likewise.
14109 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14110 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14111 * tree-tailcall.c (process_assignment): Likewise.
14112 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14113 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14114 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14115 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14116 (vectorizable_assignment): Likewise.
14117 (vectorizable_shift): Likewise.
14118 (vectorizable_operation): Likewise.
14119 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14121 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14123 * match.pd: Add pow (C, x) simplification.
14125 2017-08-21 Richard Biener <rguenther@suse.de>
14127 PR tree-optimization/81900
14128 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14129 for blocks with abnormal predecessors.
14130 (compute_antic): Do not set visited flag prematurely.
14132 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14135 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14137 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14139 * stor-layout.h (vector_type_mode): Move to...
14140 * tree.h (vector_type_mode): ...here.
14141 * stor-layout.c (vector_type_mode): Move to...
14142 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14144 2017-08-21 Richard Biener <rguenther@suse.de>
14146 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14147 register_external_die hooks.
14148 (debug_false_tree_charstarstar_uhwistar): Declare.
14149 (debug_nothing_tree_charstar_uhwi): Likewise.
14150 * debug.c (do_nothing_debug_hooks): Adjust.
14151 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14152 (debug_nothing_tree_charstar_uhwi): Likewise.
14153 * dbxout.c (dbx_debug_hooks): Adjust.
14154 (xcoff_debug_hooks): Likewise.
14155 * sdbout.c (sdb_debug_hooks): Likewise.
14156 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14157 * dwarf2out.c (macinfo_label_base): New global.
14158 (dwarf2out_register_external_die): New function for the
14159 register_external_die hook.
14160 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14161 (dwarf2_debug_hooks): Use them.
14162 (dwarf2_lineno_debug_hooks): Adjust.
14163 (struct die_struct): Add with_offset flag.
14164 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14165 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14166 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14167 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14168 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14169 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14170 defining section names for the early LTO debug variants.
14171 (reset_indirect_string): New helper.
14172 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14173 (print_dw_val): Add support for offsetted symbol references.
14174 (get_ultimate_context): Split out from is_cxx.
14175 (is_cxx): Use get_ultimate_context.
14176 (is_fortran): Add decl overload.
14177 (compute_comp_unit_symbol): Split out worker from
14178 compute_section_prefix.
14179 (compute_section_prefix): Call compute_comp_unit_symbol and
14180 set comdat_type_p here.
14181 (output_die): Skip DIE symbol output for the LTO added one.
14182 Handle DIE symbol references with offset.
14183 (output_comp_unit): Guard section name mangling properly.
14184 For LTO debug sections emit a symbol at the section beginning
14185 which we use to refer to its DIEs.
14186 (add_abstract_origin_attribute): For DIEs registered via
14187 dwarf2out_register_external_die directly refer to the early
14188 DIE rather than indirectly through the shadow one we created.
14189 Remove obsolete call to dwarf2out_abstract_function for
14190 non-function/block origins.
14191 (gen_array_type_die): When generating early LTO debug do
14192 not emit DW_AT_string_length.
14193 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14194 late when in LTO. As suggested place a gcc_unreachable for
14195 the DECL_ABSTRACT_P case.
14196 (gen_subprogram_die): Avoid another specification DIE
14197 for early built declarations/definitions for the late LTO case.
14198 (gen_variable_die): Add type references for late duplicated VLA dies
14200 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14201 we have the abstract instance already.
14202 (process_scope_var): Adjust decl DIE contexts in LTO which
14203 first puts them in limbo.
14204 (gen_decl_die): Do not generate type DIEs late apart from
14205 types for VLAs or for decls we do not yet have a DIE. Do not
14206 call dwarf2out_abstract_function late.
14207 (dwarf2out_early_global_decl): Make sure to create DIEs
14208 for abstract instances of a decl first.
14209 (dwarf2out_late_global_decl): Adjust comment.
14210 (output_macinfo_op): With multiple macro sections use
14211 macinfo_label_base to distinguish labels.
14212 (output_macinfo): Likewise. Update macinfo_label_base.
14213 Pass in the line info label.
14214 (note_variable_value_in_expr): When generating LTO resolve
14215 all variable values here by generating DIEs as needed.
14216 (init_sections_and_labels): Add early LTO debug flag parameter
14217 and generate different sections and names if set. Add generation
14218 counter for the labels so we can have multiple of them.
14219 (reset_dies): Helper to allow DIEs to be output multiple times.
14220 (dwarf2out_finish): When outputting DIEs to the fat part of an
14221 LTO object first reset DIEs.
14222 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14223 (modified_type_die): Check for decl_ultimate_origin being self
14225 (gen_type_die_with_usage): Likewise.
14226 (gen_typedef_die): Allow decl_ultimate_origin being self.
14227 (set_decl_abstract_flags): Remove.
14228 (set_block_abstract_flags): Likewise.
14229 (dwarf2out_abstract_function): Treat the early generated DIEs
14230 as the abstract copy and only add DW_AT_inline and
14231 DW_AT_artificial here and call set_decl_origin_self.
14232 If the DIE has an abstract origin don't do anything.
14233 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14234 if we have none yet (Go fails to build one, PR78628).
14235 (variably_modified_type_p): Prevent endless recursion for Ada
14236 cyclic pointer types.
14237 * lto-streamer-in.c: Include debug.h.
14238 (dref_queue): New global.
14239 (lto_read_tree_1): Stream in DIE references.
14240 (lto_input_tree): Register DIE references.
14241 (input_function): Stream DECL_DEBUG_ARGS.
14242 * lto-streamer-out.c: Include debug.h.
14243 (lto_write_tree_1): Output DIE references.
14244 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14245 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14246 (output_function): Stream DECL_DEBUG_ARGS.
14247 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14248 Stream DECL_ABSTRACT_ORIGIN.
14249 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14250 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14251 DECL_CONTEXT for file-scope decls.
14252 * lto-streamer.h (struct dref_entry): Declare.
14253 (dref_queue): Likewise.
14254 * cfgexpand.c (pass_expand::execute): Do not call the
14255 outlining_inline_function hook here.
14256 * lto-wrapper.c (debug_obj): New global.
14257 (tool_cleanup): Unlink it if required.
14258 (debug_objcopy): New function.
14259 (run_gcc): Handle early debug sections in the IL files by
14260 extracting them to separate files, partially linkin them and
14261 feeding the result back as result to the linker.
14262 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14263 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14264 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14265 sections into a separate segment.
14266 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14268 (darwin_asm_dwarf_section): Likewise.
14269 (darwin_asm_output_dwarf_offset): Likewise.
14270 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14272 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14273 Alan Hayward <alan.hayward@arm.com>
14274 David Sherwood <david.sherwood@arm.com>
14276 * read-md.h (md_reader::record_potential_iterator_use): Replace
14277 pointer argument with an rtx and an index.
14278 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14279 (apply_mode_iterator): Likewise.
14280 (apply_code_iterator): Likewise.
14281 (apply_int_iterator): Likewise.
14282 (apply_subst_iterator): Likewise.
14283 (record_iterator_use): Likewise.
14284 (record_attribute_use): Likewise.
14285 (md_reader::record_potential_iterator_use): Likewise. Update calls
14286 to record_iterator_use and apply_iterator.
14287 (iterator_use): Replace ptr with x and index.
14288 (attribute_use): Likewise.
14289 (apply_attribute_uses): Update calls to apply_iterator.
14290 (apply_iterators): Likewise. Update initialization of iterator_use.
14291 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14292 and record_potential_iterator_use.
14293 (rtx_reader::read_rtx_operand): Likewise.
14295 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14296 Alan Hayward <alan.hayward@arm.com>
14297 David Sherwood <david.sherwood@arm.com>
14299 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14302 2017-08-21 Richard Biener <rguenther@suse.de>
14304 PR middle-end/81884
14305 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14306 at struct end conservatively when comparing common bases.
14308 2017-08-21 Richard Biener <rguenther@suse.de>
14310 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14311 (mem_ref_in_stmt): Remove.
14312 (determine_max_movement): Use ref index to get at the reference.
14313 (invariantness_dom_walker::before_dom_children): Deal with
14314 lim data already initialized.
14315 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14317 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14319 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14320 (TARGET_ISA_ROUND): Ditto.
14321 (TARGET_ROUND): Ditto.
14322 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14323 * config/i386/i386.md: Ditto.
14324 * config/i386/sse.md: Ditto.
14325 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14326 with OPTION_MASK_ISA_SSE4_1.
14328 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14331 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14332 __builtin_ia32_lzcnt_u16.
14334 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14337 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14338 (rs6000_set_current_function): Rewrite function to use it.
14340 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14343 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14344 and TYPE_WARN_IF_NOT_ALIGN.
14345 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14346 (handle_warn_if_not_align): New.
14347 (place_union_field): Call handle_warn_if_not_align.
14348 (place_field): Call handle_warn_if_not_align.
14349 Copy TYPE_WARN_IF_NOT_ALIGN.
14350 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14351 (layout_type): Likewise.
14352 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14354 (tree_decl_common): Add warn_if_not_align.
14355 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14356 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14357 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14358 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14359 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14360 * doc/extend.texi: Document warn_if_not_aligned attribute.
14361 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14363 2017-08-17 Martin Liska <mliska@suse.cz>
14366 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14367 (get_data_dependence): Use it as pointer type.
14368 (distribute_loop): Likewise.
14370 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14372 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14373 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14374 (p8_vmrgow_<mode>_direct): New define_insn.
14375 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14376 handle endianness for vmrgew and vmrgow permute patterns.
14378 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14380 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14381 * config/rs6000/rs6000-cpus.def: Remove comment.
14382 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14383 (POWERPC_MASKS): Likewise.
14384 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14385 use of TARGET_VSX_TIMODE.
14386 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14387 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14388 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14389 (rs6000_option_override_internal): Remove dead code.
14390 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14391 (rs6000_legitimize_reload_address): Likewise.
14392 (rs6000_legitimate_address_p): Likewise.
14393 (rs6000_opt_masks): Delete "vsx-timode".
14394 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14395 from function comment.
14396 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14397 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14398 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14400 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14401 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14402 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14403 (VSX_TI): Likewise.
14405 (define_peephole2): Likewise.
14407 2017-08-17 Martin Sebor <msebor@redhat.com>
14410 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14411 past the end of an array.
14412 (test_pp_format): Add test cases.
14414 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14416 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14417 missing ECF_NOTHROW flags.
14419 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14422 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14423 operands residing in integer registers.
14424 (*vsx_le_perm_load_<mode>): Likewise.
14425 (*vsx_le_perm_store_<mode>): Likewise.
14426 (define_peephole2): Add peepholes to optimize the above.
14428 2017-08-17 Marek Polacek <polacek@redhat.com>
14430 PR middle-end/81814
14431 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14432 to mimic what shorten_compare did. Change the return type to bool.
14433 (fold_cond_expr_with_comparison): Update call to
14434 operand_equal_for_comparison_p.
14435 (fold_ternary_loc): Likewise.
14437 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14439 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14441 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14443 2017-08-17 Richard Biener <rguenther@suse.de>
14445 * tree-ssa-structalias.c (solve_graph): When propagating
14446 to successors update the graphs succ edges and avoid duplicate work.
14448 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14451 * config/i386/i386.c (ix86_option_override_internal): Save target
14452 specific options after ix86_stack_protector_guard_reg was changed.
14454 2017-08-17 Richard Biener <rguenther@suse.de>
14456 PR tree-optimization/81827
14457 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14458 (new_var_info): Initialize it conservatively.
14459 (get_call_vi): Mark register vars.
14460 (new_scalar_tmp_constraint_exp): Likewise.
14461 (handle_rhs_call): Likewise.
14462 (handle_const_call): Likewise.
14463 (create_function_info_for): Likewise.
14464 (solve_constraints): Sort varinfos to separate register from
14465 non-register vars to pack points-to solution bitmaps during
14468 2017-08-17 Marek Polacek <polacek@redhat.com>
14470 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14472 2017-08-17 Richard Biener <rguenther@suse.de>
14474 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14475 to true when overflow is undefined and we saturated the result.
14477 2017-08-17 Alan Modra <amodra@gmail.com>
14480 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14481 Don't use store multiple if only one reg needs saving.
14482 (interesting_frame_related_regno): New function.
14483 (rs6000_frame_related): Don't emit frame info for regs that
14485 (rs6000_emit_epilogue): Likewise.
14487 2017-08-16 Nathan Sidwell <nathan@acm.org>
14489 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14490 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14491 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14492 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14493 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14494 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14495 (verify_type): Adjust for TYPE_BINFO move.
14496 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14497 process TYPE_BINFO directly.
14498 (hash_tree): Likewise.
14499 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14501 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14504 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14506 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14508 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14511 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14512 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14513 Add ix86_binary_operator_ok to insn constraint.
14514 (*iordi_1_bts): Ditto.
14515 (*xordi_1_btc): Ditto.
14516 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14517 Update corresponding peephole2 pattern.
14521 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14523 PR tree-optimization/81832
14524 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14525 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14527 2017-08-16 Marek Polacek <polacek@redhat.com>
14530 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14531 regarding size_zero_node.
14533 2017-08-16 Martin Liska <mliska@suse.cz>
14536 * config.gcc: Respect previously set extra_objs in case
14539 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14541 PR tree-optimization/81835
14542 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14543 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14544 not depend on the phi.
14546 2017-08-16 Alan Modra <amodra@gmail.com>
14548 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14551 2017-08-16 Alan Modra <amodra@gmail.com>
14553 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14554 (save_reg_p): ..into this. Update all callers.
14555 (first_reg_to_save): Simplify.
14557 2017-08-16 Alan Modra <amodra@gmail.com>
14559 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14562 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14566 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14567 constant count if that count is less than 32.
14569 2017-08-15 Nathan Sidwell <nathan@acm.org>
14571 * gcc.c (execute): Emit friendlier message if inferior is killed
14572 by an external cause.
14574 2017-08-15 Richard Biener <rguenther@suse.de>
14576 PR tree-optimization/81790
14577 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14578 CONSTRUCTORs from simplifying and VN.
14580 2017-08-14 Martin Sebor <msebor@redhat.com>
14582 * builtin-attrs.def: Add comments.
14584 2017-08-14 Martin Sebor <msebor@redhat.com>
14587 * doc/extend.texi (attribute nonstring): Document new attribute.
14589 2017-08-14 Martin Sebor <msebor@redhat.com>
14592 * tree-diagnostic.c (default_tree_printer): Handle %G.
14593 * gimple-pretty-print.h (percent_G_format): Declare new function.
14594 * gimple-pretty-print.c (percent_G_format): Define.
14595 * tree-pretty-print.c (percent_K_format): Add argument.
14597 2017-08-14 Martin Sebor <msebor@redhat.com>
14599 PR translation/79998
14600 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14601 Remove a stray space.
14603 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14606 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14607 (*iordi_1_bts): Ditto.
14608 (*xordi_1_btc): Ditto.
14610 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14613 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14614 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14616 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14617 quoted strings, and make more translator-friendly.
14618 (darwin_rs6000_override_options): Likewise.
14619 (rs6000_option_override_internal): Likewise.
14620 (rs6000_return_in_memory): Fix overlong line.
14621 (init_cmulative_args): Use quoted strings, and make more
14622 translator-friendly.
14623 (rs6000_pass_by_reference): Fix overlong line.
14624 (def_builtin): Use quoted strings.
14625 (altivec_expand_predicate_builtin): Use quoted strings, and make
14626 more translator-friendly.
14627 (htm_expand_builtin): Use quoted strings.
14628 (cpu_expand_builtin): Use quoted strings, and make more
14629 translator-friendly.
14630 (altivec_expand_builtin): Likewise.
14631 (paired_expand_predicate_builtin): Likewise.
14632 (rs6000_invalid_builtin): Likewise.
14633 (builtin_function_type): Use quoted strings.
14634 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14635 more translator-friendly.
14636 (rs6000_trampoline_init): Likewise.
14637 (rs6000_handle_altivec_attribute): Likewise.
14638 (rs6000_inner_target_options): Use quoted strings.
14639 (rs6000_disable_incompatible_switches): Likewise.
14640 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14641 strings, and make more translator-friendly.
14642 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14644 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14646 PR tree-optimization/81799
14647 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14648 cond_expr to simple gimple operand.
14650 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14652 PR middle-end/46932
14653 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14655 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14659 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14660 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14661 TARGET_GASISR_PROLOGUES.
14662 * config/avr/avr.c (avr_option_override): Same.
14663 (avr_pass_pre_proep::execute): Same.
14665 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14668 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14669 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14671 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14673 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14674 define_insn_and_split. Split to a memory load from 0 in
14675 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14676 using PTR mode iterator.
14677 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14678 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14679 (*add_tp_<mode>): Redefine as define_insn_and_split.
14680 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14681 address space. Merge with *add_tp_x32 using PTR mode iterator.
14682 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14683 Split to an add with a memory load from 0 in
14684 DEFAULT_TLS_SEG_REG address space.
14686 2017-08-12 Andrew Pinski <apinski@cavium.com>
14688 * config/aarch64/aarch64-option-extensions.def (rdma):
14689 Fix feature string to what Linux prints out in /proc/cpuinfo.
14691 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14694 * dwarf2out.c (modified_type_die): For C typedef types that have
14695 an ultimate origin, process the ultimate origin instead of the
14697 (gen_typedef_die): Assert that input DECLs have no ultimate
14699 (gen_type_die_with_usage): For typedef variants that have an
14700 ultimate origin, just call gen_decl_die on the original DECL.
14701 (process_scope_var): Avoid creating DIEs for local typedefs and
14702 concrete static variables.
14704 2017-08-12 Alan Modra <amodra@gmail.com>
14708 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14709 match gnu-user.h startfile.
14710 (ENDFILE_LINUX_SPEC): Similarly.
14712 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14715 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14717 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14719 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14720 * config/aarch64/aarch64.md (mov<mode>): Change.
14721 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14722 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14723 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14725 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14727 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14728 for storage order barriers.
14730 2017-08-11 Martin Liska <mliska@suse.cz>
14733 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14734 variables of void type.
14736 2017-08-11 Martin Liska <mliska@suse.cz>
14738 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14739 TARGET_SUPPORTS_ALIASES.
14740 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14741 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14742 (optimize_weakref): Likewise.
14743 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14744 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14745 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14747 2017-08-11 Martin Liska <mliska@suse.cz>
14750 * config/i386/i386.c (make_resolver_func): Do complete
14751 refactoring of the function.
14753 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14756 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14757 * config/i386/i386.c (ix86_stack_protect_guard): Use
14758 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14759 * doc/invoke.texi (x86 Options): Document
14760 -mstack-protector-guard-symbol= option.
14762 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14764 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14765 * config/i386/i386.c (ix86_split_stack_guard): New function.
14766 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14767 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14768 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14769 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14770 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14771 (split_stack_space_check): Call ix86_split_stack_guard.
14773 2017-08-10 Martin Sebor <msebor@redhat.com>
14775 * print-tree.c (print_node): Print location using the established
14777 Replace spaces with colons.
14778 (debug_raw, debug): Ditto.
14780 2017-08-10 Martin Sebor <msebor@redhat.com>
14783 * pretty-print.c (pp_format): Correct the handling of %s precision.
14785 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14788 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14790 (ix86_finalize_stack_frame_flags): This. Also clear
14791 frame_pointer_needed if -fno-omit-frame-pointer is used without
14793 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14794 with ix86_finalize_stack_frame_flags.
14795 (ix86_expand_epilogue): Likewise.
14796 (ix86_expand_split_stack_prologue): Likewise.
14797 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14799 2017-08-10 Martin Liska <mliska@suse.cz>
14802 * c-attribs.c (handle_target_attribute):
14803 Report warning for an empty string argument of target attribute.
14805 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14808 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14810 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14811 or DECL_NONLOCAL labels.
14812 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14813 or DECL_NONLOCAL labels here.
14815 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14817 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14818 to indicate when early gimple folding has been disabled.
14819 (rs6000_gimple_fold_builtin): Add debug content.
14820 (rs6000_invalid_builtin): Fix whitespace.
14821 (rs6000_expand_builtin): Fix whitespace.
14822 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14824 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14827 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14828 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14830 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14832 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14833 (qdf24xx): Likewise.
14834 * config/aarch64/aarch64.md: Include falkor.md.
14835 * config/aarch64/falkor.md: New.
14837 2017-08-09 Marek Polacek <polacek@redhat.com>
14840 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14841 * diagnostic.c (emit_diagnostic): Add a comment.
14842 (emit_diagnostic_valist): New function.
14844 2017-08-09 Marek Polacek <polacek@redhat.com>
14847 * input.c (make_location): New overload.
14848 * input.h (make_location): Declare.
14850 2017-08-08 Alan Modra <amodra@gmail.com>
14851 H.J. Lu <hongjiu.lu@intel.com>
14854 * gcc.c (NO_PIE_SPEC): Delete.
14855 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14857 (LINK_PIE_SPEC): ..to here.
14858 (LINK_COMMAND_SPEC): Support -no-pie.
14859 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14860 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14862 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14863 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14864 (ENDFILE_CRTEND_SPEC): Similarly.
14866 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14869 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14870 (mstack-protector-guard-offset=): Ditto.
14871 * config/i386/i386.c (ix86_option_override): Handle
14872 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14874 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14875 ix86_stack_protect_guard_offset variables.
14876 (TARGET_STACK_PROTECT_GUARD): Always define.
14877 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14878 and -mstack-protector-guard-offset= options.
14880 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14882 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14883 boundary case for the last candidate.
14885 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14887 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14888 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14890 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14892 PR middle-end/19706
14893 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14894 * config/aarch64/aarch64-builtins.c
14895 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14896 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14897 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14899 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14900 Andrew Pinski <pinskia@gmail.com>
14902 PR middle-end/19706
14903 * internal-fn.def (XORSIGN): New.
14904 * optabs.def (xorsign_optab): New.
14905 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14906 (convert_expand_mult_copysign): New.
14907 (pass_optimize_widening_mul::execute): Call
14908 convert_expand_mult_copysign.
14910 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14912 PR tree-optimization/81354
14913 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14914 Insert on edges rather than explicitly creating landing pads.
14915 (analyze_candidates_and_replace): Commit edge inserts.
14917 2017-08-08 Richard Biener <rguenther@suse.de>
14919 PR middle-end/81719
14920 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14921 (expand_simple_operations): Also look through ADDR_EXPRs with
14922 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14924 2017-08-08 Richard Biener <rguenther@suse.de>
14926 PR tree-optimization/81723
14927 * tree-vect-slp.c (struct bst_traits): New hash traits.
14928 (bst_fail): New global.
14929 (vect_build_slp_tree_2): New worker, split out from ...
14930 (vect_build_slp_tree): ... this now wrapping it with using
14931 bst_fail set to cache SLP tree build fails. Properly handle
14933 (vect_analyze_slp_instance): Allocate and free bst_fail.
14935 2017-08-08 Martin Liska <mliska@suse.cz>
14938 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14939 LABEL_DECLs that can be from a different function.
14941 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14943 PR tree-optimization/81744
14944 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14945 loop's number of iterations.
14947 2017-08-08 Martin Liska <mliska@suse.cz>
14949 * asan.c: Include header files.
14950 * attribs.c (build_decl_attribute_variant): New function moved
14952 (build_type_attribute_qual_variant): Likewise.
14953 (cmp_attrib_identifiers): Likewise.
14954 (simple_cst_list_equal): Likewise.
14955 (omp_declare_simd_clauses_equal): Likewise.
14956 (attribute_value_equal): Likewise.
14957 (comp_type_attributes): Likewise.
14958 (build_type_attribute_variant): Likewise.
14959 (lookup_ident_attribute): Likewise.
14960 (remove_attribute): Likewise.
14961 (merge_attributes): Likewise.
14962 (merge_type_attributes): Likewise.
14963 (merge_decl_attributes): Likewise.
14964 (merge_dllimport_decl_attributes): Likewise.
14965 (handle_dll_attribute): Likewise.
14966 (attribute_list_equal): Likewise.
14967 (attribute_list_contained): Likewise.
14968 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14969 (lookup_attribute_by_prefix): Likewise.
14970 * bb-reorder.c: Include header files.
14971 * builtins.c: Likewise.
14972 * calls.c: Likewise.
14973 * cfgexpand.c: Likewise.
14974 * cgraph.c: Likewise.
14975 * cgraphunit.c: Likewise.
14976 * convert.c: Likewise.
14977 * dwarf2out.c: Likewise.
14978 * final.c: Likewise.
14979 * fold-const.c: Likewise.
14980 * function.c: Likewise.
14981 * gimple-expr.c: Likewise.
14982 * gimple-fold.c: Likewise.
14983 * gimple-pretty-print.c: Likewise.
14984 * gimple.c: Likewise.
14985 * gimplify.c: Likewise.
14986 * hsa-common.c: Likewise.
14987 * hsa-gen.c: Likewise.
14988 * internal-fn.c: Likewise.
14989 * ipa-chkp.c: Likewise.
14990 * ipa-cp.c: Likewise.
14991 * ipa-devirt.c: Likewise.
14992 * ipa-fnsummary.c: Likewise.
14993 * ipa-inline.c: Likewise.
14994 * ipa-visibility.c: Likewise.
14996 * lto-cgraph.c: Likewise.
14997 * omp-expand.c: Likewise.
14998 * omp-general.c: Likewise.
14999 * omp-low.c: Likewise.
15000 * omp-offload.c: Likewise.
15001 * omp-simd-clone.c: Likewise.
15002 * opts-global.c: Likewise.
15003 * passes.c: Likewise.
15004 * predict.c: Likewise.
15005 * sancov.c: Likewise.
15006 * sanopt.c: Likewise.
15007 * symtab.c: Likewise.
15008 * toplev.c: Likewise.
15009 * trans-mem.c: Likewise.
15010 * tree-chkp.c: Likewise.
15011 * tree-eh.c: Likewise.
15012 * tree-into-ssa.c: Likewise.
15013 * tree-object-size.c: Likewise.
15014 * tree-parloops.c: Likewise.
15015 * tree-profile.c: Likewise.
15016 * tree-ssa-ccp.c: Likewise.
15017 * tree-ssa-live.c: Likewise.
15018 * tree-ssa-loop.c: Likewise.
15019 * tree-ssa-sccvn.c: Likewise.
15020 * tree-ssa-structalias.c: Likewise.
15021 * tree-ssa.c: Likewise.
15022 * tree-streamer-in.c: Likewise.
15023 * tree-vectorizer.c: Likewise.
15024 * tree-vrp.c: Likewise.
15025 * tsan.c: Likewise.
15026 * ubsan.c: Likewise.
15027 * varasm.c: Likewise.
15028 * varpool.c: Likewise.
15029 * tree.c: Remove functions moved to attribs.[ch].
15030 * tree.h: Likewise.
15031 * config/aarch64/aarch64.c: Add attrs.h header file.
15032 * config/alpha/alpha.c: Likewise.
15033 * config/arc/arc.c: Likewise.
15034 * config/arm/arm.c: Likewise.
15035 * config/avr/avr.c: Likewise.
15036 * config/bfin/bfin.c: Likewise.
15037 * config/c6x/c6x.c: Likewise.
15038 * config/cr16/cr16.c: Likewise.
15039 * config/cris/cris.c: Likewise.
15040 * config/darwin.c: Likewise.
15041 * config/epiphany/epiphany.c: Likewise.
15042 * config/fr30/fr30.c: Likewise.
15043 * config/frv/frv.c: Likewise.
15044 * config/ft32/ft32.c: Likewise.
15045 * config/h8300/h8300.c: Likewise.
15046 * config/i386/winnt.c: Likewise.
15047 * config/ia64/ia64.c: Likewise.
15048 * config/iq2000/iq2000.c: Likewise.
15049 * config/lm32/lm32.c: Likewise.
15050 * config/m32c/m32c.c: Likewise.
15051 * config/m32r/m32r.c: Likewise.
15052 * config/m68k/m68k.c: Likewise.
15053 * config/mcore/mcore.c: Likewise.
15054 * config/microblaze/microblaze.c: Likewise.
15055 * config/mips/mips.c: Likewise.
15056 * config/mmix/mmix.c: Likewise.
15057 * config/mn10300/mn10300.c: Likewise.
15058 * config/moxie/moxie.c: Likewise.
15059 * config/msp430/msp430.c: Likewise.
15060 * config/nds32/nds32-isr.c: Likewise.
15061 * config/nds32/nds32.c: Likewise.
15062 * config/nios2/nios2.c: Likewise.
15063 * config/nvptx/nvptx.c: Likewise.
15064 * config/pa/pa.c: Likewise.
15065 * config/pdp11/pdp11.c: Likewise.
15066 * config/powerpcspe/powerpcspe.c: Likewise.
15067 * config/riscv/riscv.c: Likewise.
15068 * config/rl78/rl78.c: Likewise.
15069 * config/rx/rx.c: Likewise.
15070 * config/s390/s390.c: Likewise.
15071 * config/sh/sh.c: Likewise.
15072 * config/sol2.c: Likewise.
15073 * config/sparc/sparc.c: Likewise.
15074 * config/spu/spu.c: Likewise.
15075 * config/stormy16/stormy16.c: Likewise.
15076 * config/tilegx/tilegx.c: Likewise.
15077 * config/tilepro/tilepro.c: Likewise.
15078 * config/v850/v850.c: Likewise.
15079 * config/vax/vax.c: Likewise.
15080 * config/visium/visium.c: Likewise.
15081 * config/xtensa/xtensa.c: Likewise.
15083 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15086 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15087 constraints since the -mupper-regs-* switches have been
15089 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15090 into a vector from a double word element that was extracted from
15091 another vector, and eliminate extra XXPERMDI instructions.
15092 (vsx_concat_<mode>_2): Likewise.
15093 (vsx_concat_<mode>_3): Likewise.
15094 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15095 concat to allow optimizing inserts from previous extracts.
15097 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15099 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15100 memory reference to a SSP offset in TLS address space.
15101 (ix86_print_operand) <case '@'>: Remove.
15102 (ix86_print_operand_punct_valid_p): Remove '@' code.
15103 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15104 UNSPEC_SP_TLS_TEST.
15105 (stack_tls_protect_set_<mode>): Remove.
15106 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15107 (stack_tls_protect_test_<mode>): Remove.
15108 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15110 2017-08-07 Olivier Hainque <hainque@adacore.com>
15113 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15115 2017-08-07 Douglas Rupp <rupp@adacore.com>
15117 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15118 variable was referenced as multidir in command.
15120 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15123 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15126 2017-08-07 Martin Liska <mliska@suse.cz>
15128 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15129 * config/rl78/rl78.c: Add include of attribs.h.
15130 * config/sh/sh.c: Likewise.
15131 * config/v850/v850.c: Likewise.
15133 2017-08-07 Tom de Vries <tom@codesourcery.com>
15135 PR middle-end/78266
15136 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15138 2017-08-07 Martin Liska <mliska@suse.cz>
15140 * config/mips/mips.c: Include attribs.h.
15142 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15145 * doc/invoke.texi: Document change in behvaior for -Ofast for
15148 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15150 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15152 (aarch64_pop_regs): Likewise.
15153 (aarch64_gen_load_pair): Likewise.
15154 (aarch64_save_callee_saves): Likewise.
15155 (aarch64_restore_callee_saves): Likewise.
15157 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15159 * config/i386/i386.c: Revert the last change.
15161 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15164 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15166 (ix86_finalize_stack_frame_flags): This. Also clear
15167 frame_pointer_needed if -fno-omit-frame-pointer is used without
15169 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15170 with ix86_finalize_stack_frame_flags.
15171 (ix86_expand_epilogue): Likewise.
15172 (ix86_expand_split_stack_prologue): Likewise.
15174 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15177 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15178 to P_AES for Westmere.
15180 2017-08-07 Jonathan Yong <10walls@gmail.com>
15182 * config/i386/mingw.opt (fset-stack-executable): Removed.
15183 * config/i386/cygming.opt (fset-stack-executable): Moved
15185 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15187 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15189 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15191 2017-08-07 Marek Polacek <polacek@redhat.com>
15193 PR middle-end/81737
15194 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15196 2017-08-07 Martin Liska <mliska@suse.cz>
15198 * attribs.h (canonicalize_attr_name): New function.
15199 (cmp_attribs): Move from c-format.c and adjusted.
15200 (is_attribute_p): Moved from tree.h.
15201 * tree-inline.c: Add new includes.
15202 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15203 (private_is_attribute_p): Remove.
15204 (private_lookup_attribute): Likewise.
15205 (private_lookup_attribute_by_prefix): Simplify.
15206 (remove_attribute): Use is_attribute_p.
15207 * tree.h: Remove removed declarations.
15209 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15211 PR middle-end/81698
15212 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15213 instead of computing it in the function. Formatting fix.
15214 (expand_case): Don't rely on default_edge being the first edge,
15215 clear it if removing it, pass default_edge to
15216 emit_case_dispatch_table.
15217 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15220 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15222 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15223 insn in the function, emit NOP after the insn.
15225 2017-08-06 Tom de Vries <tom@codesourcery.com>
15227 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15230 2017-08-06 Tom de Vries <tom@codesourcery.com>
15232 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15235 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15237 PR tree-optimization/57371
15238 * match.pd: New pattern.
15240 2017-08-04 Marek Polacek <polacek@redhat.com>
15242 PR middle-end/81695
15243 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15244 perform the computation in offset_int.
15246 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15248 PR tree-optimization/81136
15249 * tree-vectorizer.h: Include tree-hash-traits.h.
15250 (vec_base_alignments): New typedef.
15251 (vec_info): Add a base_alignments field.
15252 (vect_record_base_alignments): Declare.
15253 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15255 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15256 (create_data_ref): Add an is_conditional_in_stmt argument.
15257 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15258 the is_conditional_in_stmt field.
15259 (data_ref_loc): Add an is_conditional_in_stmt field.
15260 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15261 (find_data_references_in_stmt): Update call to create_data_ref.
15262 (graphite_find_data_references_in_stmt): Likewise.
15263 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15264 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15265 (vect_record_base_alignment): New function.
15266 (vect_record_base_alignments): Likewise.
15267 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15268 for nested statements even if we fail to compute a misalignment.
15269 Use pooled base alignments for unconditional references.
15270 (vect_find_same_alignment_drs): Compare base addresses instead
15272 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15273 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15275 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15277 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15278 Add an explicit name for the enum. Use auto_vec for slp_instances
15279 and grouped_stores.
15280 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15282 (_bb_vec_info): Add a constructor and destructor.
15283 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15284 (destroy_loop_vec_info): Delete.
15285 (vect_destroy_datarefs): Likewise.
15286 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15287 (vec_info::vec_info): New function.
15288 (vec_info::~vec_info): Likewise.
15289 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15290 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15291 destroy_loop_vec_info.
15292 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15293 (_loop_vec_info::_loop_vec_info): ...this.
15294 (destroy_loop_vec_info): Replace with...
15295 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15296 the stmt_vec_infos. Leave handling of vec_info information to its
15297 destructor. Remove explicit vector releases.
15298 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15299 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15300 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15301 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15302 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15303 (destroy_bb_vec_info): Replace with...
15304 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15305 information to its destructor.
15306 (vect_slp_analyze_bb_1): Use new and delete instead of
15307 new_bb_vec_info and destroy_bb_vec_info.
15308 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15311 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15313 * tree-data-ref.h (subscript): Add access_fn field.
15314 (data_dependence_relation): Add could_be_independent_p.
15315 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15316 (same_access_functions): Move to tree-data-ref.c.
15317 * tree-data-ref.c (ref_contains_union_access_p): New function.
15318 (access_fn_component_p): Likewise.
15319 (access_fn_components_comparable_p): Likewise.
15320 (dr_analyze_indices): Add a reference to access_fn_component_p.
15321 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15323 (constant_access_functions): Likewise.
15324 (add_other_self_distances): Likewise.
15325 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15326 (initialize_data_dependence_relation): Use XCNEW and remove
15327 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15328 of access functions that have the same type. Allow the
15329 subsequence to end with different bases in some circumstances.
15330 Record the chosen access functions in SUB_ACCESS_FN.
15331 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15332 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15333 (subscript_dependence_tester_1): Likewise dra and drb.
15334 (build_classic_dist_vector): Update calls accordingly.
15335 (subscript_dependence_tester): Likewise.
15336 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15337 DDR_COULD_BE_INDEPENDENT_P.
15338 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15339 comp_alias_ddrs instead of may_alias_ddrs.
15340 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15342 (vect_analyze_data_ref_dependence): Use it if
15343 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15344 distance vectors if that fails.
15345 (dependence_distance_ge_vf): New function.
15346 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15347 LOOP_VINFO_MAY_ALIAS_DDRS.
15349 2017-08-04 Richard Biener <rguenther@suse.de>
15351 PR middle-end/81705
15352 * fold-const.c (fold_binary_loc): Properly restrict
15353 minus_var0 && minus_var1 case when associating undefined overflow
15356 2017-08-04 Tom de Vries <tom@codesourcery.com>
15358 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15360 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15362 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15363 Don't start diagnostic messages with a capital letter.
15364 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15366 (rs6000_invalid_builtin): Likewise.
15367 (rs6000_trampoline_init): Likewise.
15369 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15372 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15373 after setting changeable df flags.
15375 2017-08-03 Richard Biener <rguenther@suse.de>
15377 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15378 up if the use is in USE - X.
15380 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15382 * toplev.c (dumpfile.h): New include.
15383 (internal_error_reentered): New static function. Use it...
15384 (internal_error_function): ...here to handle reentered internal_error.
15386 2017-08-03 Richard Biener <rguenther@suse.de>
15388 PR middle-end/81148
15389 * fold-const.c (split_tree): Add minus_var and minus_con
15390 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15391 here but always use minus_*.
15392 (associate_trees): Assert we never associate with MINUS_EXPR
15393 and NULL first operand. Do not recurse for PLUS_EXPR operands
15394 when associating as MINUS_EXPR either.
15395 (fold_binary_loc): Track minus_var and minus_con.
15397 2017-08-03 Tom de Vries <tom@codesourcery.com>
15400 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15401 ACCEL_COMPILER, apply finish_options on
15402 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15404 2017-08-03 Tom de Vries <tom@codesourcery.com>
15407 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15408 function_entry_patch_area_size > 0.
15410 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15413 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15414 instead of 10??LU, perform unit multiplication in wide_int,
15415 don't change alloc_object_size_limit if the limit is larger
15418 PR tree-optimization/81655
15419 PR tree-optimization/81588
15420 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15421 the case when ranges[i].low and high are 1 for unsigned type with
15424 PR middle-end/81052
15425 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15426 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15428 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15430 * tree-vrp.h: Add include guard.
15432 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15435 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15436 (ud2): New insn pattern.
15437 * config/i386/i386.c (ix86_expand_epilogue):
15438 For naked functions, generate ud2 instead of trap insn.
15440 2017-08-02 Marek Polacek <polacek@redhat.com>
15443 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15445 2017-08-02 Tom de Vries <tom@codesourcery.com>
15446 Cesar Philippidis <cesar@codesourcery.com>
15448 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15449 Add missing edge probabilities.
15451 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15453 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15454 Correct endianness.
15456 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15458 PR middle-end/79499
15459 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15460 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15461 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15462 prologue_seq sequences - if any.
15464 2017-08-02 Richard Biener <rguenther@suse.de>
15466 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15467 via vectors if supported, integer extracts via punning if supported
15468 or otherwise vector extracts.
15470 2017-08-02 Richard Biener <rguenther@suse.de>
15472 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15474 (bitmap_insert_into_set): ... this.
15476 2017-08-02 Richard Biener <rguenther@suse.de>
15478 PR tree-optimization/81633
15480 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15482 PR tree-optimization/71752
15483 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15485 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15487 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15488 (machine_function::call_ms2sysv_pad_out): Remove field.
15489 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15490 (ix86_compute_frame_layout): Likewise.
15492 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15495 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15496 attribute with interrupt attribute.
15498 2017-08-01 Andrew Pinski <apinski@cavium.com>
15500 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15501 BIT_INSERT_EXPR's operand 1
15502 to see if the types precision matches.
15504 2017-08-01 Martin Liska <mliska@suse.cz>
15506 PR middle-end/70140
15507 * builtins.c (expand_builtin_memcpy_args): Remove.
15508 (expand_builtin_memcpy): Call newly added function
15509 expand_builtin_memory_copy_args.
15510 (expand_builtin_memcpy_with_bounds): Likewise.
15511 (expand_builtin_mempcpy): Remove last argument.
15512 (expand_builtin_mempcpy_with_bounds): Likewise.
15513 (expand_builtin_memory_copy_args): New function created from
15514 expand_builtin_mempcpy_args with small modifications.
15515 (expand_builtin_mempcpy_args): Remove.
15516 (expand_builtin_stpcpy): Remove unused argument.
15517 (expand_builtin): Likewise.
15518 (expand_builtin_with_bounds): Likewise.
15520 2017-08-01 Martin Liska <mliska@suse.cz>
15523 Make mempcpy more optimal (PR middle-end/70140).
15525 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15528 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15529 __builtin_vec_cmpne verify both arguments are compatible vectors
15530 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15531 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15532 move computation of aligned to after checking the argument types.
15536 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15537 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15540 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15541 Doug Rupp <rupp@adacore.com>
15542 Olivier Hainque <hainque@adacore.com>
15544 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15545 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15547 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15548 for TARGET_OS_CPP_BUILTIN.
15549 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15550 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15552 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15553 passing required abi options to the assembler for EABI configurations.
15554 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15555 of .text.hot and .text.unlikely sections for kernel modules when
15556 using ARM style exceptions.
15557 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15558 options. Add EXTRA_CC1_SPEC.
15559 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15561 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15563 (ARM_TARGET2_DWARF_FORMAT): Define.
15564 * config/arm/t-vxworks: Adjust multilib control to removal of the
15565 Diab command line options.
15567 2017-08-01 Martin Liska <mliska@suse.cz>
15569 PR gcov-profile/81561
15570 * gcov.c (unblock): Make unblocking safe as we need to preserve
15571 index correspondence of blocks and block_lists.
15573 2017-08-01 Richard Biener <rguenther@suse.de>
15575 PR tree-optimization/81181
15576 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15577 (compute_antic): ... end of iteration here.
15579 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15581 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15582 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15583 (ftree-slp-vectorize): Likewise.
15584 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15585 can no longer be set independent of flag_tree_loop_vectorize.
15586 * omp-general.c (emp_max_vf): Likewise.
15587 * opts.c (enable_fdo_optimizations): Remove references to
15588 flag_tree_vectorize, these are now implicit.
15589 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15590 and leave it for the options machinery.
15592 2017-08-01 Martin Liska <mliska@suse.cz>
15594 PR middle-end/70140
15595 * builtins.c (expand_builtin_memcpy_args): Remove.
15596 (expand_builtin_memcpy): Call newly added function
15597 expand_builtin_memory_copy_args.
15598 (expand_builtin_memcpy_with_bounds): Likewise.
15599 (expand_builtin_mempcpy): Remove last argument.
15600 (expand_builtin_mempcpy_with_bounds): Likewise.
15601 (expand_builtin_memory_copy_args): New function created from
15602 expand_builtin_mempcpy_args with small modifications.
15603 (expand_builtin_mempcpy_args): Remove.
15604 (expand_builtin_stpcpy): Remove unused argument.
15605 (expand_builtin): Likewise.
15606 (expand_builtin_with_bounds): Likewise.
15608 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15611 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15612 print "ds:" only for immediates in generic address space.
15614 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15617 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15618 (ix86_function_ok_for_sibcall): Return false for naked functions.
15620 2017-08-01 Richard Biener <rguenther@suse.de>
15622 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15623 (compute_antic): Seed worklist with exit block predecessors.
15624 * cfganal.c (dfs_find_deadend): For a cycle return the source
15625 of the edge closing it.
15627 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15629 * config/aarch64/aarch64.c
15630 (aarch64_can_const_movi_rtx_p): Move 0 check.
15632 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15634 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15635 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15637 * match.pd: Ditto in address comparison pattern.
15639 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15641 PR tree-optimization/81627
15642 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15643 closed ssa form for store-store chain.
15645 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15647 PR tree-optimization/81620
15648 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15649 for store-store chain.
15651 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15653 PR tree-optimization/81588
15654 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15655 ranges[i].in_p, invert comparison code ccode. For >/>=,
15656 swap rhs1 and rhs2 and comparison code unconditionally,
15657 for </<= don't do that. Don't swap rhs1/rhs2 again if
15658 ranges[i].in_p, instead invert comparison code ccode if
15659 opcode or oe->rank is BIT_IOR_EXPR.
15662 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15663 a direct optab to conversion optab.
15664 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15665 with GET_MODE_INNER as last argument instead of optab_handler.
15666 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15667 vector extraction if possible and optab is available.
15668 * expr.c (store_constructor): Use convert_optab_handler instead
15669 of optab_handler. Use vector initialization from smaller
15670 vectors if possible and optab is available.
15671 * tree-vect-stmts.c (vectorizable_load): Likewise.
15672 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15673 now have two modes.
15674 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15675 of vec_init from half-sized vectors with the same element mode.
15676 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15677 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15678 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15679 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15680 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15681 after mode in gen_vec_extract* calls.
15682 (vec_extract<mode>): Renamed to ...
15683 (vec_extract<mode><ssescalarmodelower>): ... this.
15684 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15685 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15686 element mode after mode in gen_vec_init* calls.
15687 (VEC_INIT_HALF_MODE): New mode iterator.
15688 (vec_init<mode>): Renamed to ...
15689 (vec_init<mode><ssescalarmodelower>): ... this.
15690 (vec_init<mode><ssehalfvecmodelower>): New expander.
15691 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15692 (vec_extractv2sfsf): ... this.
15693 (vec_initv2sf): Renamed to ...
15694 (vec_initv2sfsf): ... this.
15695 (vec_extractv2si): Renamed to ...
15696 (vec_extractv2sisi): ... this.
15697 (vec_initv2si): Renamed to ...
15698 (vec_initv2sisi): ... this.
15699 (vec_extractv4hi): Renamed to ...
15700 (vec_extractv4hihi): ... this.
15701 (vec_initv4hi): Renamed to ...
15702 (vec_initv4hihi): ... this.
15703 (vec_extractv8qi): Renamed to ...
15704 (vec_extractv8qiqi): ... this.
15705 (vec_initv8qi): Renamed to ...
15706 (vec_initv8qiqi): ... this.
15707 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15708 (vec_init<mode>): Renamed to ...
15709 (vec_init<mode><VEC_base_l>): ... this.
15710 (vec_extract<mode>): Renamed to ...
15711 (vec_extract<mode><VEC_base_l>): ... this.
15712 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15713 (vec_initv2sfsf): ... this.
15714 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15715 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15716 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15717 element mode after mode in gen_vec_init* calls.
15718 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15719 (vec_init<mode><Vel>): ... this.
15720 (vec_extract<mode>): Renamed to ...
15721 (vec_extract<mode><Vel>): ... this.
15722 * config/aarch64/iterators.md (Vel): New mode attribute.
15723 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15724 Add element mode after mode in gen_vec_extract* calls.
15725 * config/s390/vector.md (non_vec_l): New mode attribute.
15726 (vec_extract<mode>): Renamed to ...
15727 (vec_extract<mode><non_vec_l>): ... this.
15728 (vec_init<mode>): Renamed to ...
15729 (vec_init<mode><non_vec_l>): ... this.
15730 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15731 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15733 * config/arm/iterators.md (V_elem_l): New mode attribute.
15734 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15735 (vec_extract<mode><V_elem_l>): ... this.
15736 (vec_extractv2di): Renamed to ...
15737 (vec_extractv2didi): ... this.
15738 (vec_init<mode>): Renamed to ...
15739 (vec_init<mode><V_elem_l>): ... this.
15740 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15741 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15742 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15743 Add element mode after gen_vec_extract* calls.
15744 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15745 (vec_init<mode><unitmode>): ... this.
15746 (vec_extract<mode>): Renamed to ...
15747 (vec_extract<mode><unitmode>): ... this.
15748 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15749 (vec_init<mode><unitmode>): ... this.
15750 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15751 (vec_initv2sfsf): ... this.
15752 (vec_extractv2sf): Renamed to ...
15753 (vec_extractv2sfsf): ... this.
15754 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15755 Add element mode after gen_vec_extract* calls.
15756 * config/mips/mips.md (unitmode): New mode iterator.
15757 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15758 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15759 * config/spu/spu.md (inner_l): New mode attribute.
15760 (vec_init<mode>): Renamed to ...
15761 (vec_init<mode><inner_l>): ... this.
15762 (vec_extract<mode>): Renamed to ...
15763 (vec_extract<mode><inner_l>): ... this.
15764 * config/sparc/sparc.md (veltmode): New mode iterator.
15765 (vec_init<VMALL:mode>): Renamed to ...
15766 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15767 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15768 (vec_initv2sisi): ... this.
15769 (vec_initv2sf): Renamed to ...
15770 (vec_initv2sfsf): ... this.
15771 (vec_extractv2sf): Renamed to ...
15772 (vec_extractv2sfsf): ... this.
15773 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15774 (vec_init<mode>): Renamed to ...
15775 (vec_init<mode><VEC_base_l>): ... this.
15776 (vec_extract<mode>): Renamed to ...
15777 (vec_extract<mode><VEC_base_l>): ... this.
15778 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15779 (vec_initv2sfsf): ... this.
15780 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15781 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15782 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15783 gen_vec_init* calls.
15785 2017-08-01 Richard Biener <rguenther@suse.de>
15787 PR tree-optimization/81297
15788 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15789 TREE_OVERFLOW from INTEGER_CSTs.
15791 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15793 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15795 2017-07-31 Carl Love <cel@us.ibm.com>
15797 * config/rs6000/rs6000-c: Add support for built-in functions
15798 vector signed char vec_xl_be (signed long long, signed char *);
15799 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15800 vector signed int vec_xl_be (signed long long, signed int *);
15801 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15802 vector signed long long vec_xl_be (signed long long, signed long long *);
15803 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15804 vector signed short vec_xl_be (signed long long, signed short *);
15805 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15806 vector double vec_xl_be (signed long long, double *);
15807 vector float vec_xl_be (signed long long, float *);
15808 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15809 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15810 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15812 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15813 (altivec_expand_builtin): Add switch statement to call
15814 altivec_expand_xl_be for each builtin.
15815 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15816 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15817 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15818 __builtin_vsx_le_be_v16qi.
15819 * doc/extend.texi: Update the built-in documentation file for the
15820 new built-in functions.
15822 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15825 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15827 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15829 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15831 * config.gcc: Add z14.
15832 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15833 CPU model numbers for z13s and z14.
15834 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15836 * config/s390/s390-opts.h (enum processor_type): Rename
15837 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15838 * config/s390/s390.c (processor_table): Add field for CPU name to
15839 be passed to Binutils.
15840 (s390_asm_output_machine_for_arch): Use the new field in
15841 processor_table for Binutils.
15842 (s390_expand_builtin): Replace arch12 with z14.
15843 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15844 (s390_get_sched_attrmask): Likewise.
15845 (s390_get_unit_mask): Likewise.
15846 * config/s390/s390.opt: Add z14 to processor_type enum.
15848 2017-07-31 Martin Jambor <mjambor@suse.cz>
15851 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15852 regardless of optimization level.
15854 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15855 Martin Liska <mliska@suse.cz>
15857 * predict.def: Remove old comment and adjust probability.
15858 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15859 PREDICT statements.
15861 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15864 * config/i386/i386.c (ix86_function_naked): New function.
15865 (ix86_can_use_return_insn_p): Return false for naked functions.
15866 (ix86_expand_prologue): Skip prologue for naked functions.
15867 (ix86_expand_epilogue): Skip epilogue for naked functions
15868 and emit trap instruction.
15869 (ix86_warn_func_return): New function.
15870 (ix86_attribute_table): Add "naked" attribute specification.
15871 (TARGET_WARN_FUNC_RETURN): Define.
15872 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15874 2017-07-31 Martin Liska <mliska@suse.cz>
15876 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15877 (dump_gimple_bb_header): Always dump BB info.
15878 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15880 2017-07-31 Martin Liska <mliska@suse.cz>
15883 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15884 also with current_function_decl non-null equality.
15886 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15889 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15890 change type to the element type, instead add eltype variable and
15891 use it where we are interested in the element type.
15893 PR tree-optimization/81603
15894 * ipa-polymorphic-call.c
15895 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15896 offset arithmetic in offset_int, bail out if the resulting bit offset
15897 doesn't fit into shwi.
15899 2017-07-31 Martin Liska <mliska@suse.cz>
15901 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15902 (gimplify_save_expr): Fix comment.
15904 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15907 * config/i386/i386.c (ix86_function_arg): Update arguments for
15909 (ix86_compute_frame_layout): Set the initial stack offset to
15910 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15911 INCOMING_FRAME_SP_OFFSET.
15912 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15913 stack before exception handler returns.
15914 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15915 the 'ERROR_CODE' for exception handler.
15917 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15919 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15920 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15921 (ASM_OUTPUT_REG_POP): Ditto.
15922 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15923 instead of asm_fprintf to output pure string.
15925 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15927 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15928 to imported_module_or_decl hook.
15929 (debug_nothing_tree_tree_tree_bool): Remove.
15930 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15931 * debug.c (do_nothing_debug_hooks): Use
15932 debug_nothing_tree_tree_tree_bool_bool instead of
15933 debug_nothing_tree_tree_tree_bool.
15934 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15935 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15936 * sdbout.c (sdb_debug_hooks): Likewise.
15937 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15938 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15940 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15941 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15942 attribute, don't add anything.
15944 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15946 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15947 (fold_build2_stat_loc): Likewise.
15948 (fold_build3_stat_loc): Likewise.
15949 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15950 (fold_build1_loc): Remove macro.
15951 (fold_build2_loc): Likewise.
15952 (fold_build3_loc): Likewise.
15954 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15956 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15957 (gimple_build_debug_bind_source_stat): Likewise.
15958 * gimple.h (gimple_build_debug_bind): Remove macro.
15959 (gimple_build_debug_bind_source): Likewise.
15961 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15963 * bitmap.c (bitmap_alloc): Adjust.
15964 (bitmap_gc_alloc): Likewise.
15965 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15967 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15969 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15970 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15971 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15972 (bitmap_gc_alloc_stat): Likewise.
15973 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15975 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15977 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15978 * rtl.h (shallow_copy_rtx): Remove macro.
15980 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15982 * emit-rtl.c (gen_raw_REG): Adjust.
15983 * gengenrtl.c (gendef): Likewise.
15984 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15985 * rtl.h (rtx_alloc): Remove macro.
15987 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15989 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15990 (build_tree_list_stat): Likewise.
15991 * tree.h (build_tree_list): Remove macro.
15992 (build_tree_list_vec): Likewise.
15994 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15996 * tree.c (make_vector_stat): Remove _stat from name.
15997 (build_vector_stat): Likewise.
15998 * tree.h (make_vector_stat): Remove macro.
15999 (build_vector_stat): Likewise.
16001 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16003 * tree.h (build_var_debug_value): Remove prototype.
16005 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16007 * tree.c (tree_cons_stat): Remove _stat from name.
16008 * tree.h (tree_cons): Remove macro.
16010 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16012 * tree.c (build_vl_exp_stat): Remove _stat from name.
16013 * tree.h (build_vl_exp): Remove macro.
16015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16017 * tree.c (build_decl_stat): Remove _stat from name.
16018 * tree.h (build_decl): Remove macro.
16020 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16022 * gimple.c (gimple_build_with_ops_stat): Adjust.
16023 (gimple_alloc_stat): Remove _stat from name.
16024 * gimple.h (gimple_alloc): Remove macro.
16026 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16028 * tree.c (make_tree_vec_stat): Remove _stat from name.
16029 (grow_tree_vec_stat): Likewise.
16030 * tree.h (make_tree_vec_stat): Adjust prototype.
16031 (grow_tree_vec_stat): Likewise.
16032 (make_tree_vec): Remove macro.
16033 (grow_tree_vec): Likewise.
16035 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16037 * fold-const.c (fold_build1_stat_loc): Adjust.
16038 (fold_build2_stat_loc): Likewise.
16039 (fold_build3_stat_loc): Likewise.
16040 * tree.c (build0_stat): Remove _stat from name.
16041 (build1_stat): Likewise.
16042 (build2_stat): Likewise.
16043 (build3_stat): Likewise.
16044 (build4_stat): Likewise.
16045 (build5_stat): Likewise.
16046 * tree.h (build1_loc): Remove macro, and rename _stat function
16048 (build2_loc): Likewise.
16049 (build3_loc): Likewise.
16050 (build4_loc): Likewise.
16051 (build5_loc): Likewise.
16053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16055 * tree.c (make_int_cst_stat): Remove _stat from name.
16056 * tree.h (make_int_cst_stat): Adjust prototype.
16057 (make_int_cst): Remove macro.
16059 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16061 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16062 * tree.h (make_tree_binfo_stat): Adjust prototype.
16063 (make_tree_binfo): Remove.
16065 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16067 * tree.c (copy_node_stat): Rename to copy_node.
16068 (build_distinct_type_copy): Adjust.
16069 * tree.h (copy_node_stat): Adjust prototype.
16070 (copy_node): Remove macro.
16072 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16074 * tree.c (make_node_stat): rename to make_node.
16075 (build_tree_list_stat): Adjust.
16076 (build0_stat): Likewise.
16077 (build2_stat): Likewise.
16078 (build3_stat): Likewise.
16079 (build4_stat): Likewise.
16080 (build5_stat): Likewise.
16081 (build_decl_stat): Likewise.
16082 * tree.h (make_node_stat): Adjust prototype.
16083 (make_node): remove macro.
16085 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16087 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16088 (PPC_FEATURE2_SCV): Likewise.
16089 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16091 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16093 * config/aarch64/aarch64.c
16094 (aarch64_internal_mov_immediate): Add new special pattern.
16095 * config/aarch64/aarch64.md (*movdi_aarch64):
16096 Add reg/32bit const mov case.
16098 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16099 Richard Sandiford <richard.sandiford@linaro.org>
16101 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16102 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16103 Add integer and movi cases.
16104 (movi-split-hf-df-sf split, fp16): New.
16105 (enabled): Added TARGET_FP_F16INST.
16106 * config/aarch64/iterators.md (GPF_HF): New.
16107 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16109 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16111 * config/aarch64/aarch64.c
16112 (aarch64_simd_container_mode): Add prototype.
16113 (aarch64_expand_mov_immediate): Add HI support.
16114 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16115 (aarch64_can_const_movi_rtx_p): New.
16116 (aarch64_preferred_reload_class):
16117 Remove restrictions of using FP registers for certain SIMD operations.
16118 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16119 (aarch64_valid_floating_const): Add integer move validation.
16120 (aarch64_simd_imm_scalar_p): Remove.
16121 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16122 (aarch64_legitimate_constant_p): Expand list of supported cases.
16123 * config/aarch64/aarch64-protos.h
16124 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16125 (aarch64_reinterpret_float_as_int): New.
16126 (aarch64_simd_imm_scalar_p): Remove.
16127 * config/aarch64/constraints.md (Uvi): New.
16128 (Dd): Split into Ds and new Dd.
16129 * config/aarch64/aarch64.md (*movsi_aarch64):
16131 (*movdi_aarch64): Add SIMD mov case.
16133 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16135 * tree-predcom.c: (struct chain): Handle store-store chain in which
16136 stores for elimination only store loop invariant values.
16137 (execute_pred_commoning_chain): Ditto.
16138 (prepare_initializers_chain_store_elim): Ditto.
16139 (prepare_finalizers): Ditto.
16140 (is_inv_store_elimination_chain): New function.
16141 (initialize_root_vars_store_elim_1): New function.
16143 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16145 * tree-predcom.c: Revise general description of the pass.
16146 (enum chain_type): New enum type for store elimination.
16147 (struct chain): New field supporting store elimination.
16148 (struct component): Ditto.
16149 (dump_chain): Dump store-stores chain.
16150 (release_chain): Release resources.
16151 (split_data_refs_to_components): Compute and create component
16152 contains only stores for elimination.
16153 (get_chain_last_ref_at): New function.
16154 (make_invariant_chain): Initialization.
16155 (make_rooted_chain): Specify chain type in parameter and record it.
16156 (add_looparound_copies): Skip for store-stores chain.
16157 (determine_roots_comp): Compute type of chain and pass it to
16159 (initialize_root_vars_store_elim_2): New function.
16160 (finalize_eliminated_stores): New function.
16161 (remove_stmt): Handle store for elimination.
16162 (execute_pred_commoning_chain): Execute predictive commoning on
16163 store-store chains.
16164 (determine_unroll_factor): Skip unroll for store-stores chain.
16165 (prepare_initializers_chain_store_elim): New function.
16166 (prepare_initializers_chain): Hanlde store-store chain.
16167 (prepare_finalizers_chain, prepare_finalizers): New function.
16168 (tree_predictive_commoning_loop): Return integer value indicating
16169 if loop is unrolled or lcssa form is corrupted.
16170 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16172 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16174 * tree-predcom.c (initialize_root): Delete.
16175 (execute_pred_commoning_chain): Initialize root vars and replace
16176 reference of non-combined chain directly, rather than call above
16179 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16181 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16182 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16184 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16186 * tree-predcom.c (struct chain): New field init_seq.
16187 (release_chain): Release init_seq.
16188 (prepare_initializers_chain): Record intialization stmts in above
16190 (insert_init_seqs): New function.
16191 (tree_predictive_commoning_loop): Call insert_init_seqs.
16193 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16195 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16197 2017-07-28 Richard Biener <rguenther@suse.de>
16199 * match.pd: Remove superfluous :c.
16200 * genmatch.c (simplify::id): Add member.
16201 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16203 (current_id): New global.
16204 (dt_node::parent): Move from ...
16205 (dt_operand::parent): ... here. Add for_id member.
16206 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16207 (decision_tree::find_node): Relax order requirement when
16208 merging DT_TRUE nodes to ones inbetween the current simplify
16209 and the one we try to merge with. Add diagnostic whenever
16210 we need to enforce pattern order by not merging.
16211 (decision_tree::insert): Set current_id.
16212 (decision_tree::print_node): Dump parent node and for_id.
16213 (parser::last_id): Add member.
16214 (parser::push_simplify): Assign unique id.
16215 (parser::parser): Initialize last_id.
16217 2017-07-28 Martin Liska <mliska@suse.cz>
16220 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16221 gimple_build_debug_bind.
16223 2017-07-28 Richard Biener <rguenther@suse.de>
16225 PR tree-optimization/81502
16226 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16228 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16230 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16231 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16232 for BIT_FIELD_REF args.
16233 * fold-const.c (make_bit_field_ref): Likewise.
16234 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16236 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16239 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16240 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16241 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16242 Or it into SANITIZER_UNDEFINED.
16243 * ubsan.c: Include gimple-fold.h and varasm.h.
16244 (ubsan_expand_ptr_ifn): New function.
16245 (instrument_pointer_overflow): New function.
16246 (maybe_instrument_pointer_overflow): New function.
16247 (instrument_object_size): Formatting fix.
16248 (pass_ubsan::execute): Call instrument_pointer_overflow
16249 and maybe_instrument_pointer_overflow.
16250 * internal-fn.c (expand_UBSAN_PTR): New function.
16251 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16252 * sanitizer.def (__ubsan_handle_pointer_overflow,
16253 __ubsan_handle_pointer_overflow_abort): New builtins.
16254 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16255 * internal-fn.def (UBSAN_PTR): New internal function.
16256 * opts.c (sanitizer_opts): Add pointer-overflow.
16257 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16258 * fold-const.c (build_range_check): Compute pointer range check in
16259 integral type if pointer arithmetics would be needed. Formatting
16262 2017-07-28 Martin Liska <mliska@suse.cz>
16265 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16266 parameters that are of a variable-length.
16268 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16270 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16272 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16273 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16274 (CRT_CALL_STATIC_FUNCTION): Likewise.
16275 (ASM_DEFAULT_SPEC): New define.
16276 (ASM_SPEC32): Likewise.
16277 (ASM_SPEC64): Likewise.
16278 (ASM_SPEC_COMMON): Likewise.
16279 (ASM_SPEC): Likewise.
16280 (INVALID_64BIT): Likewise.
16281 (LINK_OS_DEFAULT_SPEC): Likewise.
16282 (LINK_OS_SPEC32): Likewise.
16283 (LINK_OS_SPEC64): Likewise.
16284 (POWERPC_LINUX): Likewise.
16285 (PTRDIFF_TYPE): Likewise.
16286 (RESTORE_FP_PREFIX): Likewise.
16287 (RESTORE_FP_SUFFIX): Likewise.
16288 (SAVE_FP_PREFIX): Likewise.
16289 (SAVE_FP_SUFFIX): Likewise.
16290 (SIZE_TYPE): Likewise.
16291 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16292 (TARGET_64BIT): Likewise.
16293 (TARGET_64BIT): Likewise.
16294 (TARGET_AIX): Likewise.
16295 (WCHAR_TYPE_SIZE): Likewise.
16296 (WCHAR_TYPE): Undefine.
16297 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16298 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16299 (CPP_OS_RTEMS_SPEC): Delete.
16300 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16301 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16303 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16305 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16307 PR tree-optimization/81578
16308 * tree-parloops.c (build_new_reduction): Bail out if
16309 reduction_code isn't one of the standard OpenMP reductions.
16310 Move the details printing after that decision.
16312 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16314 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16315 related to reload_in_progress.
16316 (splat_input_operand): Likewise.
16317 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16319 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16321 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16322 (TARGET_INSTANTIATE_DECLS): Likewise.
16323 (legitimate_indexed_address_p): Delete reload_in_progress code.
16324 (rs6000_debug_legitimate_address_p): Likewise.
16325 (rs6000_eliminate_indexed_memrefs): Likewise.
16326 (rs6000_emit_le_vsx_store): Likewise.
16327 (rs6000_emit_move_si_sf_subreg): Likewise.
16328 (rs6000_emit_move): Likewise.
16329 (register_to_reg_type): Likewise.
16330 (rs6000_pre_atomic_barrier): Likewise.
16331 (rs6000_machopic_legitimize_pic_address): Likewise.
16332 (rs6000_allocate_stack_temp): Likewise.
16333 (rs6000_address_for_fpconvert): Likewise.
16334 (rs6000_address_for_altivec): Likewise.
16335 (rs6000_secondary_memory_needed_rtx): Delete function.
16336 (rs6000_check_sdmode): Likewise.
16337 (rs6000_alloc_sdmode_stack_slot): Likewise.
16338 (rs6000_instantiate_decls): Likewise.
16339 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16340 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16341 Delete reload_in_progress.
16342 (*vec_reload_and_plus_<mptrsize>): Likewise.
16343 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16344 (vsx_div_v2di): Likewise.
16345 (vsx_udiv_v2di): Likewise.
16347 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16349 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16350 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16351 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16352 (rs6000_debug_reg_global): Delete print of LRA status.
16353 (rs6000_option_override_internal): Delete dead LRA related code.
16354 (rs6000_lra_p): Delete function.
16355 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16357 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16359 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16360 * config/riscv/rtems.h: New file.
16362 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16363 Sudakshina Das <sudi.das@arm.com>
16365 * config/aarch64/aarch64.md
16366 (define_split for and<mode>3nr_compare): Move
16367 non aarch64_logical_operand to a register.
16368 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16369 register immediate operand to a register.
16370 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16372 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16374 PR middle-end/81564
16375 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16377 2017-07-27 Richard Biener <rguenther@suse.de>
16379 PR tree-optimization/81573
16380 PR tree-optimization/81494
16381 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16382 multi defuse cycle case.
16384 2017-07-27 Richard Biener <rguenther@suse.de>
16386 PR tree-optimization/81571
16387 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16390 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16392 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16393 earlier and only if MASK_FPU is set. Adjust formatting.
16395 2017-07-27 Martin Liska <mliska@suse.cz>
16397 * opt-functions.awk: Add validation of value of Init.
16398 * optc-gen.awk: Pass new argument.
16400 2017-07-27 Martin Liska <mliska@suse.cz>
16402 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16403 Fix wrong condition.
16405 2017-07-27 Martin Liska <mliska@suse.cz>
16407 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16408 BBs and edges seen by autoFDO.
16410 2017-07-27 Richard Biener <rguenther@suse.de>
16412 PR tree-optimization/81502
16413 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16414 with incompatible but same sized type.
16415 (execute_update_addresses_taken): Likewise.
16417 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16419 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16420 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16422 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16425 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16426 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16427 Change s_operand to memory_operand.
16429 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16431 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16432 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16433 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16434 Emit instructions rather than returning an expression. Handle TFmode
16435 and KFmode by casting to TImode.
16436 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16437 (rs6000_emit_le_vsx_store): Likewise.
16438 * config/rs6000/vsx.md (VSX_TI): New iterator.
16439 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16440 (*vsx_le_undo_permute_<mode>): Likewise.
16441 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16442 emit the split sequence.
16443 (*vsx_le_perm_store_<mode>): Likewise.
16445 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16447 PR tree-optimization/81555
16448 PR tree-optimization/81556
16449 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16450 if true, force CHANGED for the recursive invocation.
16451 (reassociate_bb): Remember original length of ops array, pass
16452 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16454 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16455 attributes for noipa attribute. For naked attribute use
16456 lookup_attribute first before lookup_attribute_spec.
16457 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16459 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16460 for functions with noipa attribute.
16461 (cgraph_externally_visible_p): Return true for functions with noipa
16463 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16464 for functions with noipa attribute.
16465 * doc/extend.texi: Document noipa function attribute.
16466 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16467 also for functions with noipa attribute.
16468 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16470 2017-07-26 Andrew Pinski <apinski@cavium.com>
16472 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16473 vec_unalign_load_cost and vec_unalign_store_cost.
16475 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16477 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16478 -mvsx-small-integer option.
16479 (ISA_3_0_MASKS_IEEE): Likewise.
16480 (OTHER_VSX_VECTOR_MASKS): Likewise.
16481 (POWERPC_MASKS): Likewise.
16482 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16483 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16484 code, only testing for DImode being allowed in non-VSX floating
16486 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16487 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16489 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16490 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16491 TARGET_P8_VECTOR test.
16492 (rs6000_secondary_reload_simple_move): Likewise.
16493 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16494 since TARGET_P9_VECTOR was already tested.
16495 (rs6000_opt_masks): Remove -mvsx-small-integer.
16496 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16497 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16499 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16500 test for TARGET_VEXTRACTUB was used, and that uses
16502 (p9 extract splitter): Likewise.
16503 (vsx_extract_<mode>_di_p9): Likewise.
16504 (vsx_extract_<mode>_store_p9): Likewise.
16505 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16506 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16507 the elimination of TARGET_VSX_SMALL_INTEGER.
16508 (vsx_extract_<mode>_p8): Likewise.
16509 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16510 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16511 (vsx_set_<mode>_p9): Likewise.
16512 (vsx_set_v4sf_p9): Likewise.
16513 (vsx_set_v4sf_p9_zero): Likewise.
16514 (vsx_insert_extract_v4sf_p9): Likewise.
16515 (vsx_insert_extract_v4sf_p9_2): Likewise.
16516 * config/rs6000/rs6000.md (sign extend splitter): Change
16517 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16518 (floatsi<mode>2_lfiwax_mem): Likewise.
16519 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16520 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16521 since a test for TARGET_P9_VECTOR was used.
16522 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16523 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16524 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16525 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16526 TARGET_P8_VECTOR test.
16527 (fix_trunc<mode>si2_stfiwx): Likewise.
16528 (fix_trunc<mode>si2_internal): Likewise.
16529 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16530 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16532 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16533 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16534 TARGET_P8_VECTOR test.
16535 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16536 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16537 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16539 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16540 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16541 since a test for TARGET_P9_VECTOR was used.
16542 (splitter for loading small constants): Likewise.
16544 2017-07-26 Andrew Pinski <apinski@cavium.com>
16546 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16549 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16552 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16553 (fp_valid_at): Likewise.
16555 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16557 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16558 (qdf24xx_addrcost_table): Likewise.
16559 (cortexa57_tunings): Update to use generic_branch_cost.
16560 (cortexa72_tunings): Likewise.
16561 (cortexa73_tunings): Likewise.
16562 (qdf24xx_tunings): Likewise.
16564 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16566 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16567 (thunderx2t99_branch_cost): Likewise.
16568 (cortexa35_tunings): Update to use generic_branch_cost.
16569 (cortexa53_tunings): Likewise.
16570 (cortexa57_tunings): Likewise.
16571 (cortexa72_tunings): Likewise.
16572 (cortexa73_tunings): Likewise.
16573 (thunderx2t99_tunings): Likewise.
16575 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16577 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16578 (sparc_option_override): Honour MASK_FSMULD.
16579 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16580 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16581 * config/sparc/sparc.opt (mfsmuld): New option.
16582 * doc/invoke.texi (mfsmuld): Document option.
16584 2017-07-26 Marek Polacek <polacek@redhat.com>
16586 PR middle-end/70992
16587 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16589 2017-07-26 Richard Biener <rguenther@suse.de>
16591 * gimple-match-head.c (do_valueize): Return OP if valueize
16593 (get_def): New helper to get at the def stmt of a SSA name
16594 if valueize allows.
16595 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16596 do_valueize to get at the def stmt.
16597 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16599 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16601 PR middle-end/46932
16602 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16604 2017-07-26 Martin Liska <mliska@suse.cz>
16607 * function.c (expand_function_start): Make expansion of
16608 nonlocal_goto_save_area after parm_birth_insn.
16610 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16612 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16613 from all CPU target flags enable members.
16615 2017-07-26 Richard Biener <rguenther@suse.de>
16617 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16618 (decision_tree::gen): Make 'type' const.
16619 (write_predicate): Likewise.
16621 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16623 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16624 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16625 (rs6000_option_override_internal): Likewise.
16626 (rs6000_expand_vector_set): Likewise.
16627 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16628 (TARGET_UPPER_REGS_SF): Likewise.
16629 (TARGET_UPPER_REGS_DI): Likewise.
16630 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16631 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16632 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16633 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16634 (Splitters for DI constants in Altivec registers): Likewise.
16635 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16636 (vsx_set_v4sf_p9): Likewise.
16637 (vsx_set_v4sf_p9_zero): Likewise.
16638 (vsx_insert_extract_v4sf_p9): Likewise.
16639 (vsx_insert_extract_v4sf_p9_2): Likewise.
16641 2017-07-25 Carl Love <cel@us.ibm.com>
16643 * doc/extend.texi: Update the built-in documentation file for the
16644 existing built-in functions
16645 vector signed char vec_cnttz (vector signed char);
16646 vector unsigned char vec_cnttz (vector unsigned char);
16647 vector signed short vec_cnttz (vector signed short);
16648 vector unsigned short vec_cnttz (vector unsigned short);
16649 vector signed int vec_cnttz (vector signed int);
16650 vector unsigned int vec_cnttz (vector unsigned int);
16651 vector signed long long vec_cnttz (vector signed long long);
16652 vector unsigned long long vec_cnttz (vector unsigned long long);
16654 2017-07-25 Andrew Pinski <apinski@cavium.com>
16656 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16657 accesses where the use is for the first operand of a BIT_INSERT.
16659 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16662 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16663 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16665 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16667 * config/i386/gstabs.h: Delete.
16668 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16670 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16672 * config/i386/i386.c (ix86_decompose_address): Do not check for
16673 register RTX when looking at index_reg or base_reg.
16674 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16676 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16678 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16679 to update EH info here.
16681 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16683 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16685 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16687 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16689 2017-07-25 Torsten Duwe <duwe@suse.de>
16691 * common.opt: Introduce -fpatchable-function-entry
16692 command line option, and its variables function_entry_patch_area_size
16693 and function_entry_patch_area_start.
16694 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16695 including a two-value parser.
16696 * target.def (print_patchable_function_entry): New target hook.
16697 * targhooks.h (default_print_patchable_function_entry): New function.
16698 * targhooks.c (default_print_patchable_function_entry): Likewise.
16699 * toplev.c (process_options): Switch off IPA-RA if
16700 patchable function entries are being generated.
16701 * varasm.c (assemble_start_function): Look at the
16702 patchable-function-entry command line switch and current
16703 function attributes and maybe generate NOP instructions by
16704 calling the print_patchable_function_entry hook.
16705 * doc/extend.texi: Document patchable_function_entry attribute.
16706 * doc/invoke.texi: Document -fpatchable_function_entry
16707 command line option.
16708 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16710 * doc/tm.texi: Re-generate.
16712 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16715 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16716 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16718 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16720 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16722 2017-07-25 Richard Biener <rguenther@suse.de>
16724 PR tree-optimization/81455
16725 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16726 not walk in cycles when looking for guards.
16728 2017-07-25 Richard Biener <rguenther@suse.de>
16730 PR tree-optimization/81529
16731 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16732 when optimizing backedge uses.
16734 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16736 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16738 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16739 to dl_section_ref. On AIX, append an expression to subtract
16740 the size of the section length to dl_section_ref.
16742 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16744 * configure.ac: If any of the config.* scripts fail, exit 1.
16745 * configure: Regenerate.
16747 2017-07-25 Richard Biener <rguenther@suse.de>
16749 PR middle-end/81546
16750 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16751 of immediate uses, be more verbose on errors.
16753 2017-07-25 Richard Biener <rguenther@suse.de>
16755 PR tree-optimization/81510
16756 * tree-vect-loop.c (vect_is_simple_reduction): When the
16757 reduction stmt is not inside the loop bail out.
16759 2017-07-25 Richard Biener <rguenther@suse.de>
16761 PR tree-optimization/81303
16762 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16763 profitability check against LOOP_VINFO_NITERSM1.
16765 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16767 * domwalk.c (cmp_bb_postorder): Simplify.
16768 (sort_bbs_postorder): New function. Use it...
16769 (dom_walker::walk): ...here to optimize common cases.
16771 2017-07-25 Martin Liska <mliska@suse.cz>
16774 * ipa-visibility.c (function_and_variable_visibility): Make the
16775 redirection just on target that supports aliasing.
16776 Fix GNU coding style.
16778 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16781 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16783 * config/arm/rtems.h: Likewise.
16784 * config/bfin/rtems.h: Likewise.
16785 * config/i386/rtemself.h: Likewise.
16786 * config/lm32/rtems.h: Likewise.
16787 * config/m32c/rtems.h: Likewise.
16788 * config/m68k/rtemself.h: Likewise.
16789 * config/microblaze/rtems.h: Likewise.
16790 * config/mips/rtems.h: Likewise.
16791 * config/moxie/rtems.h: Likewise.
16792 * config/nios2/rtems.h: Likewise.
16793 * config/powerpcspe/rtems.h: Likewise.
16794 * config/rs6000/rtems.h: Likewise.
16795 * config/rtems.h: Likewise.
16796 * config/sh/rtems.h: Likewise.
16797 * config/sh/rtemself.h: Likewise.
16798 * config/sparc/rtemself.h: Likewise.
16800 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16803 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16804 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16805 * tree-ssa-structalias.c (alias_get_name): Same.
16807 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16810 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16811 instructions if no du chain is found.
16813 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16815 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16817 2017-07-25 Richard Biener <rguenther@suse.de>
16819 PR middle-end/81505
16820 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16823 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16825 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16826 upper-regs options.
16827 (ISA_2_7_MASKS_SERVER): Likewise.
16828 (ISA_3_0_MASKS_IEEE): Likewise.
16829 (OTHER_P8_VECTOR_MASKS): Likewise.
16830 (OTHER_VSX_VECTOR_MASKS): Likewise.
16831 (POWERPC_MASKS): Likewise.
16832 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16833 duplicate list of options.
16834 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16835 explicit -mupper-regs options.
16836 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16837 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16838 alias for -mupper-regs-df.
16839 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16840 (rs6000_init_hard_regno_mode_ok): Likewise.
16841 (rs6000_option_override_internal): Likewise.
16842 (rs6000_opt_masks): Likewise.
16843 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16844 options in terms of whether -mvsx or -mpower8-vector was used.
16845 (TARGET_UPPER_REGS_DI): Likewise.
16846 (TARGET_UPPER_REGS_SF): Likewise.
16847 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16848 -mupper-regs-* options.
16850 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16852 * passes.c (emergency_dump_function): Print some empty lines and a
16853 header before the RTL dump.
16855 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16857 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16859 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16862 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16863 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16865 2017-07-24 Carl Love <cel@us.ibm.com>
16867 * config/rs6000/rs6000-c.c: Add support for built-in functions
16868 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16869 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16870 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16871 vec_extract_fp_from_shortl): Add defines for the two builtins.
16872 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16873 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16875 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16876 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16877 * doc/extend.texi: Update the built-in documentation file for the
16878 new built-in function.
16880 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16883 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16885 * doc/generic.texi: Likewise.
16886 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16887 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16889 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16891 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16892 (aarch64_mls_elt_merge<mode>): Likewise.
16894 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16896 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16897 having __cxa_atexit.
16899 2017-07-23 Michael Collison <michael.collison@arm.com>
16901 * config/arm/arm.c (arm_option_override): Deprecate
16902 use of -mstructure-size-boundary.
16903 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16904 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16906 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16909 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16910 Reduce cost estimate for direct moves.
16912 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16915 * config/i386/i386.c (ix86_option_override_internal): Disable
16916 BMI, BMI2 and TBM instructions for -m16.
16918 2017-07-21 Carl Love <cel@us.ibm.com>
16920 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16921 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16922 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16923 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16924 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16925 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16926 VMULOSW): New enum "unspec" values.
16927 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16928 altivec_vmulosw): New patterns.
16929 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16930 VMULOSW): Add definitions.
16932 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16934 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16935 (qdf24xx): Likewise.
16936 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16937 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16938 (AARCH64_FL_V8_1): Renumber.
16939 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16940 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16941 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16942 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16943 rdma to feature modifiers list.
16945 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16947 PR middle-end/56727
16948 * ipa-visibility (function_and_variable_visibility): Convert
16949 recursive PLT call to direct call if appropriate.
16951 2017-07-21 Andrew Pinski <apinski@cavium.com>
16953 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16954 operand 1 to see if the types precision matches.
16955 * fold-const.c (operand_equal_p): Likewise.
16957 2017-07-21 Richard Biener <rguenther@suse.de>
16959 PR tree-optimization/81303
16960 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16961 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16962 (vect_peeling_hash_get_lowest_cost): Adjust.
16963 (vect_enhance_data_refs_alignment): Likewise. Use
16964 vect_get_peeling_costs_all_drs to compute the penalty for no
16965 peeling to match up costs.
16967 2017-07-21 Richard Biener <rguenther@suse.de>
16969 PR tree-optimization/81500
16970 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16971 we didn't identify a reduction path.
16973 2017-07-21 Tom de Vries <tom@codesourcery.com>
16974 Cesar Philippidis <cesar@codesourcery.com>
16976 PR gcov-profile/81442
16977 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16980 2017-07-21 Tom de Vries <tom@codesourcery.com>
16983 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16985 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16986 nvptx_override_options_after_change.
16988 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16990 * dwarf2out.c (output_file_names): Avoid double testing for
16991 dwarf_version >= 5.
16993 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16995 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16997 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16999 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17001 (try_crossjump_to_edge): Do not punt on partitioned functions.
17003 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17005 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17006 Put all BBs reachable only via paths crossing cold region to cold
17008 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17010 2016-07-21 Richard Biener <rguenther@suse.de>
17012 PR tree-optimization/81303
17013 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17014 into account prologue and epilogue iterations when raising
17015 min_profitable_iters to sth at least covering one vector iteration.
17017 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17019 * config/arm/arm.c (arm_test_cpu_arch_dat):
17022 2017-07-20 Nathan Sidwell <nathan@acm.org>
17024 Remove TYPE_METHODS.
17025 * tree.h (TYPE_METHODS): Delete.
17026 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17027 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17028 (dbxout_type_methods): Scan TYPE_FIELDS.
17029 (dbxout_type): Don't check TYPE_METHODS here.
17030 * function.c (use_register_for_decl): Always ignore register for
17031 class types when not optimizing.
17032 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17033 * tree.c (free_lang_data_in_type): Stitch out member functions and
17034 templates from TYPE_FIELDS.
17035 (build_distinct_type_copy, verify_type_variant,
17036 verify_type): Member fns are on TYPE_FIELDS.
17037 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17038 * tree-pretty-print.c (dump_generic_node): Likewise.
17040 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17043 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17044 V2TImode and V4TImode.
17045 (ix86_expand_vector_extract): Likewise.
17046 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17047 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17048 (ssescalarmode): Handle V4TImode and V2TImode.
17049 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17050 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17051 (VEXTRACTI128_MODE): New mode iterator.
17052 (splitter for *vec_extractv?ti first element): New.
17053 (VEC_INIT_MODE): New mode iterator.
17054 (vec_init<mode>): Consolidate 3 expanders into one using
17055 VEC_INIT_MODE mode iterator.
17057 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17059 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17060 non_spilled_static_chain_regno_p.
17062 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17064 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17066 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17068 * bb-reorder.c (connect_traces): Allow copying of blocks within
17071 2017-07-20 Richard Biener <rguenther@suse.de>
17073 * gimple.h (gimple_phi_result): Add gphi * overload.
17074 (gimple_phi_result_ptr): Likewise.
17075 (gimple_phi_arg): Likewise. Adjust index assert to only
17076 allow actual argument accesses rather than all slots available
17078 (gimple_phi_arg_def): Add gphi * overload.
17079 * tree-phinodes.c (make_phi_node): Initialize only actual
17081 (resize_phi_node): Clear memory not covered by old node,
17082 do not initialize excess argument slots.
17083 (reserve_phi_args_for_new_edge): Initialize new argument slot
17086 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17088 PR tree-optimization/81388
17090 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17092 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17093 by removing computation of may_be_zero.
17095 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17096 Tom de Vries <tom@codesourcery.com>
17098 PR middle-end/81030
17099 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17100 when gimple level profile disagrees with what RTL expander did.
17102 2017-07-20 Richard Biener <rguenther@suse.de>
17104 PR tree-optimization/61171
17105 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17106 (vect_analyze_stmt): Add slp instance parameter.
17107 (vectorizable_reduction): Likewise.
17108 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17109 (vect_is_simple_reduction): Deal with chains not detected
17110 as SLP reduction chain, specifically not properly associated
17111 chains containing a mix of plus/minus.
17112 (get_reduction_op): Remove.
17113 (get_initial_defs_for_reduction): Simplify, pass in whether
17114 this is a reduction chain, pass in the SLP node for the PHIs.
17115 (vect_create_epilog_for_reduction): Get the SLP instance as
17117 (vectorizable_reduction): Get the SLP instance as arg.
17118 During analysis remember the SLP node with the PHIs in the
17119 instance. Simplify getting at the vectorized reduction PHIs.
17120 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17121 through SLP instance.
17122 (vect_slp_analyze_operations): Likewise.
17123 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17124 (vect_transform_stmt): Likewise.
17126 2017-07-20 Tom de Vries <tom@codesourcery.com>
17128 PR tree-optimization/81489
17129 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17130 read of phi arg location to before loop that modifies phi.
17132 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17134 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17137 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17139 PR middle-end/81331
17140 * except.c (execute): Fix ordering issue.
17142 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17144 PR rtl-optimization/81423
17145 * combine.c (make_compound_operation_int): Don't try to optimize
17146 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17148 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17150 PR rtl-optimization/81423
17151 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17152 with a constant that is -1 in the truncated to mode.
17154 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17156 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17157 (determine_unlikely_bbs): ... here.
17158 * predict.h (propagate_unlikely_bbs_forward): Declare.
17159 * cfgexpand.c (pass_expand::execute): Use it.
17160 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17162 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17163 propagate_unlikely_bbs_forward.
17165 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17167 PR middle-end/81331
17168 * except.c (maybe_add_nop_after_section_switch): New function.
17171 2017-07-19 Tom de Vries <tom@codesourcery.com>
17173 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17175 2017-07-19 Tom de Vries <tom@codesourcery.com>
17177 * gimple.h (gimple_phi_arg): Make assert more strict.
17179 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17181 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17182 * config/rs6000/mmintrin.h: New file.
17183 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17185 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17187 PR tree-optimization/81346
17188 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17190 2017-07-19 Tom de Vries <tom@codesourcery.com>
17192 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17194 2017-07-19 Tom de Vries <tom@codesourcery.com>
17196 * config/nvptx/nvptx-modes.def: Add V2DImode.
17197 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17198 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17199 (nvptx_output_mov_insn): Handle lack of mov.b128.
17200 (nvptx_print_operand): Handle 'H' and 'L' codes.
17201 (nvptx_vector_mode_supported): Allow V2DImode.
17202 (nvptx_preferred_simd_mode): New function.
17203 (nvptx_data_alignment): New function.
17204 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17205 nvptx_preferred_simd_mode.
17206 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17208 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17210 2017-07-19 Tom de Vries <tom@codesourcery.com>
17212 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17213 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17214 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17215 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17216 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17217 (mov<VECIM>_insn): New define_insn.
17218 (define_expand "mov<VECIM>): New define_expand.
17220 2017-07-19 Tom de Vries <tom@codesourcery.com>
17222 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17224 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17226 PR tree-optimization/81346
17227 * fold-const.h (fold_div_compare, range_check_type): Declare.
17228 * fold-const.c (range_check_type): New function.
17229 (build_range_check): Use range_check_type.
17230 (fold_div_compare): No longer static, rewritten into
17231 a match.pd helper function.
17232 (fold_comparison): Don't call fold_div_compare here.
17233 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17234 as helper function.
17236 2017-07-19 Nathan Sidwell <nathan@acm.org>
17238 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17239 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17240 * tree.c (find_decls_types_r, verify_type): Use
17241 TYPE_{MIN,MAX}_VALUE_RAW.
17242 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17243 (hash_tree): Likewise.
17244 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17246 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17249 2017-07-18 Tom de Vries <tom@codesourcery.com>
17251 PR middle-end/81464
17252 * omp-expand.c (expand_omp_for_static_chunk): Handle
17253 equal-argument loop exit phi.
17255 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17258 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17259 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17260 operand 2 predicate.
17261 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17262 operand 2 predicate.
17263 (ror,rol -> rorx splitters): Use const_int_operand as
17264 operand 2 predicate.
17266 2017-06-18 Richard Biener <rguenther@suse.de>
17268 PR tree-optimization/81410
17269 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17270 the gap in the ! slp_perm SLP case after each group.
17272 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17274 PR middle-end/81463
17275 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17278 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17280 PR middle-end/81462
17281 * predict.c (set_even_probabilities): Cleanup; do not affect
17282 probabilities that are already known.
17283 (combine_predictions_for_bb): Call even when count is set.
17285 2017-07-18 Nathan Sidwell <nathan@acm.org>
17287 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17290 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17293 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17294 optimization for loop niter analysis.
17296 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17299 * config/avr/avr.c (avr_optimize_casesi): Don't use
17300 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17302 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17304 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17305 body_cost_vec from _vect_peel_extended_info.
17306 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17307 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17310 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17312 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17314 2017-07-18 Richard Biener <rguenther@suse.de>
17316 PR tree-optimization/80620
17317 PR tree-optimization/81403
17318 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17319 info when re-using a VN table entry.
17321 2017-07-18 Richard Biener <rguenther@suse.de>
17323 PR tree-optimization/81418
17324 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17325 vectype_in. Verify that with lane-reducing reduction operations
17326 we have a single def-use cycle.
17328 2017-07-17 Carl Love <cel@us.ibm.com>
17330 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17332 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17333 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17334 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17335 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17336 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17337 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17338 VMULOSW): New enum "unspec" values.
17339 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17340 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17341 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17342 altivec_vmulosw): New patterns.
17343 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17344 VMULOSW): Add definitions.
17345 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17347 * config/alpha/alpha.c: Include predict.h.
17349 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17351 * tree-vrp.c (compare_assert_loc): Fix comparison function
17352 to return predictable results.
17354 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17356 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17358 (subdi3): Likewise.
17359 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17360 * doc/invoke.texi (mexpand-adddi): Update text.
17362 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17364 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17365 that also clobbers the CC register. The old expand code is moved
17367 (*arc_clzsi2): ... here.
17368 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17369 the CC register. The old expand code is moved to ...
17370 (arc_ctzsi2): ... here.
17372 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17374 * config/arc/arc.opt (mindexed-loads): Use initial value
17375 TARGET_INDEXED_LOADS_DEFAULT.
17376 (mauto-modify-reg): Use initial value
17377 TARGET_AUTO_MODIFY_REG_DEFAULT.
17378 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17379 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17380 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17381 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17383 2017-07-17 Martin Liska <mliska@suse.cz>
17386 * opts.c (finish_options): Do not allow -fgnu-tm
17387 w/ -fsanitize={kernel-,}address. Say sorry.
17389 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17392 * tree-loop-distribution.c (classify_partition): Only assert on
17393 numer of iterations.
17394 (merge_dep_scc_partitions): Delete prameter. Update function call.
17395 (distribute_loop): Remove code handling loop with unknown niters.
17396 (pass_loop_distribution::execute): Skip loop with unknown niters.
17398 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17401 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17402 function sort_partitions_by_post_order.
17404 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17406 PR tree-optimization/81374
17407 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17408 the max index of basic blocks, rather than number of basic blocks.
17410 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17412 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17414 (arc_legitimate_pic_operand_p): Likewise.
17415 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17417 (arc_needs_pcl_p): Likewise.
17418 (arc_legitimate_pc_offset_p): Likewise.
17419 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17420 function is also used in constrains.md.
17421 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17422 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17423 PLUS. Only return true/false in known cases, otherwise assert.
17424 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17425 is already called in arc_legitimate_constant_p.
17426 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17428 (LEGITIMATE_PIC_OPERAND_P): Use
17429 arc_raw_symbolic_reference_mentioned_p function.
17430 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17435 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17436 Andrew Burgess <andrew.burgess@embecosm.com>
17438 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17439 (arc_return_address_register): New function.
17440 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17441 (arc_handle_fndecl_attribute): Add naked attribute.
17442 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17443 (TARGET_WARN_FUNC_RETURN): Likewise.
17444 (arc_allocate_stack_slots_for_args): New function.
17445 (arc_warn_func_return): Likewise.
17446 (machine_function): Change type fn_type.
17447 (arc_compute_function_type): Consider new naked function type,
17448 change function return type.
17449 (arc_must_save_register): Adapt to handle new
17450 arc_compute_function_type's return type.
17451 (arc_expand_prologue): Likewise.
17452 (arc_expand_epilogue): Likewise.
17453 (arc_return_address_regs): Delete.
17454 (arc_return_address_register): New function.
17455 (arc_epilogue_uses): Use above function.
17456 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17457 (arc_function_type): Change encoding, add naked type.
17458 (ARC_INTERRUPT_P): Change to handle the new encoding.
17459 (ARC_FAST_INTERRUPT_P): Likewise.
17460 (ARC_NORMAL_P): Define.
17461 (ARC_NAKED_P): Likewise.
17462 (arc_compute_function_type): Delete prototype.
17463 * config/arc/arc.md (in_ret_delay_slot): Use
17464 arc_return_address_register function.
17465 (simple_return): Likewise.
17466 (p_return_i): Likewise.
17468 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17470 PR tree-optimization/81428
17471 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17472 can't be built for those types.
17474 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17476 Remove stuff dead since r239246.
17478 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17479 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17480 (avr_inform_devices): Remove dead stuff.
17482 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17484 * config/arm/arm_neon.h: Fix softp typo.
17486 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17488 PR tree-optimization/81365
17489 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17490 aggregate moves onto bb predecessor edges, make sure there are no
17491 loads that could alias the lhs in between the start of bb and the
17494 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17497 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17498 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17499 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17501 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17503 PR tree-optimization/81396
17504 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17505 (init_symbolic_number): Initialize it to 1.
17506 (perform_symbolic_merge): Add n_ops from both operands into the new
17508 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17509 without base_addr as useless if they need more than one operation.
17510 (bswap_replace): Handle !bswap case for NULL base_addr.
17512 2017-07-17 Tom de Vries <tom@codesourcery.com>
17515 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17518 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17520 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17521 conditional builtin define __FIX_LEON3FT_B2BST.
17523 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17525 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17526 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17529 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17531 PR rtl-optimization/81424
17532 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17533 to remove potential trapping from operands if -fnon-call-exceptions.
17535 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17537 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17538 profile_proability for scalling.
17539 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17541 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17543 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17545 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17547 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17548 fixpoint arithmetics.
17550 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17552 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17553 fixpoint arithmetics.
17555 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17557 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17558 fixpoint arithmetics.
17560 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17562 * profile-count.h (profile_probability::from_reg_br_prob_note,
17563 profile_probability::to_reg_br_prob_note): New functions.
17564 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17565 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17566 * predict.c (probability_reliable_p): Update.
17567 (edge_probability_reliable_p): Update.
17568 (br_prob_note_reliable_p): Update.
17569 (invert_br_probabilities): Update.
17570 (add_reg_br_prob_note): New function.
17571 (combine_predictions_for_insn): Update.
17572 * asan.c (asan_clear_shadow): Update.
17573 * cfgbuild.c (compute_outgoing_frequencies): Update.
17574 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17575 (update_br_prob_note): Update.
17576 (rtl_verify_edges): Update.
17577 (purge_dead_edges): Update.
17578 (fixup_reorder_chain): Update.
17579 * emit-rtl.c (try_split): Update.
17580 * ifcvt.c (cond_exec_process_insns): Update.
17581 (cond_exec_process_if_block): Update.
17582 (dead_or_predicable): Update.
17583 * internal-fn.c (expand_addsub_overflow): Update.
17584 (expand_neg_overflow): Update.
17585 (expand_mul_overflow): Update.
17586 * loop-doloop.c (doloop_modify): Update.
17587 * loop-unroll.c (compare_and_jump_seq): Update.
17588 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17589 * predict.h: Update.
17590 * reorg.c (mostly_true_jump): Update.
17592 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17593 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17594 * config/arc/arc.c: (emit_unlikely_jump): Update.
17595 * config/arm/arm.c: (emit_unlikely_jump): Update.
17596 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17597 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17598 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17599 (ix86_print_operand): Update.
17600 (ix86_split_fp_branch): Update.
17601 (predict_jump): Update.
17602 * config/ia64/ia64.c (ia64_print_operand): Update.
17603 * config/mmix/mmix.c (mmix_print_operand): Update.
17604 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17605 (rs6000_expand_split_stack_prologue): Update.
17606 * config/rs6000/rs6000.c: Update.
17607 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17608 (s390_expand_vec_movstr): Update.
17609 (s390_expand_cs_tdsi): Update.
17610 (s390_expand_split_stack_prologue): Update.
17611 * config/sh/sh.c (sh_print_operand): Update.
17612 (expand_cbranchsi4): Update.
17613 (expand_cbranchdi4): Update.
17614 * config/sparc/sparc.c (output_v9branch): Update.
17615 * config/spu/spu.c (get_branch_target): Update.
17616 (ea_load_store_inline): Update.
17617 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17618 * config/tilepro/tilepro.c: Update.
17620 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17622 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17623 (gimplify_save_expr): Likewise.
17625 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17627 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17629 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17631 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17633 * timevar.def (TV_IPA_FNSUMMARY): Define.
17635 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17637 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17638 to back store errata sensitive sequence from being generated.
17639 (sqrtdf2_fix): Likewise.
17641 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17643 * tree-ssa-threadupdate.c (compute_path_counts,
17644 update_joiner_offpath_counts): Use profile_probability.
17646 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17649 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17651 * config/arm/arm-c.c (arm_cpu_builtins): Define
17652 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17654 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17656 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17657 array entries to represent __ieee128 versions of the
17658 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17659 scalar_extract_sig, and scalar_insert_exp built-in functions.
17660 (altivec_resolve_overloaded_builtin): Add special case handling
17661 for the __builtin_scalar_insert_exp function, as represented by
17662 the P9V_BUILTIN_VEC_VSIEDP constant.
17663 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17664 exponent support for __ieee128 argument.
17665 (VSESQP): Add scalar extract signature support for __ieee128
17667 (VSTDCNQP): Add scalar test negative support for __ieee128
17669 (VSIEQP): Add scalar insert exponent support for __int128 argument
17670 with __ieee128 result.
17671 (VSIEQPF): Add scalar insert exponent support for __ieee128
17672 argument with __ieee128 result.
17673 (VSTDCQP): Add scalar test data class support for __ieee128
17675 (VSTDCNQP): Add overload support for scalar test negative with
17676 __ieee128 argument.
17677 (VSTDCQP): Add overload support for scalar test data class
17678 __ieee128 argument.
17679 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17680 UNSPEC_VSX_SXSIGDP.
17681 (UNSPEC_VSX_SIEXPQP): New constant.
17682 (xsxexpqp): New insn for VSX scalar extract exponent quad
17684 (xsxsigqp): New insn for VSX scalar extract significand quad
17686 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17687 precision with floating point argument.
17688 (xststdcqp): New expand for VSX scalar test data class quad
17690 (xststdcnegqp): New expand for VSX scalar test negative quad
17692 (xststdcqp): New insn to match expansions for VSX scalar test data
17693 class quad precision and VSX scalar test negative quad precision.
17694 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17695 special case operand checking to enforce that second operand of
17696 VSX scalar test data class with quad precision argument is a 7-bit
17698 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17699 prototypes and descriptions of __ieee128 versions of
17700 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17701 scalar_test_data_class, and scalar_test_neg built-in functions.
17703 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17705 PR tree-optimization/81162
17706 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17707 replace a negate with an add.
17709 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17711 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17713 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17715 * config/arm/arm-c.c (arm_cpu_builtins): Define
17716 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17718 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17720 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17721 (armv8-r): Set ARM Cortex-R52 as default CPU.
17722 * config/arm/arm-tables.opt: Regenerate.
17723 * config/arm/arm-tune.md: Regenerate.
17724 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17726 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17727 extension for -mcpu=cortex-r52.
17729 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17731 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17732 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17733 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17734 (fp-armv8): Define it as FP_ARMv8 only.
17735 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17736 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17738 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17739 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17740 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17741 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17742 than TARGET_FPU_ARMV8.
17743 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17744 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17745 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17747 * config/arm/neon.md (neon_vrint): Likewise.
17748 (neon_vcvt): Likewise.
17749 (neon_<fmaxmin_op><mode>): Likewise.
17750 (<fmaxmin><mode>3): Likewise.
17751 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17752 * config/arm/predicates.md (arm_cond_move_operator): Check against
17753 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17755 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17757 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17758 to top of function.
17760 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17762 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17763 loop in comment with memset.
17765 2017-07-14 Martin Liska <mliska@suse.cz>
17767 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17768 * dwarf2out.c (is_java): Remove the function.
17769 (output_pubname): Remove usage of the function.
17770 (lower_bound_default): Remove usage of DW_LANG_Java.
17771 (gen_compile_unit_die): Likewise.
17772 * gcc.c: Remove compiler defaults for .java and .zip files.
17773 * gimple-expr.c (remove_suffix): Change as there's no longer
17774 extension than 4-letter one.
17775 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17776 (gimplify_save_expr): Likewise.
17777 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17778 as it's possible even for other languages than Java.
17779 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17780 * lto-opts.c (lto_write_options): Remove reference to Java.
17781 * opts.c (strip_off_ending): Update file extension handling.
17782 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17783 * tree-eh.c (lower_resx): Likewise.
17784 * tree.c (free_lang_data_in_type): Remove dead code.
17785 (find_decls_types_r): Likewise.
17786 (build_common_builtin_nodes): Remove Java from a comment.
17787 (verify_type): Remove dead code.
17788 * varasm.c (assemble_external): Remove Java from a comment.
17790 2017-07-14 Martin Liska <mliska@suse.cz>
17792 * opts.c (finish_options): Add quotes.
17793 (common_handle_option): Likewise.
17795 2017-07-14 Martin Liska <mliska@suse.cz>
17797 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17798 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17799 Remove N_SO_PASCAL.
17800 * dwarf2out.c (lower_bound_default): Do not handle
17802 (gen_compile_unit_die): Likewise.
17803 * gcc.c: Remove default extension binding for GNU Pascal.
17804 * stmt.c: Remove Pascal language from a comment.
17805 * xcoffout.c: Likewise.
17807 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17810 * diagnostic-show-locus.c (fixit_cmp): New function.
17811 (layout::layout): Sort m_fixit_hints.
17812 (column_range::column_range): Assert that the values are valid.
17813 (struct char_span): New struct.
17814 (correction::overwrite): New method.
17815 (struct source_line): New struct.
17816 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17817 calls in terms of classes source_line and char_span, and
17818 correction::overwrite.
17819 (selftest::test_overlapped_fixit_printing_2): New function.
17820 (selftest::diagnostic_show_locus_c_tests): Call it.
17822 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17824 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17825 early if there is no lhs.
17827 2017-07-13 Martin Liska <mliska@suse.cz>
17829 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17830 (gen_reference_type_die): Likewise.
17831 * stor-layout.c: Remove Pascal-related comment.
17833 2017-07-13 Martin Liska <mliska@suse.cz>
17835 * opts.c (finish_options): Add quotes to error messages.
17836 (parse_sanitizer_options): Likewise.
17838 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17840 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17842 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17844 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17846 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17848 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17850 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17852 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17855 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17856 provides the hardware capability bits, define the macro
17857 __BUILTIN_CPU_SUPPORTS__.
17858 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17859 if GLIBC does not provide the hardware capability bits. Add a
17860 gcc_unreachable call if the built-in cpu function is neither
17861 __builtin_cpu_is nor __builtin_cpu_supports.
17862 (rs6000_get_function_versions_dispatcher): Change the warning
17863 that an old GLIBC is used which does not export the capability
17864 bits to be an error.
17865 * doc/extend.texi (target_clones attribute): Document the
17866 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17867 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17868 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17869 the macros defined by GCC if the newer GLIBC is available.
17871 2017-07-12 Jeff Law <law@redhat.com>
17873 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17874 remaining includes slightly.
17875 * config/riscv/riscv-builtins.c: Include profile-count.h.
17877 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17880 * config/avr/avr.c (avr_set_current_function): In diagnostic
17881 messages: Quote keywords and (parts of) identifiers.
17882 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17885 2017-07-12 Carl Love <cel@us.ibm.com>
17887 * config/rs6000/rs6000-c.c: Add support for built-in functions
17888 vector bool char vec_revb (vector bool char);
17889 vector bool short vec_revb (vector short char);
17890 vector bool int vec_revb (vector bool int);
17891 vector bool long long vec_revb (vector bool long long);
17892 * doc/extend.texi: Update the built-in documentation file for the
17893 new built-in functions.
17895 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17897 * config/s390/s390.md: Remove movcc splitter.
17899 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17901 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17902 load/store on condition.
17904 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17907 * config/avr/avr.c (avr_encode_section_info)
17908 [progmem && !TREE_READONLY]: Error if progmem object needs
17911 2017-07-11 Michael Collison <michael.collison@arm.com>
17913 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17916 2017-07-11 Carl Love <cel@us.ibm.com>
17918 * config/rs6000/rs6000-c.c: Add support for builtins
17919 vector unsigned int vec_parity_lsbb (vector signed int);
17920 vector unsigned int vec_parity_lsbb (vector unsigned int);
17921 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17922 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17923 vector unsigned long long vec_parity_lsbb (vector signed long long);
17924 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17925 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17926 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17927 * doc/extend.texi: Update the built-in documentation file for the
17928 new built-in functions.
17930 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17932 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17933 (layout::m_primary_loc): New field.
17934 (layout::layout): Initialize new field. Move location filtering
17935 logic from here to...
17936 (layout::maybe_add_location_range): ...this new method. Add
17937 support for filtering to just the lines already specified by other
17939 (layout::will_show_line_p): New method.
17940 (gcc_rich_location::add_location_if_nearby): New method.
17941 (selftest::test_add_location_if_nearby): New test function.
17942 (selftest::diagnostic_show_locus_c_tests): Call it.
17943 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17946 2017-07-11 Tom de Vries <tom@codesourcery.com>
17948 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17949 (bb_first_real_insn): New function.
17950 (nvptx_single): Add extra initialization of broadcasted condition
17953 2017-07-11 Nathan Sidwell <nathan@acm.org>
17955 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17957 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17959 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17960 Binutils doc as TEXI will mess them up.
17961 * doc/invoke.texi (AVR Options): Same here.
17963 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17965 * config/sparc/sparc.opt (mfix-ut700): New option.
17966 (mfix-gr712rc): Likewise.
17967 (sparc_fix_b2bst): New variable.
17968 * doc/invoke.texi (SPARC options): Document them.
17969 (ARM options): Fix warnings.
17970 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17971 instructions to prevent sequences that can trigger the store-store
17972 errata for certain LEON3FT processors.
17973 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17974 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17975 * config/sparc/sparc.md (fix_b2bst): New attribute.
17976 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17978 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17981 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17983 (*rsqrtsf2_sse): Ditto.
17985 (div<mode>3): Macroize insn from divdf3 and divsf3
17986 using MODEF mode iterator.
17988 2017-07-10 Martin Sebor <msebor@redhat.com>
17990 PR tree-optimization/80397
17991 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17992 instead of testing for equality to INTEGER_TYPE.
17994 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17996 * config.gcc: Remove uclibc from arc target spec.
17998 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18000 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18002 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18005 * lto-wrapper.c (remove_option): New function.
18006 (merge_and_complain): Merge PIC/PIE options more realistically.
18008 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18010 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18014 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18015 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18016 * config.in: Regenerate.
18017 * configure: Regenerate.
18018 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18019 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18020 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18021 (TARGET_GASISR_PROLOGUES): ...target mask.
18022 * common/config/avr/avr-common.c
18023 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18024 Set -mgas-isr-prologues.
18025 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18026 INSERT_PASS_BEFORE for it.
18027 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18028 * config/avr/avr.c (avr_option_override)
18029 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18030 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18031 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18032 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18033 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18035 (make_avr_pass_pre_proep): New function.
18036 (emit_push_sfr) <treg>: Add argument to function and use it
18037 instead of TMP_REG.
18038 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18039 and set machine->gasisr.yes.
18040 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18041 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18042 __gcc_isr.n_pushed to .L__stack_usage.
18043 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18044 (avr_asm_final_postscan_insn): ...this new static function.
18045 * config/avr/avr.h (machine_function)
18046 <is_no_gccisr, use_L__stack_usage>: New fields.
18047 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18048 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18049 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18050 (gasisr, *gasisr): New expander and insn.
18051 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18052 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18053 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18055 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18057 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18060 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18062 Move jump-tables out of .text again.
18065 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18066 (ASM_OUTPUT_ADDR_VEC): New function.
18067 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18068 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18069 INSN_ADDRESSes as asm comment.
18070 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18071 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18072 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18073 * config/avr/avr.md (*tablejump): Adjust comment.
18074 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18075 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18077 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18078 (avr_output_addr_vec): New proto.
18079 (avr_log_t) <insn_addresses>: New field.
18081 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18084 * config/i386/i386.c (ix86_function_arg_advance): Set
18085 outgoing_args_on_stack to true if there are outgoing arguments
18087 (ix86_function_arg): Likewise.
18088 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18089 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18090 * config/i386/i386.h (machine_function): Add
18091 outgoing_args_on_stack.
18093 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18095 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18096 supporting pthreds.
18097 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18099 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18101 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18102 (REAL_H): Remove $(MACHMODE_H).
18103 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18105 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18106 $(MACHMODE_H) and double-int.h.
18107 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18109 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18112 2017-07-07 Andrew Pinski <apinski@cavium.com>
18114 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18115 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18117 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18119 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18120 Add warning if GCC was not configured to link against a GLIBC that
18121 exports the hardware capability bits.
18122 (make_resolver_func): Make resolver function private and not a
18123 COMDAT function. Create the name with clone_function_name instead
18124 of make_unique_name.
18127 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18128 correct operand in doing the split.
18130 2017-07-07 Carl Love <cel@us.ibm.com>
18132 * config/rs6000/rs6000-c: Add support for built-in function
18133 vector unsigned short vec_pack_to_short_fp32 (vector float,
18135 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18136 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18137 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18138 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18139 (convert_4f32_8i16): Add define_expand.
18140 * doc/extend.texi: Update the built-in documentation file for the
18141 new built-in function.
18143 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18145 * config/sparc/m8.md: New file.
18146 * config/sparc/sparc.md: Include m8.md.
18148 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18150 * config/sparc/sparc.opt: New option -mvis4b.
18151 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18152 (sparc_option_override): Handle VIS4B.
18153 (enum sparc_builtins): Define
18154 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18155 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18156 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18157 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18158 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18159 (check_constant_argument): New function.
18160 (sparc_vis_init_builtins): Define builtins
18161 __builtin_vis_dictunpack{8,16,32},
18162 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18163 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18164 __builtin_vis_fpcmpde{8,16,32}shl and
18165 __builtin_vis_fpcmpur{8,16,32}shl.
18166 (sparc_expand_builtin): Check that the constant operands to
18167 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18168 constant and in range.
18169 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18171 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18172 (SPARC_IMM5_P): Likewise.
18173 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18174 (enabled): Handle vis4b.
18175 (UNSPEC_DICTUNPACK): New unspec.
18176 (UNSPEC_FPCMPSHL): Likewise.
18177 (UNSPEC_FPUCMPSHL): Likewise.
18178 (UNSPEC_FPCMPDESHL): Likewise.
18179 (UNSPEC_FPCMPURSHL): Likewise.
18180 (cpu_feature): New CPU feature `vis4b'.
18181 (dictunpack{8,16,32}): New insns.
18182 (FPCSMODE): New mode iterator.
18183 (fpcscond): New code iterator.
18184 (fpcsucond): Likewise.
18185 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18186 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18187 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18188 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18189 * config/sparc/constraints.md: Define constraints `q' for unsigned
18190 2-bit integer constants and `t' for unsigned 5-bit integer
18192 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18194 (imm5_operand_dictunpack16): Likewise.
18195 (imm5_operand_dictunpack32): Likewise.
18196 (imm2_operand): Likewise.
18197 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18198 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18199 ditunpack* and fpcmp*shl builtins.
18201 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18203 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18204 * config.in: Add HAVE_AS_SPARC6 define.
18205 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18207 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18209 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18210 (CPP_CPU_SPEC): Handle m8.
18211 (ASM_CPU_SPEC): Likewise.
18212 * config/sparc/sparc-opts.h (enum processor_type): Add
18214 * config/sparc/sparc.c (m8_costs): New struct.
18215 (sparc_option_override): Handle TARGET_CPU_m8.
18216 (sparc32_initialize_trampoline): Likewise.
18217 (sparc64_initialize_trampoline): Likewise.
18218 (sparc_issue_rate): Likewise.
18219 (sparc_register_move_cost): Likewise.
18220 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18221 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18222 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18223 (CPP_CPU_SPEC): Handle M8.
18224 (ASM_CPU_SPEC): Likewise.
18225 (AS_M8_FLAG): Define.
18226 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18227 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18228 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18230 * configure: Regenerate.
18231 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18234 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18236 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18238 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18239 ("*movdi_insn_sp32"): Do not set v3pipe.
18240 ("*movsi_insn"): Likewise.
18241 ("*movdi_insn_sp64"): Likewise.
18242 ("*movsf_insn"): Likewise.
18243 ("*movdf_insn_sp32"): Likewise.
18244 ("*movdf_insn_sp64"): Likewise.
18245 ("*zero_extendsidi2_insn_sp64"): Likewise.
18246 ("*sign_extendsidi2_insn"): Likewise.
18247 ("*mov<VM32:mode>_insn"): Likewise.
18248 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18249 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18250 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18251 ("<vlop:code><VL:mode>3"): Likewise.
18252 ("*not_<vlop:code><VL:mode>3"): Likewise.
18253 ("*nand<VL:mode>_vis"): Likewise.
18254 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18255 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18256 ("one_cmpl<VL:mode>2"): Likewise.
18257 ("faligndata<VM64:mode>_vis"): Likewise.
18258 ("alignaddrsi_vis"): Likewise.
18259 ("alignaddrdi_vis"): Likweise.
18260 ("alignaddrlsi_vis"): Likewise.
18261 ("alignaddrldi_vis"): Likewise.
18262 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18263 ("bmaskdi_vis"): Likewise.
18264 ("bmasksi_vis"): Likewise.
18265 ("bshuffle<VM64:mode>_vis"): Likewise.
18266 ("cmask8<P:mode>_vis"): Likewise.
18267 ("cmask16<P:mode>_vis"): Likewise.
18268 ("cmask32<P:mode>_vis"): Likewise.
18269 ("pdistn<P:mode>_vis"): Likewise.
18270 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18272 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18274 * config/sparc/sparc.md ("subtype"): New insn attribute.
18275 ("*wrgsr_sp64"): Set insn subtype.
18276 ("*rdgsr_sp64"): Likewise.
18277 ("alignaddrsi_vis"): Likewise.
18278 ("alignaddrdi_vis"): Likewise.
18279 ("alignaddrlsi_vis"): Likewise.
18280 ("alignaddrldi_vis"): Likewise.
18281 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18282 ("fexpand_vis"): Likewise.
18283 ("fpmerge_vis"): Likewise.
18284 ("faligndata<VM64:mode>_vis"): Likewise.
18285 ("bshuffle<VM64:mode>_vis"): Likewise.
18286 ("cmask8<P:mode>_vis"): Likewise.
18287 ("cmask16<P:mode>_vis"): Likewise.
18288 ("cmask32<P:mode>_vis"): Likewise.
18289 ("fchksm16_vis"): Likewise.
18290 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18291 ("fmean16_vis"): Likewise.
18292 ("fp<plusminus_insn>64_vis"): Likewise.
18293 ("<plusminus_insn>v8qi3"): Likewise.
18294 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18295 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18296 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18297 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18298 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18299 ("*movqi_insn"): Likewise.
18300 ("*movhi_insn"): Likewise.
18301 ("*movsi_insn"): Likewise.
18302 ("movsi_pic_gotdata_op"): Likewise.
18303 ("*movdi_insn_sp32"): Likewise.
18304 ("*movdi_insn_sp64"): Likewise.
18305 ("movdi_pic_gotdata_op"): Likewise.
18306 ("*movsf_insn"): Likewise.
18307 ("*movdf_insn_sp32"): Likewise.
18308 ("*movdf_insn_sp64"): Likewise.
18309 ("*zero_extendhisi2_insn"): Likewise.
18310 ("*zero_extendqihi2_insn"): Likewise.
18311 ("*zero_extendqisi2_insn"): Likewise.
18312 ("*zero_extendqidi2_insn"): Likewise.
18313 ("*zero_extendhidi2_insn"): Likewise.
18314 ("*zero_extendsidi2_insn_sp64"): Likewise.
18315 ("ldfsr"): Likewise.
18316 ("prefetch_64"): Likewise.
18317 ("prefetch_32"): Likewise.
18318 ("tie_ld32"): Likewise.
18319 ("tie_ld64"): Likewise.
18320 ("*tldo_ldub_sp32"): Likewise.
18321 ("*tldo_ldub1_sp32"): Likewise.
18322 ("*tldo_ldub2_sp32"): Likewise.
18323 ("*tldo_ldub_sp64"): Likewise.
18324 ("*tldo_ldub1_sp64"): Likewise.
18325 ("*tldo_ldub2_sp64"): Likewise.
18326 ("*tldo_ldub3_sp64"): Likewise.
18327 ("*tldo_lduh_sp32"): Likewise.
18328 ("*tldo_lduh1_sp32"): Likewise.
18329 ("*tldo_lduh_sp64"): Likewise.
18330 ("*tldo_lduh1_sp64"): Likewise.
18331 ("*tldo_lduh2_sp64"): Likewise.
18332 ("*tldo_lduw_sp32"): Likewise.
18333 ("*tldo_lduw_sp64"): Likewise.
18334 ("*tldo_lduw1_sp64"): Likewise.
18335 ("*tldo_ldx_sp64"): Likewise.
18336 ("*mov<VM32:mode>_insn"): Likewise.
18337 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18338 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18340 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18342 * config/sparc/sparc.md ("type"): New insn type viscmp.
18343 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18345 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18346 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18347 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18348 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18350 ("n7_vis_logical_11cycle"): Likewise.
18351 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18352 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18353 * config/sparc/niagara.md ("niag_vis"): Likewise.
18354 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18355 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18357 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18359 * config/sparc/sparc.md: New instruction type `bmask'.
18360 (bmaskdi_vis): Use the `bmask' type.
18361 (bmasksi_vis): Likewise.
18362 * config/sparc/ultra3.md (us3_array): Likewise.
18363 * config/sparc/niagara7.md (n7_array): Likewise.
18364 * config/sparc/niagara4.md (n4_array): Likewise.
18365 * config/sparc/niagara2.md (niag2_vis): Likewise.
18366 (niag3_vis): Likewise.
18367 * config/sparc/niagara.md (niag_vis): Likewise.
18369 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18371 * ipa-comdats.c: Remove optimize check from gate.
18372 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18373 for functions not optimized.
18374 (ipa_fn_summary_read): Skip optimize check.
18375 (ipa_fn_summary_write): Likewise.
18376 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18378 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18380 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18381 (check_callers): Give up on caller which is not optimized.
18382 (inline_small_functions): Likewise.
18383 (ipa_inline): Do not give up when not optimizing.
18384 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18385 away unoptimizes cdtors.
18386 (whole_program_function_and_variable_visibility): Do
18387 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18388 * ipa.c (process_references): Do not check optimize.
18389 (symbol_table::remove_unreachable_nodes): Update optimize check.
18390 (set_writeonly_bit): Update optimize check.
18391 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18392 (pass_ipa_single_use::gate): Remove.
18394 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18396 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18397 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18398 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18399 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18400 permute_load, permute_store, adjust_extract, adjust_splat,
18401 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18402 replace_swap_with_copy, dump_swap_insn_table,
18403 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18404 recombine_lvx_pattern, recombine_stvx_pattern,
18405 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18406 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18407 to file rs6000-p8swap.c.
18408 * config/rs6000/rs6000-p8swap.c: New file.
18409 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18410 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18411 and rs6000*-*-* targets.
18413 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18415 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18417 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18419 * lto-wrapper.c (merge_and_complain): Do not merge
18420 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18421 fsigned_zeros, ftrapping_math, fwrapv.
18422 (append_compiler_options): Do not track these options.
18423 (append_linker_options): Likewie
18425 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18427 * cgraphunit.c (cgraph_node::finalize_function): When
18428 !flag_toplevel_reorde set no_reorder flag.
18429 (varpool_node::finalize_decl): Likewise.
18430 (symbol_table::compile): Drop no toplevel reorder path.
18432 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18434 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18435 edges; zero probability is not better than uninitialized.
18437 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18439 * asan.h (asan_sanitize_allocas_p): Declare.
18440 * asan.c (asan_sanitize_allocas_p): New function.
18441 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18442 (handle_builtin_alloca): Likewise.
18443 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18444 if !asan_sanitize_allocas_p.
18445 * params.def (asan-instrument-allocas): Add new option.
18446 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18447 * opts.c (common_handle_option): Disable allocas sanitization for
18450 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18452 * asan.c: Include gimple-fold.h.
18453 (get_last_alloca_addr): New function.
18454 (handle_builtin_stackrestore): Likewise.
18455 (handle_builtin_alloca): Likewise.
18456 (asan_emit_allocas_unpoison): Likewise.
18457 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18458 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18459 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18460 (instrument_builtin_call): Pass gimple iterator to
18461 get_mem_refs_of_builtin_call.
18462 (last_alloca_addr): New global.
18463 * asan.h (asan_emit_allocas_unpoison): Declare.
18464 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18465 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18466 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18467 if function calls alloca.
18468 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18469 * gimple-fold.h (replace_call_with_value): Declare.
18470 * internal-fn.c: Include asan.h.
18471 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18472 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18474 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18476 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18477 (C_SELFTEST_FLAGS): New.
18478 (CPP_SELFTEST_FLAGS): New.
18479 (SELFTEST_DEPS): New, from deps of s-selftest.
18480 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18481 (CPP_SELFTEST_DEPS): New.
18482 (selftest): Add dependency on s-selftest-c++.
18483 (s-selftest): Rename to...
18484 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18485 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18486 than SELFTEST_FLAGS.
18487 (selftest-gdb): Rename to...
18488 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18490 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18491 (selftest-valgrind): Rename to...
18492 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18494 (selftest-valgrind): Reintroduce as an alias for
18495 selftest-c-valgrind.
18496 (s-selftest-c++): New.
18497 (selftest-c++-gdb): New.
18498 (selftest-c++-valgrind): New.
18500 2017-07-06 Olivier Hainque <hainque@adacore.com>
18502 * gcc.c (process_command): When deciding if undefined variables
18503 should be ignored when processing specs, accept "gcc -v" as well.
18505 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18507 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18508 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18510 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18512 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18513 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18514 * config/arm/arm-tables.opt: Regenerate.
18515 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18517 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18519 2017-07-06 Carl Love <cel@us.ibm.com>
18521 * ChangeLog: Clean up from mid air collision
18523 2017-07-06 Carl Love <cel@us.ibm.com>
18525 * config/rs6000/rs6000-c.c: Add support for built-in functions
18526 vector signed int vec_subc (vector signed int, vector signed int);
18527 vector signed __int128 vec_subc (vector signed __int128,
18528 vector signed __int128);
18529 vector unsigned __int128 vec_subc (vector unsigned __int128,
18530 vector unsigned __int128);
18531 vector signed int vec_sube (vector signed int, vector signed int,
18532 vector signed int);
18533 vector unsigned int vec_sube (vector unsigned int,
18534 vector unsigned int,
18535 vector unsigned int);
18536 vector signed __int128 vec_sube (vector signed __int128,
18537 vector signed __int128,
18538 vector signed__int128);
18539 vector unsigned __int128 vec_sube (vector unsigned __int128,
18540 vector unsigned __int128,
18541 vector unsigned __int128);
18542 vector signed int vec_subec (vector signed int, vector signed int,
18543 vector signed int);
18544 vector unsigned int vec_subec (vector unsigned int,
18545 vector unsigned int,
18546 vector unsigned int);
18547 vector signed __int128 vec_subec (vector signed __int128,
18548 vector signed __int128,
18549 vector signed__int128);
18550 vector unsigned __int128 vec_subec (vector unsigned __int128,
18551 vector unsigned __int128,
18552 vector unsigned __int128);
18553 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18554 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18555 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18556 BU_ALTIVEC_OVERLOAD_X definitions.
18557 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18558 * doc/extend.texi: Update the built-in documentation file for the new
18559 built-in functions.
18561 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18564 * diagnostic-show-locus.c (layout::layout): Use start and finish
18565 spelling location for the start and finish of each range.
18566 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18567 Add unused aspect param.
18568 * input.c (expand_location_1): Add "aspect" param, and use it
18569 to access the correct part of the location.
18570 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18572 (expand_location_to_spelling_point): Likewise.
18573 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18574 param, and pass it to expand_location_1.
18576 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18578 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18579 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18580 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18581 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18582 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18583 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18584 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18585 _mm_maskz_getmant_ss): New intrinsics.
18586 (__builtin_ia32_getexpss128_mask): Changed to ...
18587 __builtin_ia32_getexpss128_round ... this.
18588 (__builtin_ia32_getexpsd128_mask): Changed to ...
18589 __builtin_ia32_getexpsd128_round ... this.
18590 * config/i386/i386-builtin-types.def
18591 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18592 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18593 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18594 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18595 __builtin_ia32_getmantss_mask_round): New builtins.
18596 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18597 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18598 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18599 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18600 * config/i386/sse.md
18601 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18602 avx512f_sgetexp<mode><mask_scalar_name>
18603 <round_saeonly_scalar_name> ... this.
18604 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18605 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18606 vgetexp<ssescalarmodesuffix>
18607 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18608 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18609 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18610 avx512f_vgetmant<mode><mask_scalar_name>
18611 <round_saeonly_scalar_name> ... this.
18612 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18613 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18614 vgetmant<ssescalarmodesuffix>
18615 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18616 %0<mask_scalar_operand4>, %1, %2
18617 <round_saeonly_scalar_mask_op4>, %3} ... this.
18618 * config/i386/subst.md (mask_scalar_operand4,
18619 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18620 round_saeonly_scalar_nimm_predicate): New subst attributes.
18622 2017-07-06 Julia Koval <julia.koval@intel.com>
18624 * config/i386/i386.c (ix86_erase_embedded_rounding):
18625 Remove code for old rounding pattern.
18627 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18629 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18631 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18633 * doc/sourcebuild.texi (Test Directives, Variants of
18634 dg-require-support): Add documentation for dg-require-stack-check.
18636 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18638 * config/i386/subst.md (mask_scalar, round_scalar,
18639 round_saeonly_scalar): New meta-templates.
18640 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18641 round_scalar_mask_operand3, round_scalar_mask_op3,
18642 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18643 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18644 round_saeonly_scalar_constraint,
18645 round_saeonly_scalar_prefix): New subst attribute.
18646 * config/i386/sse.md
18647 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18648 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18649 <round_scalar_name> ... this.
18650 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18651 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18652 <round_scalar_name> ... this.
18653 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18654 <sse>_vm<code><mode>3<mask_scalar_name>
18655 <round_saeonly_scalar_name> ... this.
18656 (v<plusminus_mnemonic><ssescalarmodesuffix>
18657 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18658 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18659 v<plusminus_mnemonic><ssescalarmodesuffix>
18660 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18661 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18662 (v<multdiv_mnemonic><ssescalarmodesuffix>
18663 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18664 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18665 v<multdiv_mnemonic><ssescalarmodesuffix>
18666 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18667 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18668 (v<maxmin_float><ssescalarmodesuffix>
18669 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18670 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18671 v<maxmin_float><ssescalarmodesuffix>
18672 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18673 %0<mask_scalar_operand3>, %1, %<iptr>2
18674 <round_saeonly_scalar_mask_op3>} ... this.
18676 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18678 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18679 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18681 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18682 Alan Hayward <alan.hayward@arm.com>
18683 David Sherwood <david.sherwood@arm.com>
18685 * combine.c (simplify_if_then_else): Remove "enum" before
18687 * compare-elim.c (can_eliminate_compare): Likewise.
18688 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18690 (aarch64_lookup_simd_builtin_type): Likewise.
18691 (aarch64_simd_builtin_type): Likewise.
18692 (aarch64_init_simd_builtin_types): Likewise.
18693 (aarch64_simd_expand_args): Likewise.
18694 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18696 (aarch64_reverse_mask): Likewise.
18697 (aarch64_simd_emit_reg_reg_move): Likewise.
18698 (aarch64_gen_adjusted_ldpstp): Likewise.
18699 (aarch64_ccmp_mode_to_code): Likewise.
18700 (aarch64_operands_ok_for_ldpstp): Likewise.
18701 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18702 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18704 (aarch64_min_divisions_for_recip_mul): Likewise.
18705 (aarch64_reassociation_width): Likewise.
18706 (aarch64_get_condition_code_1): Likewise.
18707 (aarch64_simd_emit_reg_reg_move): Likewise.
18708 (aarch64_simd_attr_length_rglist): Likewise.
18709 (aarch64_reverse_mask): Likewise.
18710 (aarch64_operands_ok_for_ldpstp): Likewise.
18711 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18712 (aarch64_gen_adjusted_ldpstp): Likewise.
18713 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18715 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18716 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18717 (arm_lookup_simd_builtin_type): Likewise.
18718 (arm_simd_builtin_type): Likewise.
18719 (arm_init_simd_builtin_types): Likewise.
18720 (arm_expand_builtin_args): Likewise.
18721 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18722 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18723 (ft32_setup_incoming_varargs): Likewise.
18724 (ft32_function_arg): Likewise.
18725 (ft32_function_arg_advance): Likewise.
18726 (ft32_pass_by_reference): Likewise.
18727 (ft32_arg_partial_bytes): Likewise.
18728 (ft32_valid_pointer_mode): Likewise.
18729 (ft32_addr_space_pointer_mode): Likewise.
18730 (ft32_addr_space_legitimate_address_p): Likewise.
18731 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18733 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18734 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18735 (iamcu_alignment): Likewise.
18736 (canonicalize_vector_int_perm): Likewise.
18737 (ix86_noce_conversion_profitable_p): Likewise.
18738 (ix86_mpx_bound_mode): Likewise.
18739 (ix86_operands_ok_for_move_multiple): Likewise.
18740 * config/microblaze/microblaze-protos.h
18741 (microblaze_expand_conditional_branch_reg): Likewise.
18742 * config/microblaze/microblaze.c
18743 (microblaze_expand_conditional_branch_reg): Likewise.
18744 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18746 (rs6000_reassociation_width): Likewise.
18747 (rs6000_invalid_binary_op): Likewise.
18748 (fusion_p9_p): Likewise.
18749 (emit_fusion_p9_load): Likewise.
18750 (emit_fusion_p9_store): Likewise.
18751 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18753 (riscv_hard_regno_mode_ok_p): Likewise.
18754 (riscv_address_insns): Likewise.
18755 (riscv_split_symbol): Likewise.
18756 (riscv_legitimize_move): Likewise.
18757 (riscv_function_value): Likewise.
18758 (riscv_hard_regno_nregs): Likewise.
18759 (riscv_expand_builtin): Likewise.
18760 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18761 (riscv_build_integer): Likewise.
18762 (riscv_split_integer): Likewise.
18763 (riscv_legitimate_constant_p): Likewise.
18764 (riscv_cannot_force_const_mem): Likewise.
18765 (riscv_regno_mode_ok_for_base_p): Likewise.
18766 (riscv_valid_base_register_p): Likewise.
18767 (riscv_valid_offset_p): Likewise.
18768 (riscv_valid_lo_sum_p): Likewise.
18769 (riscv_classify_address): Likewise.
18770 (riscv_legitimate_address_p): Likewise.
18771 (riscv_address_insns): Likewise.
18772 (riscv_load_store_insns): Likewise.
18773 (riscv_force_binary): Likewise.
18774 (riscv_split_symbol): Likewise.
18775 (riscv_force_address): Likewise.
18776 (riscv_legitimize_address): Likewise.
18777 (riscv_move_integer): Likewise.
18778 (riscv_legitimize_const_move): Likewise.
18779 (riscv_legitimize_move): Likewise.
18780 (riscv_address_cost): Likewise.
18781 (riscv_subword): Likewise.
18782 (riscv_output_move): Likewise.
18783 (riscv_canonicalize_int_order_test): Likewise.
18784 (riscv_emit_int_order_test): Likewise.
18785 (riscv_function_arg_boundary): Likewise.
18786 (riscv_pass_mode_in_fpr_p): Likewise.
18787 (riscv_pass_fpr_single): Likewise.
18788 (riscv_pass_fpr_pair): Likewise.
18789 (riscv_get_arg_info): Likewise.
18790 (riscv_function_arg): Likewise.
18791 (riscv_function_arg_advance): Likewise.
18792 (riscv_arg_partial_bytes): Likewise.
18793 (riscv_function_value): Likewise.
18794 (riscv_pass_by_reference): Likewise.
18795 (riscv_setup_incoming_varargs): Likewise.
18796 (riscv_print_operand): Likewise.
18797 (riscv_elf_select_rtx_section): Likewise.
18798 (riscv_save_restore_reg): Likewise.
18799 (riscv_for_each_saved_reg): Likewise.
18800 (riscv_register_move_cost): Likewise.
18801 (riscv_hard_regno_mode_ok_p): Likewise.
18802 (riscv_hard_regno_nregs): Likewise.
18803 (riscv_class_max_nregs): Likewise.
18804 (riscv_memory_move_cost): Likewise.
18805 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18806 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18807 (rl78_addr_space_address_mode): Likewise.
18808 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18810 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18811 (rs6000_reassociation_width): Likewise.
18812 (rs6000_invalid_binary_op): Likewise.
18813 (fusion_p9_p): Likewise.
18814 (emit_fusion_p9_load): Likewise.
18815 (emit_fusion_p9_store): Likewise.
18816 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18817 (ok_for_simple_move_operands): Likewise.
18818 (ok_for_simple_move_strict_operands): Likewise.
18819 (ok_for_simple_arith_logic_operands): Likewise.
18820 (visium_legitimize_reload_address): Likewise.
18821 (visium_select_cc_mode): Likewise.
18822 (output_cbranch): Likewise.
18823 (visium_split_double_move): Likewise.
18824 (visium_expand_copysign): Likewise.
18825 (visium_expand_int_cstore): Likewise.
18826 (visium_expand_fp_cstore): Likewise.
18827 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18828 (visium_function_arg): Likewise.
18829 (visium_function_arg_advance): Likewise.
18830 (visium_libcall_value): Likewise.
18831 (visium_setup_incoming_varargs): Likewise.
18832 (visium_legitimate_constant_p): Likewise.
18833 (visium_legitimate_address_p): Likewise.
18834 (visium_legitimize_address): Likewise.
18835 (visium_secondary_reload): Likewise.
18836 (visium_register_move_cost): Likewise.
18837 (visium_memory_move_cost): Likewise.
18838 (prepare_move_operands): Likewise.
18839 (ok_for_simple_move_operands): Likewise.
18840 (ok_for_simple_move_strict_operands): Likewise.
18841 (ok_for_simple_arith_logic_operands): Likewise.
18842 (visium_function_value_1): Likewise.
18843 (rtx_ok_for_offset_p): Likewise.
18844 (visium_legitimize_reload_address): Likewise.
18845 (visium_split_double_move): Likewise.
18846 (visium_expand_copysign): Likewise.
18847 (visium_expand_int_cstore): Likewise.
18848 (visium_expand_fp_cstore): Likewise.
18849 (visium_split_cstore): Likewise.
18850 (visium_select_cc_mode): Likewise.
18851 (visium_split_cbranch): Likewise.
18852 (output_cbranch): Likewise.
18853 (visium_print_operand_address): Likewise.
18854 * expmed.c (flip_storage_order): Likewise.
18855 * expmed.h (emit_cstore): Likewise.
18856 (flip_storage_order): Likewise.
18857 * genrecog.c (validate_pattern): Likewise.
18858 * hsa-gen.c (gen_hsa_addr): Likewise.
18859 * internal-fn.c (expand_arith_overflow): Likewise.
18860 * ira-color.c (allocno_copy_cost_saving): Likewise.
18861 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18862 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18863 (process_invariant_for_inheritance): Likewise.
18864 * lra-eliminations.c (move_plus_up): Likewise.
18865 * omp-low.c (lower_oacc_reductions): Likewise.
18866 * simplify-rtx.c (simplify_subreg): Likewise.
18867 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18868 (TARGET_CHKP_BOUND_MODE): Likewise..
18869 * targhooks.c (default_chkp_bound_mode): Likewise.
18870 (default_setup_incoming_vararg_bounds): Likewise.
18871 * targhooks.h (default_chkp_bound_mode): Likewise.
18872 (default_setup_incoming_vararg_bounds): Likewise.
18873 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18874 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18875 (have_whole_vector_shift): Likewise.
18876 * tree-vect-stmts.c (vectorizable_load): Likewise.
18877 * doc/tm.texi: Regenerate.
18879 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18881 Graceful degrade if Binutils PR21472 is not available.
18884 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18885 .rodata in flash test fails.
18886 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18887 * confgure: Regenerate.
18888 * config.in: Regenerate.
18889 * config/avr/avr.c (avr_asm_named_section)
18890 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18891 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18892 (avr_asm_init_sections): Same.
18894 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18896 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18897 (fma<VH:mode>4_intrinsic): Likewise.
18898 (*fmsub<VCVTF:mode>4): Likewise.
18899 (*fmsub<VH:mode>4_intrinsic): Likewise.
18901 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18904 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18905 Don't depend on "optimize > 0".
18906 (out_movhi_r_mr, out_movqi_mr_r): Same.
18907 (out_movhi_mr_r, out_movqi_r_mr): Same.
18908 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18909 io_address_operand on "optimize > 0".
18911 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18913 * tree-loop-distribution.c: Add general explanantion on the pass.
18914 (generate_loops_for_partition): Mark distributed loop.
18915 (pg_add_dependence_edges): New parameter. Handle alias data
18916 dependence specially and record it in the parameter if asked.
18917 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18918 (init_partition_graph_vertices, add_partition_graph_edge): New.
18919 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18920 (free_partition_graph_vdata, build_partition_graph): New.
18921 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18922 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18923 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18924 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18925 (version_for_distribution_p, finalize_partitions): New.
18926 (distribute_loop): Handle alias data dependence specially. Factor
18927 out loop fusion code as functions and call these functions.
18929 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18931 * tree-loop-distribution.c (classify_partition): New parameter and
18932 better handle reduction statement.
18933 (rdg_build_partitions): Revise comment.
18934 (distribute_loop): Compute statements in all partitions and pass it
18935 to classify_partition.
18937 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18939 * tree-loop-distribution.c (enum partition_type): New.
18940 (struct partition): New field type.
18941 (partition_merge_into): Add parameter. Update partition type.
18942 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18943 (build_rdg_partition_for_vertex): Compute partition type.
18944 (rdg_build_partitions): Dump partition type.
18945 (distribute_loop): Update calls to partition_merge_into.
18947 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18949 * tree-loop-distribution.c (struct ddr_hasher): New.
18950 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18952 (classify_partition): Call get_data_dependence.
18953 (pg_add_dependence_edges): Ditto.
18954 (distribute_loop): Release data dependence hash table.
18956 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18958 * tree-loop-distribution.c (ref_base_address): Delete.
18959 (similar_memory_accesses): Rename ...
18960 (share_memory_accesses): ... to this. Check if partitions access
18961 the same memory reference.
18962 (distribute_loop): Call share_memory_accesses.
18964 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18966 * tree-loop-distribution.c (struct partition): New field recording
18967 its data reference.
18968 (partition_alloc, partition_free): Init and release data refs.
18969 (partition_merge_into): Merge data refs.
18970 (build_rdg_partition_for_vertex): Collect data refs for partition.
18971 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18973 (distribute_loop): Remve data refs from vertice data of partition
18976 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18978 * tree-loop-distribution.c (params.h): Include header file.
18979 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18980 (datarefs_vec): New global var.
18981 (create_rdg_vertices): Use datarefs_vec directly.
18982 (free_rdg): Don't free data references.
18983 (build_rdg): Update use. Don't free data references.
18984 (distribute_loop): Compute global variable for data references.
18985 Bail out if there are too many data references.
18987 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18989 * tree-loop-distribution.c (loop_nest): New global var.
18990 (build_rdg): Use loop directly, rather than loop nest.
18991 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18993 (distribute_loop): Compute global variable loop nest. Update use.
18995 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18997 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18998 (partition_merge_into): New parameter. Dump reason for fusion.
18999 (distribute_loop): Update use of partition_merge_into.
19001 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19003 * tree-loop-distribution.c (bb_top_order_index): New.
19004 (bb_top_order_index_size, bb_top_order_cmp): New.
19005 (stmts_from_loop): Use topological order.
19006 (pass_loop_distribution::execute): Compute and release topological
19007 order for basic blocks.
19009 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19011 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19014 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19016 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19017 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19018 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19019 * internal-fn.def (LOOP_DIST_ALIAS): New.
19020 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19021 (fold_loop_internal_call): ... this.
19022 (vect_loop_dist_alias_call): New function.
19023 (set_uid_loop_bbs): Call fold_loop_internal_call.
19024 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19027 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19030 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19031 Require dead FLAGS_REG at the beginning of a peephole.
19033 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19036 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19037 arguments in the call to __builtin_ia32_sbb_u32.
19038 (_subborrow_u64): Swap _X and _Y arguments in the call to
19039 __builtin_ia32_sbb_u64.
19041 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19044 * tree-vrp.c (compare_assert_loc): Turn into a function template
19045 with stable template parameter. Only test if a->e is NULL,
19046 !a->e == !b->e has been verified already. Use e == NULL or
19047 e != NULL instead of e or ! e tests. If stable is true, don't use
19048 iterative_hash_expr, on the other side allow a or b or both NULL
19049 and sort the NULLs last.
19050 (process_assert_insertions): Sort using compare_assert_loc<false>
19051 instead of compare_assert_loc, later sort using
19052 compare_assert_loc<true> before calling process_assert_insertions_for
19053 in a loop. Use break instead of continue once seen NULL pointer.
19055 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19057 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19058 Cortex-R7 and Cortex-R8 processors.
19060 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19062 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19063 uninitialized while src is not.
19065 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19067 * common/config/arm/arm-common.c: Adjust include path for
19069 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19070 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19071 (arm-cpu-data.h): Likewise.
19072 (arm-cpu-cdata.h): Likewise.
19073 * config/arm/arm-cpu.h: Delete.
19074 * config/arm/arm-cpu-cdata.h: Delete.
19075 * config/arm/arm-cpu-data.h: Delete.
19077 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19079 * config/arm/arm-cpus.in (cortex-a55): New.
19080 (cortex-a75): Likewise.
19081 (cortex-a75.cortex-a55): Likewise.
19082 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19084 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19085 * config/arm/arm-cpu-cdata.h: Regenerate.
19086 * config/arm/arm-cpu-data.h: Regenerate.
19087 * config/arm/arm-cpu.h: Regenerate.
19088 * config/arm/arm-tables.opt: Regenerate.
19089 * config/arm/arm-tune.md: Regenerate.
19091 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19093 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19095 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19097 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19100 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19102 PR tree-optimization/81292
19103 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19104 full_string_p, also call adjust_related_strinfos if the adjustment
19105 is simple, otherwise invalidate related strinfos.
19107 2017-07-04 Martin Liska <mliska@suse.cz>
19110 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19111 newly created variable as DECL_IGNORED_P.
19113 2017-07-04 Martin Liska <mliska@suse.cz>
19116 * ipa-inline.c (inline_small_functions):
19117 Use xstrdup_for_dump.
19119 2017-07-04 Tom de Vries <tom@codesourcery.com>
19121 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19123 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19126 * config/darwin.c (darwin_function_switched_text_sections):
19129 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19131 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19133 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19135 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19137 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19139 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19140 min_profitable_iters, and th as inclusive lower bounds.
19141 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19142 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19143 for min_profitable_iters and min_profitable_estimate.
19144 (vect_transform_loop): Treat th as an inclusive lower bound.
19145 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19147 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19150 * config/darwin.c (darwin_function_switched_text_sections):
19151 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19152 in two pieces, and suppress the use of buf.
19154 2017-07-03 Nathan Sidwell <nathan@acm.org>
19156 * hash-table.h (hash_table_mod1): Fix indentation.
19158 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19160 PR middle-end/81290
19161 * predict.c (force_edge_cold): Be more careful about propagation
19163 * profile-count.h (profile_probability::guessed,
19164 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19166 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19168 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19170 * doc/invoke.texi (rcpc architecture extension): Document it.
19172 2017-07-03 Richard Biener <rguenther@suse.de>
19174 PR tree-optimization/60510
19175 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19176 the scalar reduction PHI and use it.
19177 (vectorizable_reduction): Properly guard the single_defuse_cycle
19178 path for non-SLP reduction chains where we cannot use it.
19179 Rework reduc_def/index and vector type deduction. Rework
19180 vector operand gathering during reduction op code-gen.
19181 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19182 chains dissolve the chain and leave it to non-SLP reduction
19185 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19187 * tree-data-ref.h (dr_alignment): Declare.
19188 * tree-data-ref.c (dr_alignment): New function.
19189 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19190 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19192 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19194 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19196 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19197 and base_misalignment fields.
19198 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19199 * tree-data-ref.c: Include builtins.h.
19200 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19201 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19202 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19203 * tree-vect-data-refs.c: Include tree-cfg.h.
19204 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19205 fields instead of calculating an alignment here.
19206 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19207 innermost_loop_behavior fields.
19209 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19211 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19213 (DR_STEP_ALIGNMENT): New macro.
19214 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19215 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19216 (create_data_ref): Print it.
19217 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19218 to tell whether the step preserves vector (mis)alignment.
19219 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19220 Move the check for an integer step and generalise to all INTEGER_CST.
19221 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19222 Print the outer step alignment.
19224 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19226 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19227 with offset_alignment.
19228 (DR_ALIGNED_TO): Delete.
19229 (DR_OFFSET_ALIGNMENT): New macro.
19230 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19231 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19232 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19233 (create_data_ref): Likewise.
19234 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19235 (vect_analyze_data_refs): Likewise.
19236 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19237 creating dummy innermost behavior.
19239 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19241 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19242 with a "innermost_loop_behavior *" and refeence tree.
19243 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19244 (create_data_ref): Update call accordingly.
19245 * tree-predcom.c (find_looparound_phi): Likewise.
19247 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19249 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19250 fields with dr_wrt_vec_loop.
19251 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19252 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19253 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19254 (vect_dr_behavior): New function.
19255 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19256 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19257 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19258 track whether the step preserves the misalignment.
19259 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19260 Use vect_dr_behavior.
19261 (vect_setup_realignment): Update call accordingly.
19262 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19263 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19264 call to vect_create_addr_base_for_vector_ref.
19265 (vect_create_cond_for_align_checks): Likewise.
19266 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19267 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19268 (vect_recog_mask_conversion_pattern): Likewise.
19269 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19270 (new_stmt_vec_info): Remove redundant zeroing.
19272 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19274 * common/config/arm/arm-common.c (arm_be8_option): New function.
19275 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19276 (ISA_ARMv6): Add isa_bit_be8.
19277 * config/arm/arm.h (arm_be8_option): Add prototype.
19278 (BE8_SPEC_FUNCTION): New define.
19279 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19280 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19281 (mlittle-endian): Similarly.
19282 (mbe8, mbe32): New options.
19283 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19284 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19286 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19288 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19290 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19292 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19293 (cleanup_tree_cfg_bb): Use it.
19294 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19296 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19298 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19301 * loop-doloop.c (add_test): Update profile.
19303 2017-07-03 Martin Liska <mliska@suse.cz>
19306 * sanopt.c (rewrite_usage_of_param): New function.
19307 (sanitize_rewrite_addressable_params): Likewise.
19308 (pass_sanopt::execute): Call rewrite_usage_of_param.
19310 2017-07-03 Richard Biener <rguenther@suse.de>
19312 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19313 back to using VIEW_CONVERT_EXPR.
19315 2017-07-03 Martin Liska <mliska@suse.cz>
19318 * doc/extend.texi: Document when a resolver function is
19319 generated for target_clones.
19321 2017-07-03 Martin Liska <mliska@suse.cz>
19323 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19324 and shadow memory of auto variables which are subject of
19325 use-after-scope sanitization.
19326 (asan_expand_mark_ifn): Add do set only when is_poison.
19328 2016-07-03 Richard Biener <rguenther@suse.de>
19330 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19332 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19333 (vectorizable_reduction): Perform reduction PHI creation when
19334 visiting a reduction PHI and adjust and simplify code generation
19335 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19336 (vect_transform_loop): Visit reduction PHIs.
19337 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19338 defs into the SLP tree.
19339 (vect_build_slp_tree): Reduction defs terminate the recursion.
19340 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19342 (vect_get_vec_defs_for_stmt_copy): Export.
19343 (vect_get_vec_defs): Likewise.
19344 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19346 (vect_get_vec_defs_for_stmt_copy): Declare.
19347 (vect_get_vec_defs): Likewise.
19349 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19351 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19352 parameter with a "loop" parameter and use it instead of the
19353 loop containing DR_STMT. Don't check simple_iv when doing
19354 BB analysis. Describe the two analysis modes in the comment.
19356 2017-07-03 Tom de Vries <tom@codesourcery.com>
19358 PR tree-optimization/69468
19359 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19360 (find_same_succ_bb): Handle ignore_edge_flags.
19362 2017-07-03 Tom de Vries <tom@codesourcery.com>
19364 PR tree-optimization/81192
19365 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19367 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19369 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19371 2017-07-03 Tom de Vries <tom@codesourcery.com>
19373 PR tree-optimization/81192
19374 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19375 BB_SAME_SUCC (bb) == NULL.
19377 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19379 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19382 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19384 * dumpfile.c: Include profile-count.h
19385 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19387 (insert_cond_bb): Update profile.
19388 * tree-cfg.h (insert_cond_bb): Update prototype.
19389 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19390 * tree-dump.c: Do not include tree-cfg.
19392 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19394 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19396 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19398 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19401 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19403 * tree-complex.c (expand_complex_div_wide): update profile.
19405 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19406 Alan Hayward <alan.hayward@arm.com>
19407 David Sherwood <david.sherwood@arm.com>
19409 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19410 (CORETYPES_H): New define.
19411 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19412 (insn-modes-inline.h, s-modes-inline-h): New rules.
19413 (generated_files): Add insn-modes-inline.h.
19414 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19415 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19416 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19417 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19418 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19419 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19420 (build/gencodes.o, build/genconditions.o): Likewise.
19421 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19422 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19423 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19424 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19425 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19426 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19427 * coretypes.h: Include everything up to real.h for generators.
19428 Include insn-modes.h first. Include wide-int-print.h after
19429 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19430 * machmode.h: Don't include insn-modes.h here.
19431 * function-tests.c: Remove includes of signop.h, machmode.h,
19432 double-int.h and wide-int.h.
19434 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19436 * optc-save-gen.awk: Likewise.
19437 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19438 * godump.c: Remove include of wide-int-print.h.
19439 * pretty-print.h: Likewise.
19440 * wide-int-print.cc: Likewise.
19441 * wide-int.cc: Likewise.
19442 * hash-map-tests.c: Remove include of signop.h.
19443 * hash-set-tests.c: Likewise.
19444 * rtl-tests.c: Likewise.
19445 * mkconfig.sh: Remove include of machmode.h.
19446 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19448 (emit_insn_modes_inline_h): ...this new function. Emit the code
19449 into an insn-modes-inline.h header file, adding appropriate
19450 include guards and end comments.
19451 (emit_insn_modes_c_header): Remove include of machmode.h.
19452 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19454 (main): Handle -i flag and call emit_insn_modes_inline_h when
19457 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19459 * tree-ssa-strlen.c (strinfo): Rename the length field to
19460 nonzero_chars. Add a full_string_p field.
19461 (compare_nonzero_chars, zero_length_string_p): New functions.
19462 (get_addr_stridx): Add an offset_out parameter.
19463 Use compare_nonzero_chars.
19464 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19465 (new_strinfo): Update after above changes to strinfo.
19466 (set_endptr_and_length): Set full_string_p.
19467 (get_string_length): Update after above changes to strinfo.
19468 (unshare_strinfo): Update call to new_strinfo.
19469 (maybe_invalidate): Likewise.
19470 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19471 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19472 as a uhwi instead of an shwi. Update after above changes to
19473 strinfo and new_strinfo.
19474 (zero_length_string): Assert that chainsi contains full strings.
19475 Use zero_length_string_p. Update call to new_strinfo.
19476 (adjust_related_strinfos): Update after above changes to strinfo.
19477 Copy full_string_p from origsi.
19478 (adjust_last_stmt): Use zero_length_string_p.
19479 (handle_builtin_strlen): Update after above changes to strinfo and
19480 new_strinfo. Install the lhs as the string length if the previous
19481 entry didn't describe a full string.
19482 (handle_builtin_strchr): Update after above changes to strinfo
19484 (handle_builtin_strcpy): Likewise.
19485 (handle_builtin_strcat): Likewise.
19486 (handle_builtin_malloc): Likewise.
19487 (handle_pointer_plus): Likewise.
19488 (handle_builtin_memcpy): Likewise. Track nonzero characters
19489 that aren't necessarily followed by a nul terminator.
19490 (handle_char_store): Likewise.
19492 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19494 PR tree-optimization/80769
19495 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19496 for malloc and calloc. Document the new invariant that all related
19497 strinfos have delayed lengths or none do.
19498 (verify_related_strinfos): Move earlier in file.
19499 (set_endptr_and_length): New function, split out from...
19500 (get_string_length): ...here. Also set the lengths of related
19502 (zero_length_string): Assert that chainsi has known (rather than
19504 (adjust_related_strinfos): Likewise.
19506 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19508 PR tree-optimization/81136
19509 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19510 assert that two references with the same misalignment have the same
19511 compile-time misalignment if those compile-time misalignments
19514 2017-07-01 Andi Kleen <ak@linux.intel.com>
19516 * print-tree.c (print_node): Print all attributes.
19518 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19520 * cfg.c (scale_bbs_frequencies): New function.
19521 * cfg.h (scale_bbs_frequencies): Declare it.
19522 * cfgloopanal.c (single_likely_exit): Cleanup.
19523 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19525 (scale_loop_profile): Likewise.
19526 (loop_version): Likewise.
19527 (create_empty_loop_on_edge): Update.
19528 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19529 scale_loop_frequencies, scale_loop_profile, loopify,
19530 loop_version): Update prototypes.
19531 * modulo-sched.c (sms_schedule): Update.
19532 * predict.c (unlikely_executed_edge_p): Also check probability.
19533 (probably_never_executed_edge_p): Fix typo.
19534 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19535 * tree-parloops.c (gen_parallel_loop): Update.
19536 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19537 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19538 * tree-ssa-loop-split.c (split_loop): Update.
19539 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19540 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19541 (vect_loop_versioning): Update.
19542 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19544 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19546 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19548 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19550 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19551 to keep profile consistent.
19553 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19555 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19556 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19557 * profile-count.h (max_safe_multiplier): Make unsigned.
19558 (profile_count::guessed_zero): New.
19560 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19562 * bb-reorder.c (fix_up_crossing_landing_pad,
19563 fix_crossing_conditional_branches): Use make_single_succ_edge
19564 to keep profile consistent.
19566 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19568 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19571 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19574 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19575 the right scopes, make sure cond_jump isn't preserved between multiple
19576 iterations. Search for fallthru edge whenever there are 3+ edges and
19577 use find_fallthru_edge for it.
19579 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19581 Patch by Alexander Monakov <amonakov@ispras.ru>
19582 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19583 probabilities consistently.
19585 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19587 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19588 emit_cmp_and_jump_insns.
19590 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19593 * tree-inline.c (expand_call_inline): Combine profile statuses.
19595 2017-06-30 Andrew Pinski <apinski@cavium.com>
19597 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19598 fold_stmt returned true.
19600 2017-06-30 Nathan Sidwell <nathan@acm.org>
19602 * ggc.h (empty_string): Delete.
19603 * cfgexpand.c (expand_asm_stmt): Use plain "".
19604 * optabs.c (expand_asm_memory_barrier): Likewise.
19605 * stringpool.c (empty_string): Delete.
19606 (digit_vector, digit_string): Delete.
19607 (ggc_alloc_string): Use plain "", don't optimize single digit
19608 strings. Use ggc_alloc_atomic.
19610 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19612 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19613 comparison set and one other set, use the cost of the non-comparison
19616 2017-06-30 Nathan Sidwell <nathan@acm.org>
19618 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19621 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19623 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19624 loops. Remove now unneeded calls to gimple_switch_set_label() that
19625 just set removed labels to NULL_TREE.
19627 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19629 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19630 (set_range_info): ...here. Only call set_range_info_raw if domain
19632 (set_nonzero_bits): Call set_range_info_raw.
19633 * tree-ssanames.h (set_range_info_raw): New.
19635 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19638 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19639 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19640 of nonimmediate_operand and <store_mask_constraint> instead of m
19641 for the input operand. For V8FI iterator, always split if input
19642 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19643 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19644 <store_mask_predicate> instead of register_operand and
19645 <store_mask_constraint> instead of v for the input operand. Make
19646 sure both operands aren't MEMs for if not <mask_applied>.
19648 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19650 * lto-wrapper.c (copy_file) Close both file descriptors before
19653 2017-06-30 Martin Liska <mliska@suse.cz>
19656 * multiple_target.c (create_dispatcher_calls): Make ifunc
19657 also for function that don't have calls or are not referenced.
19659 2017-06-30 Richard Biener <rguenther@suse.de>
19661 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19662 analyze the first scalar stmt. Move vector type computation
19663 for the BB case here from ...
19664 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19665 live operation processing in the SLP case properly.
19667 2017-06-30 Richard Biener <rguenther@suse.de>
19669 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19671 2017-06-30 Martin Liska <mliska@suse.cz>
19674 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19675 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19677 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19679 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19682 2017-06-30 Richard Biener <rguenther@suse.de>
19684 PR tree-optimization/81249
19685 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19686 condition reduction result to original scalar type.
19688 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19690 * profile-count.h (enum profile_quality): Fix typos and whitespace
19693 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19695 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19696 type for branch probabilities.
19698 2017-06-29 Julian Brown <julian@codesourcery.com>
19699 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19701 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19702 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19703 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19704 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19706 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19708 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19709 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19710 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19711 CC usage from generic code to here.
19712 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19713 CC usage into the target macros.
19715 2017-06-29 Maya Rashish <coypu@sdf.org>
19717 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19720 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19722 * arm/arm-builtins.c: Include profile-count.h
19723 * except.c (sjlj_emit_function_enter): Use
19724 profile_probability::unlikely.
19726 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19728 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19729 and tocrel_offset be pointer args rather than implicitly using
19731 (legitimate_constant_pool_address_p, rs6000_emit_move,
19732 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19733 tocrel_offset and use in toc_relative_expr_p call.
19734 (print_operand, print_operand_address): Use static tocrel_base_oac
19735 and tocrel_offset_oac.
19736 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19739 2017-06-29 Maya Rashish <coypu@sdf.org>
19741 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19743 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19745 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19746 objects, take into account only the alignment of 'op0' and 'mode1' if
19749 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19751 * ccmp.c (ccmp_tree_comparison_p): New function.
19752 (ccmp_candidate_p): Update to use above function.
19753 (get_compare_parts): New function.
19754 (expand_ccmp_next): Update to use new functions.
19755 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19757 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19758 take mode as argument.
19759 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19760 * expr.c (expand_expr_real_1): Pass mode as argument.
19762 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19764 * combine.c (combine_instructions): Print insns to dump_file, together
19767 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19769 * asan.c (asan_emit_stack_protection): Update.
19770 (create_cond_insert_point): Update.
19771 * auto-profile.c (afdo_propagate_circuit): Update.
19772 * basic-block.h (struct edge_def): Turn probability to
19773 profile_probability.
19774 (EDGE_FREQUENCY): Update.
19775 * bb-reorder.c (find_traces_1_round): Update.
19776 (better_edge_p): Update.
19777 (sanitize_hot_paths): Update.
19778 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19779 (make_single_succ_edge): Update.
19780 (check_bb_profile): Update.
19781 (dump_edge_info): Update.
19782 (update_bb_profile_for_threading): Update.
19783 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19784 probabilitycount to 0.
19785 * cfgbuild.c (compute_outgoing_frequencies): Update.
19786 * cfgcleanup.c (try_forward_edges): Update.
19787 (outgoing_edges_match): Update.
19788 (try_crossjump_to_edge): Update.
19789 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19790 (expand_gimple_tailcall): Update.
19791 (construct_init_block): Use make_single_succ_edge.
19792 (construct_exit_block): Use make_single_succ_edge.
19793 * cfghooks.c (verify_flow_info): Update.
19794 (redirect_edge_succ_nodup): Update.
19795 (split_edge): Update.
19796 (account_profile_record): Update.
19797 * cfgloopanal.c (single_likely_exit): Update.
19798 * cfgloopmanip.c (scale_loop_profile): Update.
19799 (set_zero_probability): Remove.
19800 (duplicate_loop_to_header_edge): Update.
19801 * cfgloopmanip.h (loop_version): Update prototype.
19802 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19803 (force_nonfallthru_and_redirect): Update.
19804 (update_br_prob_note): Update.
19805 (rtl_verify_edges): Update.
19806 (purge_dead_edges): Update.
19807 (rtl_lv_add_condition_to_bb): Update.
19808 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19809 * cgraphunit.c (init_lowered_empty_function): Update.
19810 (cgraph_node::expand_thunk): Update.
19811 * cilk-common.c: Include profile-count.h
19812 * dojump.c (inv): Remove.
19813 (jumpifnot): Update.
19814 (jumpifnot_1): Update.
19815 (do_jump_1): Update.
19817 (do_jump_by_parts_greater_rtx): Update.
19818 (do_compare_rtx_and_jump): Update.
19819 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19820 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19821 * dwarf2cfi.c: Include profile-count.h
19822 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19823 (sjlj_emit_dispatch_table): Likewise.
19824 * explow.c: Include profile-count.h
19825 * expmed.c (emit_store_flag_force): Update.
19826 (do_cmp_and_jump): Update.
19827 * expr.c (compare_by_pieces_d::generate): Update.
19828 (compare_by_pieces_d::finish_mode): Update.
19829 (emit_block_move_via_loop): Update.
19830 (store_expr_with_bounds): Update.
19831 (store_constructor): Update.
19832 (expand_expr_real_2): Update.
19833 (expand_expr_real_1): Update.
19834 * expr.h (try_casesi, try_tablejump): Update prototypes.
19835 * gimple-pretty-print.c (dump_probability): Update.
19836 (dump_profile): New.
19837 (dump_gimple_label): Update.
19838 (dump_gimple_bb_header): Update.
19839 * graph.c (draw_cfg_node_succ_edges): Update.
19840 * hsa-gen.c (convert_switch_statements): Update.
19841 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19842 (find_if_case_1): Update.
19843 (find_if_case_2): Update.
19844 * internal-fn.c (expand_arith_overflow_result_store): Update.
19845 (expand_addsub_overflow): Update.
19846 (expand_neg_overflow): Update.
19847 (expand_mul_overflow): Update.
19848 (expand_vector_ubsan_overflow): Update.
19849 * ipa-cp.c (good_cloning_opportunity_p): Update.
19850 * ipa-split.c (split_function): Use make_single_succ_edge.
19851 * ipa-utils.c (ipa_merge_profiles): Update.
19852 * loop-doloop.c (add_test): Update.
19853 (doloop_modify): Update.
19854 * loop-unroll.c (compare_and_jump_seq): Update.
19855 (unroll_loop_runtime_iterations): Update.
19856 * lra-constraints.c (lra_inheritance): Update.
19857 * lto-streamer-in.c (input_cfg): Update.
19858 * lto-streamer-out.c (output_cfg): Update.
19859 * mcf.c (adjust_cfg_counts): Update.
19860 * modulo-sched.c (sms_schedule): Update.
19861 * omp-expand.c (expand_omp_for_init_counts): Update.
19862 (extract_omp_for_update_vars): Update.
19863 (expand_omp_ordered_sink): Update.
19864 (expand_omp_for_ordered_loops): Update.
19865 (expand_omp_for_generic): Update.
19866 (expand_omp_for_static_nochunk): Update.
19867 (expand_omp_for_static_chunk): Update.
19868 (expand_cilk_for): Update.
19869 (expand_omp_simd): Update.
19870 (expand_omp_taskloop_for_outer): Update.
19871 (expand_omp_taskloop_for_inner): Update.
19872 * omp-simd-clone.c (simd_clone_adjust): Update.
19873 * optabs.c (expand_doubleword_shift): Update.
19874 (expand_abs): Update.
19875 (emit_cmp_and_jump_insn_1): Update.
19876 (expand_compare_and_swap_loop): Update.
19877 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19878 * predict.c (predictable_edge_p): Update.
19879 (edge_probability_reliable_p): Update.
19880 (set_even_probabilities): Update.
19881 (combine_predictions_for_insn): Update.
19882 (combine_predictions_for_bb): Update.
19883 (propagate_freq): Update.
19884 (estimate_bb_frequencies): Update.
19885 (force_edge_cold): Update.
19886 * profile-count.c (profile_count::dump): Add missing space into dump.
19887 (profile_count::debug): Add newline.
19888 (profile_count::differs_from_p): Explicitly convert to unsigned.
19889 (profile_count::stream_in): Update.
19890 (profile_probability::dump): New member function.
19891 (profile_probability::debug): New member function.
19892 (profile_probability::differs_from_p): New member function.
19893 (profile_probability::differs_lot_from_p): New member function.
19894 (profile_probability::stream_in): New member function.
19895 (profile_probability::stream_out): New member function.
19896 * profile-count.h (profile_count_quality): Rename to ...
19897 (profile_quality): ... this one.
19898 (profile_probability): New.
19899 (profile_count): Update.
19900 * profile.c (compute_branch_probabilities): Update.
19901 * recog.c (peep2_attempt): Update.
19902 * sched-ebb.c (schedule_ebbs): Update.
19903 * sched-rgn.c (find_single_block_region): Update.
19904 (compute_dom_prob_ps): Update.
19905 (schedule_region): Update.
19906 * sel-sched-ir.c (compute_succs_info): Update.
19907 * stmt.c (struct case_node): Update.
19908 (do_jump_if_equal): Update.
19909 (get_outgoing_edge_probs): Update.
19910 (conditional_probability): Update.
19911 (emit_case_dispatch_table): Update.
19912 (expand_case): Update.
19913 (expand_sjlj_dispatch_table): Update.
19914 (emit_case_nodes): Update.
19915 * targhooks.c: Update.
19916 * tracer.c (better_p): Update.
19917 (find_best_successor): Update.
19918 * trans-mem.c (expand_transaction): Update.
19919 * tree-call-cdce.c: Update.
19920 * tree-cfg.c (gimple_split_edge): Upate.
19921 (move_sese_region_to_fn): Upate.
19922 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19923 * tree-eh.c (lower_resx): Upate.
19924 (cleanup_empty_eh_move_lp): Upate.
19925 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19926 * tree-inline.c (copy_edges_for_bb): Update.
19927 (copy_cfg_body): Update.
19928 * tree-parloops.c (gen_parallel_loop): Update.
19929 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19930 (gimple_gen_time_profiler): Update.
19931 * tree-ssa-dce.c (remove_dead_stmt): Update.
19932 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19933 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19934 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19935 (unloop_loops): Update.
19936 (try_peel_loop): Update.
19937 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19938 * tree-ssa-loop-split.c (connect_loops): Update.
19939 (split_loop): Update.
19940 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19941 (hoist_guard): Update.
19942 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19943 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19944 (value_replacement): Update.
19945 * tree-ssa-reassoc.c (branch_fixup): Update.
19946 * tree-ssa-tail-merge.c (replace_block_by): Update.
19947 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19948 (create_edge_and_update_destination_phis): Update.
19949 (compute_path_counts): Update.
19950 (recompute_probabilities): Update.
19951 (update_joiner_offpath_counts): Update.
19952 (freqs_to_counts_path): Update.
19953 (duplicate_thread_path): Update.
19954 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19955 (struct switch_conv_info): Update.
19956 (gen_inbound_check): Update.
19957 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19958 (vect_do_peeling): Update.
19959 (vect_loop_versioning): Update.
19960 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19961 (optimize_mask_stores): Update.
19962 * ubsan.c (ubsan_expand_null_ifn): Update.
19963 * value-prof.c (gimple_divmod_fixed_value): Update.
19964 (gimple_divmod_fixed_value_transform): Update.
19965 (gimple_mod_pow2): Update.
19966 (gimple_mod_pow2_value_transform): Update.
19967 (gimple_mod_subtract): Update.
19968 (gimple_mod_subtract_transform): Update.
19969 (gimple_ic): Update.
19970 (gimple_stringop_fixed_value): Update.
19971 (gimple_stringops_transform): Update.
19972 * value-prof.h: Update.
19974 2017-06-29 Carl Love <cel@us.ibm.com>
19976 * config/rs6000/rs6000-c.c: Add support for built-in functions
19977 vector signed int vec_signed (vector float);
19978 vector signed long long vec_signed (vector double);
19979 vector signed int vec_signed2 (vector double, vector double);
19980 vector signed int vec_signede (vector double);
19981 vector signed int vec_signedo (vector double);
19982 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19983 instruction generator.
19984 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19985 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19986 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19988 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19989 vunsignede_v2df): Add define_expands.
19990 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19991 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19992 VEC_UNSIGNEDO): Add definitions.
19993 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19994 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19995 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19996 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19997 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19998 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19999 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20000 vec_unsignede, vec_unsignedo): Add builtin defines.
20001 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20003 * doc/extend.texi: Update the built-in documentation file for the
20004 new built-in functions.
20006 2017-06-29 Richard Biener <rguenther@suse.de>
20008 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20009 reduction chains to LOOP_VINFO_REDUCTIONS.
20010 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20011 SLP reductions after processing reduction chains.
20013 2017-06-29 Nathan Sidwell <nathan@acm.org>
20015 * builtins.c (fold_builtin_FUNCTION): Use
20016 lang_hooks.decl_printable_name.
20018 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20020 PR middle-end/81194
20021 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20022 with only one label.
20023 * stmt.c (expand_case): Assert NCASES is greater than one.
20025 2017-06-29 Richard Biener <rguenther@suse.de>
20027 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20029 (group_case_labels): Likewise.
20030 (find_taken_edge): Push sanity checking on val to workers...
20031 (find_taken_edge_cond_expr): ... here
20032 (find_taken_edge_switch_expr): ... and here, handle cases
20033 with just a default label.
20034 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20035 (group_case_labels): Likewise.
20036 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20037 group_case_labels does anything cleanup the CFG again.
20039 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20041 PR tree-optimization/81196
20042 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20043 exit condition comparing two IVs.
20045 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20047 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20048 profile to the dummy entry at the end of the list of architectures.
20049 * config/arm/arm-cpu-cdata.h: Regenerated.
20051 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20052 Michael Collison <michael.collison@arm.com>
20055 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20057 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20058 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20059 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20060 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20061 of shift when the shift amount is masked with constant equal to
20062 the size of the mode.
20063 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20066 2017-06-29 Martin Liska <mliska@suse.cz>
20068 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20070 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20073 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20076 (edge_growth_cache_entry::edge_growth_cache_entry): New
20078 (reset_edge_growth_cache): Update to use constructor.
20080 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20082 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20083 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20084 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20086 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20088 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20089 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20091 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20093 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20094 (*-linux-uclibc*): Add t-uclibc tmake_file.
20095 * config/t-musl: New.
20096 * config/t-uclibc: New.
20098 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20100 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20102 (gen_comm_data): Emit architectural setting of arch_prof.
20103 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20105 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20106 (armv8-m.base, armv8-m.main): Likewise.
20107 * arm-protos.h (arm_build_target): Add profile field.
20108 (arch_option): Likewise.
20109 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20111 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20112 arm_active_target.profile.
20114 2017-06-28 Richard Biener <rguenther@suse.de>
20116 PR middle-end/81227
20117 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20118 TYPE_OVERFLOW_WRAPS.
20119 * match.pd (negate_expr_p): Likewise.
20120 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20121 fold_build2, not fold_binary.
20123 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20125 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20126 Convert memory address to Pmode.
20127 (aarch64_print_operand): Assert MEM operands are always Pmode.
20129 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20132 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20133 Remove redundant if.
20134 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20135 * config/arm/aarch-common-protos.h
20136 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20137 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20139 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20142 * multiple_target.c (create_dispatcher_calls): Set the default
20143 clone to be static, not public.
20145 2017-06-28 Richard Biener <rguenther@suse.de>
20147 * tree-vect-loop.c (vectorizable_reduction): Move special
20148 cond reduction IV var creation ...
20149 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20150 parameter. Use STMT_VINFO_VECTYPE.
20151 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20154 2017-06-28 Martin Liska <mliska@suse.cz>
20157 * ipa-visibility.c (non_local_p): Handle visibility.
20159 2017-06-28 Martin Liska <mliska@suse.cz>
20162 * common.opt: Add IntegerRange to various options.
20163 * opt-functions.awk (integer_range_info): New function.
20164 * optc-gen.awk: Add integer_range_info to cl_options struct.
20165 * opts-common.c (decode_cmdline_option): Handle
20166 CL_ERR_INT_RANGE_ARG.
20167 (cmdline_handle_error): Likewise.
20168 * opts.c (print_filtered_help): Show valid interval in
20169 when --help is provided.
20170 * opts.h (struct cl_option): Add range_min and range_max fields.
20171 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20173 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20175 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20176 (x * C EQ/NE y * C): New transformation.
20178 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20180 * genmultilib (combination_space): Accept '+' in option names.
20182 2017-06-28 Martin Liska <mliska@suse.cz>
20185 * asan.c (instrument_derefs): Bail out inner references
20186 that are hard register variables.
20188 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20191 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20192 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20194 2017-06-28 Richard Biener <rguenther@suse.de>
20196 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20197 (vect_get_slp_defs): Adjust.
20198 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20200 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20202 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20203 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20204 (vectorizable_reduction): Adjust.
20205 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20207 (vect_get_slp_defs): Likewise.
20208 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20209 (vectorizable_bswap): Adjust.
20210 (vectorizable_call): Likewise.
20211 (vectorizable_conversion): Likewise.
20212 (vectorizable_assignment): Likewise.
20213 (vectorizable_shift): Likewise.
20214 (vectorizable_operation): Likewise.
20215 (vectorizable_store): Likewise.
20216 (vectorizable_condition): Likewise.
20217 (vectorizable_comparison): Likewise.
20219 2017-06-28 Michael Collison <michael.collison@arm.com>
20222 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20224 (arm_gen_movmemqi): Removed unused variable 'i'.
20225 Convert 'for' loop into 'while' loop.
20226 (arm_expand_prologue): Remove last unnecessary set of insn.
20227 (thumb_pop): Remove unused variable 'pushed_words'.
20228 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20230 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20232 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20233 * config/s390/s390-protos.h: Add prototype of
20234 s390_rel_address_ok_p.
20235 * config/s390/s390.c (s390_got_symbol): New function.
20236 (s390_rel_address_ok_p): New function.
20237 (legitimize_pic_address): Use s390_rel_address_ok_p.
20238 (s390_load_got): Use s390_got_symbol.
20239 (s390_option_override): Issue error if
20240 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20241 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20243 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20245 2017-06-27 Andrew Pinski <apinski@cavium.com>
20247 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20248 (X * copysign (1.0, X)): New pattern.
20249 (X * copysign (1.0, -X)): New pattern.
20250 (copysign (-1.0, CST)): New pattern.
20252 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20254 * genmultilib (combination_space): Remove variable.
20255 Validate reuse rules against regular expression for any sequence
20256 of multilib options in any order.
20258 2017-06-27 Michael Collison <michael.collison@arm.com>
20260 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20261 call aarch64_split_simd_combine.
20262 * (aarch64_combine_internal<mode>): Delete pattern.
20263 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20264 Allow register and subreg operands.
20266 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20268 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20269 specific need, just fallback on defaults.
20270 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20272 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20273 Olivier Hainque <hainque@adacore.com>
20275 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20277 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20278 targets. Pick a default if no particular attempt applied.
20279 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20282 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20284 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20285 (x86_64-wrs-vxworks7): Likewise.
20287 2017-06-27 Marek Polacek <polacek@redhat.com>
20290 * ubsan.c (instrument_null): Check get_base_address's result for null.
20292 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20294 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20296 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20298 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20299 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20300 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20301 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20302 New function types.
20303 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20304 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20305 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20306 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20307 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20308 BUILT_IN_FEUPDATEENV): New builtins.
20309 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20310 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20311 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20312 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20314 (builtin_structptr_types): Adjust size.
20315 * tree.c (builtin_structptr_types): Add four entries.
20317 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20318 Olivier Hainque <hainque@adacore.com>
20320 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20321 (TLS_SYM): New local macro, forcing reference to __tls__ on
20322 link command lines for VxWorks 7 RTPs, triggering initialization
20324 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20325 OS features TLS support, true for RTPs on VxWorks 7.
20326 * config/vxworks.c (vxworks_override_options): Setup emutls
20329 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20331 * predict.c (test_prediction_value_range): Use -1U instead of -1
20332 to avoid narrowing conversion warning.
20333 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20334 to avoid narrowing conversion warning.
20335 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20337 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20339 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20341 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20342 64bit configurations.
20343 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20344 (SIZE_TYPE): Likewise.
20345 * config/vxworks.c (vxworks_emutls_var_fields): Use
20346 long_unsigned_type_node instead of unsigned_type_node as the offset
20347 field type, which is "pointer" mode in emutls.c.
20349 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20352 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20354 PR middle-end/81207
20355 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20356 gimple_vuse copying separately from gimple_vdef copying.
20358 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20360 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20363 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20364 Olivier Hainque <hainque@adacore.com>
20366 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20367 for all vxworks7 targets.
20368 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20369 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20370 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20371 variations for VX6/VX7 and 32/64bits later on in ...
20372 (VXWORKS_LIB_SPEC): Leverage new macros.
20373 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20374 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20376 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20378 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20379 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20381 2017-06-26 Carl Love <cel@us.ibm.com>
20383 * config/rs6000/rs6000-c.c: Add support for built-in functions
20384 vector bool char vec_reve (vector bool char);
20385 vector signed char vec_reve (vector signed char);
20386 vector unsigned char vec_reve (vector unsigned char);
20387 vector bool int vec_reve (vector bool int);
20388 vector signed int vec_reve (vector signed int);
20389 vector unsigned int vec_reve (vector unsigned int);
20390 vector bool long long vec_reve (vector bool long long);
20391 vector signed long long vec_reve (vector signed long long);
20392 vector unsigned long long vec_reve (vector unsigned long long);
20393 vector bool short vec_reve (vector bool short);
20394 vector signed short vec_reve (vector signed short);
20395 vector double vec_reve (vector double);
20396 vector float vec_reve (vector float);
20397 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20398 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20399 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20400 (altivec_vreve): New pattern.
20401 * config/rs6000/altivec.h (vec_reve): New define.
20402 * doc/extend.texi (vec_rev): Update the built-in documentation file
20403 for the new built-in functions.
20405 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20407 PR tree-optimization/71815
20408 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20410 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20412 (slsr_process_phi): Likewise.
20413 (replace_uncond_cands_and_profitable_phis): Don't replace a
20414 multiply candidate with a stride of 1 (copy or cast).
20415 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20417 (lowest_cost_path): Likewise.
20418 (total_savings): Likewise.
20420 2017-06-26 Richard Biener <rguenther@suse.de>
20423 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20424 Use def_builtin_pure for all gather builtins.
20426 2017-06-26 Richard Biener <rguenther@suse.de>
20428 PR tree-optimization/81203
20429 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20430 non-dominating BBs.
20432 2017-06-26 Marek Polacek <polacek@redhat.com>
20435 * doc/invoke.texi: Document -Wmultistatement-macros.
20437 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20439 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20440 arm_neon_ok_no_float_abi effective target.
20442 2017-06-26 Richard Biener <rguenther@suse.de>
20444 PR tree-optimization/80928
20445 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20446 (copy_bbs): Set BB_DUPLICATED flag early.
20447 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20449 (execute_on_shrinking_pred): Likewise.
20450 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20451 BB_DUPLICATED blocks.
20452 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20453 iterate over all PHIs considering removal of *gsi.
20455 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20457 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20460 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20462 * config/rs6000/rs6000-string.c: (expand_block_clear,
20463 do_load_for_compare, select_block_compare_mode,
20464 compute_current_alignment, expand_block_compare,
20465 expand_strncmp_align_check, expand_strn_compare,
20466 expand_block_move, rs6000_output_load_multiple)
20467 Move functions related to string/block move/compare
20468 to a separate file.
20469 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20470 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20471 for this function which is now used in two files.
20472 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20473 * config.gcc: Add rs6000-string.o to extra_objs for
20474 targets powerpc*-*-* and rs6000*-*-*.
20476 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20479 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20480 32-bit, since indexed is not valid for DImode.
20481 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20482 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20483 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20484 (define_peephole2 for Altivec d-form store): Likewise.
20487 * multiple_target.c (create_dispatcher_calls): Only create the
20488 dispatcher call if the function is the default clone of a
20489 versioned function.
20491 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20493 PR middle-end/80902
20494 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20495 a call, force the call to not be a tail call.
20497 2017-06-23 Jeff Law <law@redhat.com>
20499 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20502 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20504 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20505 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20506 handling for early expansion of vector shifts (sl,sr,sra,rl).
20507 (builtin_function_type): Add vector shift right instructions
20508 to the unsigned argument list.
20510 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20512 rtl-optimizatoin/79286
20513 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20514 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20515 trap. PIC register plus a const unspec without offset can never trap.
20517 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20519 * tree.h (builtin_structptr_type): New type.
20520 (builtin_structptr_types): Declare new array.
20521 * tree.c (builtin_structptr_types): New array.
20522 (free_lang_data, build_common_tree_nodes): Use it.
20524 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20527 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20530 2017-06-22 Matt Turner <mattst88@gmail.com>
20532 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20533 Lake models to skylake case. Assume skylake for unknown
20534 models with clflushopt.
20536 2017-06-22 Jeff Law <law@redhat.com>
20538 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20539 frame sizes that do not satisfy aarch64_uimm12_shift.
20541 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20543 * profile-count.h (apply_probability,
20544 apply_scale, probability_in): Fix checks for zero.
20546 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20548 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20549 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20551 2016-06-22 Richard Biener <rguenther@suse.de>
20553 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20554 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20555 REDUC_MAX_EXPR support.
20556 (vectorizable_reduction): Likewise.
20557 (vect_create_epilog_for_reduction): Likewise.
20559 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20561 * match.pd (A / (1 << B) -> A >> B): New.
20562 * generic-match-head.c: Include optabs-tree.h.
20563 * gimple-match-head.c: Likewise.
20564 * optabs-tree.h (target_supports_op_p): New.
20565 * optabs-tree.c (target_supports_op_p): New.
20567 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20569 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20570 $gcc_cv_ld --help output.
20571 (gcc_cv_ld_demangle): Likewise.
20572 (gcc_cv_ld_eh_frame_hdr): Likewise.
20573 (gcc_cv_ld_pie): Likewise.
20574 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20575 (gcc_cv_ld_buildid): Likewise.
20576 (gcc_cv_ld_sysroot): Likewise.
20577 (ld_bndplt_support): Likewise.
20578 (ld_pushpopstate_support): Likewise.
20579 * configure: Regenerate.
20580 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20582 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20585 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20586 operands indexes to avoid gap between operands and match_dups.
20588 2017-06-21 Andrew Pinski <apinski@cavium.com>
20590 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20591 Increment Arith_shift and Arith_shift_reg by 1.
20592 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20594 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20595 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20596 (aarch64_strip_extend): Add new argument and test for it.
20597 (aarch64_cheap_mult_shift_p): New function.
20598 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20599 add a cost if it is true.
20600 Update calls to aarch64_strip_extend.
20601 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20603 2017-06-21 Andrew Pinski <apinski@cavium.com>
20605 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20607 (thunderxt88): Likewise.
20608 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20609 (thunderx_prefetch_tune): New variable.
20610 (thunderx2t99_prefetch_tune): Update for the correct values.
20611 (thunderxt88_tunings): New variable.
20612 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20613 generic_prefetch_tune.
20614 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20616 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20618 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20619 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20620 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20621 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20622 (aarch64_atomic_cas<mode>, GPI): Likewise.
20624 2017-06-21 Martin Liska <mliska@suse.cz>
20626 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20627 statements on cold and hot labels.
20628 * predict.c (tree_estimate_probability_bb): Remove the
20629 prediction from this place.
20631 2017-06-21 Martin Liska <mliska@suse.cz>
20633 PR tree-optimization/79489
20634 * gimplify.c (maybe_add_early_return_predict_stmt): New
20636 (gimplify_return_expr): Call the function.
20637 * predict.c (tree_estimate_probability_bb): Remove handling
20639 * predict.def: Update comment about early return predictor.
20640 * gimple-predict.h (is_gimple_predict): New function.
20641 * predict.def: Change default value of early return to 66.
20642 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20644 * passes.def: Put pass_strip_predict_hints to the beginning of
20647 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20649 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20650 FUNCTION_DECL declarations.
20651 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20653 (dwaf2out_decl): Likewise.
20654 * godump.c (go_early_global_decl): Skip call to the real debug hook
20655 for FUNCTION_DECL declarations.
20656 * passes.c (rest_of_decl_compilation): Skip call to the
20657 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20658 -fdump-go-spec is passed.
20660 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20662 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20663 Reorder for compactness.
20664 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20665 (def_builtin_pure, def_builtin_pure2): New functions.
20666 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20668 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20670 * match.pd (nop_convert): New predicate.
20671 ((A +- CST1) +- CST2): Allow some NOP conversions.
20673 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20676 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20677 with ctors/dtors if GOVD_SHARED is set.
20679 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20681 * config/aarch64/aarch64.md (movti_aarch64):
20682 Emit mov rather than orr.
20683 (movtf_aarch64): Likewise.
20684 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20685 Emit mov rather than orr.
20687 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20689 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20690 Swap alternatives, make integer dup more expensive.
20692 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20694 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20695 Return true for non-tls symbols.
20697 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20699 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20700 (cortex-a75): Likewise.
20701 (cortex-a75.cortex-a55): Likewise.
20702 * config/aarch64/aarch64-tune.md: Regenerate.
20703 * doc/invoke.texi (-mtune): Document new values for -mtune.
20705 2017-06-21 Tom de Vries <tom@codesourcery.com>
20707 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20708 stack_size feature.
20709 (Effective-Target Keywords, Other attributes): Suggest using
20710 dg-add-options stack_size feature to get stack limit in stack_size
20711 effective target documentation.
20713 2017-06-21 Julian Brown <julian@codesourcery.com>
20714 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20716 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20717 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20718 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20720 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20721 attribute type list for neon_multiply.
20722 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20723 attribute type list for neon_multiply.
20724 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20725 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20726 attribute type list for neon_multiply.
20727 * config/arm/types.md (crypto_pmull): Add.
20728 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20729 attribute type list.
20731 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20733 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20736 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20738 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20739 to make sure not to dereference a NULL cost_classes_ptr pointer.
20741 2017-06-20 Carl Love <cel@us.ibm.com>
20743 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20744 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20745 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20746 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20747 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20748 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20749 VMULOSW): New enum "unspec" values.
20750 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20751 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20752 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20753 altivec_vmulosw): New patterns.
20754 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20755 VMULOSW): Add definitions.
20757 2017-06-20 Julia Koval <julia.koval@intel.com>
20759 * config/i386/i386.c: Fix rounding expand for new pattern.
20760 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20762 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20764 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20765 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20767 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20769 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20772 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20774 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20775 architecture, then by implementer ID.
20776 * config/aarch64/aarch64-tune.md: Regenerate.
20778 2017-06-20 Richard Biener <rguenther@suse.de>
20780 PR middle-end/81097
20781 * fold-const.c (split_tree): Fold to type before negating.
20783 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20785 * diagnostic-show-locus.c
20786 (selftest::test_fixit_deletion_affecting_newline): New function.
20787 (selftest::diagnostic_show_locus_c_tests): Call it.
20789 2017-06-20 Andreas Schwab <schwab@suse.de>
20792 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20795 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20797 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20798 __ARM_FEATURE_COPROC according to support.
20800 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20802 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20803 Rewritten to avoid overflow for > 32-bit pointers.
20806 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20807 by removing enum keyword.
20808 (ubsan_type_descriptor): Likewise. Formatting fix.
20811 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20812 splitter): Require TARGET_SSE2 in the condition.
20814 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20817 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20818 for doing vector set of SFmode on ISA 3.0.
20819 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20820 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20822 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20823 SFmode value into a V4SF variable that was extracted from another
20824 V4SF variable without converting the element to double precision
20825 and back to single precision vector format.
20826 (vsx_insert_extract_v4sf_p9_2): Likewise.
20828 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20830 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20831 in UWHI to avoid undefined overflow.
20834 * ubsan.h (enum ubsan_encode_value_phase): New.
20835 (ubsan_encode_value): Change second argument to
20836 enum ubsan_encode_value_phase with default value of
20837 UBSAN_ENCODE_VALUE_GENERIC.
20838 * ubsan.c (ubsan_encode_value): Change second argument to
20839 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20840 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20841 create_tmp_var_raw instead of create_tmp_var and use a
20843 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20844 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20845 ubsan_encode_value callers.
20848 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20849 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20850 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20852 2017-06-19 Richard Biener <rguenther@suse.de>
20854 PR middle-end/81118
20855 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20856 estimates if we changed anything.
20858 2017-06-19 Richard Biener <rguenther@suse.de>
20860 PR tree-optimization/80887
20861 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20862 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20863 simplified lookups, then reset mprts_hook.
20864 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20866 (try_to_simplify): Likewise.
20868 2017-06-19 Martin Liska <mliska@suse.cz>
20871 * gimplify.c (gimplify_switch_expr):
20872 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20874 2017-06-19 Martin Liska <mliska@suse.cz>
20876 * doc/install.texi: Document that PGO runs in 4 stages.
20878 2017-06-19 Martin Liska <mliska@suse.cz>
20881 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20882 to dispatcher function name.
20883 * multiple_target.c (replace_function_decl): New function.
20884 (create_dispatcher_calls): Redirect both edges and references.
20886 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20888 * profile-count.c (profile_count::dump): Dump quality.
20889 (profile_count::differs_from_p): Update for unsigned val.
20890 * profile-count.h (profile_count_quality): New enum.
20891 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20893 2017-06-19 Richard Biener <rguenther@suse.de>
20895 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20896 struct function as arg.
20897 (estimate_numbers_of_iterations): Export overload with loop arg.
20898 (free_numbers_of_iterations_estimates_loop): Use an overload of
20899 free_numbers_of_iterations_estimates instead.
20900 * tree-cfg.c (remove_bb): Adjust.
20901 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20902 * tree-parloops.c (gen_parallel_loop): Likewise.
20903 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20905 (tree_unroll_loops_completely): Likewise.
20906 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20907 Use an overload instead and export.
20908 (estimated_loop_iterations): Adjust.
20909 (max_loop_iterations): Likewise.
20910 (likely_max_loop_iterations): Likewise.
20911 (estimate_numbers_of_iterations): Take struct function as arg
20913 (loop_exits_before_overflow): Adjust.
20914 (free_numbers_of_iterations_estimates_loop): Use an overload.
20915 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20916 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20918 2017-06-19 Richard Biener <rguenther@suse.de>
20921 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20922 RANGE_EXPR conservatively.
20924 2017-06-16 Carl Love <cel@us.ibm.com>
20926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20927 definitions for vec_float, vec_float2, vec_floato,
20928 vec_floate built-ins.
20929 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20930 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20932 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20933 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20934 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20935 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20936 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20937 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20938 vec_floato): Add builtin defines.
20939 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20940 Update the built-in documentation file for the new built-in
20943 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20945 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20946 (mthumb): Mark as the negative of -marm.
20948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20950 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20952 (ARM Options, -mtune): Document that this accepts the same
20953 extension options as -mcpu.
20954 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20958 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20959 permitted extensions.
20961 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20963 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20964 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20965 (armv8-m.main): Add option +nodsp.
20966 * config/arm/arm-cpu-cdata.h: Regenerated.
20968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20970 * config/arm/t-fuchsia: New file.
20971 * config.gcc (arm*-*-fuchsia*): Use it.
20973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20975 * config/arm/t-symbian: Rewrite for new option infrastructure.
20977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20979 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20980 (MULTILIB_REQUIRED): Likewise.
20982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20984 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20985 (MULTILIB_RESUE): Likewise.
20986 (MULTILIB_MATCHES): Likewise.
20987 (MULTLIB_REQUIRED): Likewise.
20989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20991 * config/arm/t-rtems: Rewrite for new option framework.
20993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20995 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20996 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20997 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20998 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20999 * config/arm/t-multilib: ... here.
21000 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21001 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21002 armv7-a and armv8*-a when A-profile libraries have not been built.
21003 * config/arm/t-rmprofile: Rewrite.
21005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21007 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21008 with a backslash. Remove the backslash after substituting unescaped
21010 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21014 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21015 the driver to pass through the default setting of -mfloat-abi.
21016 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21018 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21019 (all_feat_combs): New rule.
21020 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21022 * config/arm/t-aprofile: Rewrite.
21024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21026 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21027 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21028 fpu is not specified by the user/command-line.
21029 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21030 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21031 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21032 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21033 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21034 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21038 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21039 * config/arm/t-arm-elf: Rewritten.
21041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21043 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21044 have some floating-point instructions.
21045 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21046 (TARGET_MAYBE_HARD_FLOAT): New macro.
21047 * config/arm/arm-builtins.c (arm_init_builtins): Use
21048 TARGET_MAYBE_HARD_FLOAT.
21049 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21051 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21053 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21054 (configargs.h): Include it.
21055 (arm_print_hint_for_fpu_option): New function.
21056 (arm_parse_fpu_option): New function.
21057 (candidate_extension): New class.
21058 (arm_canon_for_multilib): New function.
21059 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21060 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21061 (ARCH_CANONICAL_SPECS): New macro.
21062 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21066 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21067 are set after handling multilib fragments. Set target_cpu_default2
21070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21072 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21074 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21075 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21077 (check_arch): Likewise.
21078 * config/arm/arm.c (arm_configure_build_target): Handle
21079 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21080 options in the default.
21082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21084 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21085 when an option is an alias of another.
21086 * config/arm/parsecpu.awk (optalias): New parser token.
21087 (gen_comm_data): Mark non-alias options as such. Emit entries
21088 for extension aliases.
21089 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21090 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21091 (armv6kz, armv6zk, armv6t2): Likewise.
21092 (armv7): Make vfpv3-d16 an alias.
21093 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21095 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21096 Sort in canonical order.
21097 (armv8-a): Sort in canonical order.
21098 (armv8.1-a, armv8.2-a): Likewise.
21099 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21101 (cortex-a9): Sort in canonical order.
21102 * config/arm/arm.c (selftests.h): Include it.
21103 (arm_test_cpu_arch_data): New function.
21104 (arm_run_self_tests): New function.
21105 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21106 (targetm): Move declaration to the end of the file.
21107 * arm-cpu-cdata.h: Regenerated.
21109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21111 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21112 call to target_mode_check describing the type of option passed.
21113 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21114 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21115 arm_parse_cpu_option_name to match parameters against list of
21117 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21118 arm_arch_core_flags data structure.
21119 * config/arm/arm-cpu_cdata.h: Regenerated.
21121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21123 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21125 (arm_print_hint_for_cpu_option): Likewise.
21126 (arm_print_hint_for_arch_option): Likewise.
21127 (arm_parse_cpu_option_name): Likewise.
21128 (arm_parse_arch_option_name): Likewise.
21129 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21130 of entries in the all_fpus list.
21131 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21132 (arm_parse_cpu_option_name): Declare.
21133 (arm_parse_arch_option_name): Declare.
21134 (arm_parse_option_features): Declare.
21135 (arm_intialize_isa): Declare.
21136 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21138 (gen_comm_data): ... here. Make definitions non-static.
21139 * config/arm/arm-cpu-data.h: Regenerated.
21140 * config/arm/arm-cpu-cdata.h: Regenerated.
21142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21144 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21145 (cpu_arch_extension): New structure.
21146 (cpu_arch_option, arch_option, cpu_option): New structures.
21147 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21148 architecture types.
21149 (gen_data): Generate new format data tables.
21150 * config/arm/arm.c (cpu_tune): New structure.
21151 (cpu_option, processors): Delete.
21152 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21153 (arm_print_hint_for_cpu_option): ... this and ...
21154 (arm_print_hint_for_arch_option): ... this.
21155 (arm_parse_arch_cpu_name): Delete. Replace with ...
21156 (arm_parse_cpu_option_name): ... this and ...
21157 (arm_parse_arch_option_name): ... this.
21158 (arm_unrecognized_feature): Change type of target parameter to
21160 (arm_parse_arch_cpu_features): Delete. Replace with ...
21161 (arm_parse_option_features): ... this.
21162 (arm_configure_build_target): Rework to use new configuration data
21164 (arm_print_tune_info): Rework for new configuration data tables.
21165 * config/arm/arm-cpu-data.h: Regenerated.
21166 * config/arm/arm-cpu.h: Regenerated.
21168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21170 * Makefile.in (OBJS): Move sbitmap.o from here ...
21171 (OBJS-libcommon): ... to here.
21173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21175 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21176 (ISA_ALL_CRYPTO): New macro.
21177 (ISA_ALL_SIMD): New macro
21178 (ISA_ALL_FP): New macro.
21179 * config/arm/arm.c (fpu_bitlist): Update initializer.
21180 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21182 (arm9e): Add fpu. Add option for nofp
21183 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21184 (arm926ej-s, arm1026ej-s): Likewise.
21185 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21186 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21187 neon-fp16, neon-vfpv4, nofp and nosimd.
21188 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21189 (cortex-a8): Add fpu. Add option for nofp.
21190 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21191 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21192 (cortex-r4f): Add fpu.
21193 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21194 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21196 (cortex-r8): Likewise.
21197 (cortex-m4): Add fpu. Add option for nofp.
21198 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21199 (cortex-a17.cortex-a7): Likewise.
21200 (cortex-a32): Add fpu. Add options for crypto and nofp.
21201 (cortex-a35, cortex-a53): Likewise.
21202 (cortex-a57): Add fpu. Add option for crypto.
21203 (cortex-a72, cortex-a73): Likewise.
21204 (exynos-m1): Likewise.
21205 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21206 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21207 (cortex-m33): Add fpu. Add option for nofp.
21208 * config/arm/arm-cpu-cdata.h: Regenerated
21209 * config/arm/arm-cpu-data.h: Regenerated.
21211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21213 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21214 (armv5te, armv5tej): Likewise.
21215 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21216 (armv7): Add options fp and vfpv3-d16.
21217 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21218 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21220 (armv7ve): Likewise.
21221 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21222 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21223 (armv8-a): Add nocrypto option.
21224 (armv8.1-a, armv8.2-a): Likewise.
21225 (armv8-m.main): add options fp, fp.dp and nofp.
21227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21229 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21231 (armv8-a+crc): Delete.
21232 (armv8.1-a): Add options simd, crypto and nofp.
21233 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21234 (armv8.2-a+fp16): Delete.
21235 (armv8-m.main): Add option dsp.
21236 (armv8-m.main+dsp): Delete.
21237 (cortex-a8): Add fpu. Add nofp option.
21238 (cortex-a9): Add fpu. Add nofp and nosimd options.
21239 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21240 link to main cpu and architecture data structures.
21241 (gen_comm_data): Only put isa attributes from the main architecture
21243 (option): New statement for architecture and CPU entries.
21244 * arm.c (struct cpu_option): New structure.
21245 (struct processors): Add entry for options.
21246 (arm_unrecognized_feature): New function.
21247 (arm_parse_arch_cpu_name): Ignore any characters after the first
21249 (arm_parse_arch_cpu_feature): New function.
21250 (arm_configure_build_target): Separate out any CPU and architecture
21251 features and parse separately. Don't error out if -mfpu=auto is
21252 used with only an architecture string.
21253 (arm_print_asm_arch_directives): New function.
21254 (arm_file_start): Call it.
21255 * config/arm/arm-cpu-cdata.h: Regenerated.
21256 * config/arm/arm-cpu-data.h: Likewise.
21257 * config/arm/arm-tables.opt: Likewise.
21259 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21261 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21262 assembler when it is not -mfpu=auto.
21264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21266 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21267 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21268 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21269 (ASM_CPU_SPEC): Rewrite.
21270 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21271 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21272 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21273 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21274 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21275 copied string is NUL-terminated. Also strip any characters prefixed
21277 (arm_rewrite_selected_arch): New function.
21278 (arm_rewrite_march): New function.
21280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21282 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21283 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21284 (march, mcpu, mtune): Convert to string-based options.
21285 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21286 (arm_parse_arch_cpu_name): New function.
21287 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21288 identify selected architecture or CPU.
21289 (arm_option_save): New function.
21290 (TARGET_OPTION_SAVE): Redefine.
21291 (arm_option_restore): Restore string options.
21292 (arm_option_print): Print string options.
21294 2017-06-16 Martin Sebor <msebor@redhat.com>
21296 PR tree-optimization/80933
21297 PR tree-optimization/80934
21298 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21299 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21300 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21301 (gimple_fold_builtin): Call them.
21303 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21305 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21306 as unlikely; update profile.
21308 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21310 * predict.c (force_edge_cold): Handle declaring edges impossible
21313 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21315 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21317 (try_unroll_loop_completely): Fix reporting.
21319 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21321 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21323 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21326 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21327 if given a non-constant argument for an intrinsic which requires a
21330 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21332 * profile.c (compare_freqs): New function.
21333 (branch_prob): Sort edge list.
21334 (find_spanning_tree): Assume that the list is priority sorted.
21336 2017-06-16 Richard Biener <rguenther@suse.de>
21338 PR tree-optimization/81090
21339 * passes.def (pass_record_bounds): Remove.
21340 * tree-pass.h (make_pass_record_bounds): Likewise.
21341 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21342 make_pass_record_bounds): Likewise.
21343 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21344 not free niter estimates at the beginning but at the end.
21345 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21347 2017-06-16 Richard Biener <rguenther@suse.de>
21349 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21350 initializer to workaround ICE in host GCC 4.8.
21352 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21354 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21356 (clone_inlined_nodes): Update.
21358 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21360 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21361 prefetch settings, and enable prefetching by default at -O3.
21363 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21365 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21366 Set flag_prefetch_loop_arrays according to tuning data.
21368 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21370 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21371 New tune structure.
21372 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21373 [Unrelated to main purpose of the patch] Place the pointer field last
21374 to enable type checking errors when tune structure are wrongly merged.
21375 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21376 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21377 (thunderx2t99_prefetch_tune): New tune constants.
21378 (tune_params *_tunings): Update all tunings (no functional change).
21379 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21380 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21381 from tunings structures.
21383 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21386 * ubsan.c (instrument_null): Add T argument, use it instead
21387 of computing it based on IS_LHS.
21388 (instrument_object_size): Likewise.
21389 (pass_ubsan::execute): Adjust instrument_null and
21390 instrument_object_size callers to pass gimple_get_lhs or
21391 gimple_assign_rhs1 result to it. Use instrument_null instead of
21392 calling get_base_address and instrument_mem_ref. Handle
21393 aggregate call arguments for object-size sanitization.
21395 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21397 PR tree-optimization/81089
21398 * tree-vrp.c (is_masked_range_test): Validate operands of
21401 2017-06-15 Martin Sebor <msebor@redhat.com>
21404 * dumpfile.c (dump_register): Avoid calling memset to initialize
21405 a class with a default ctor.
21406 * gcc.c (struct compiler): Remove const qualification.
21407 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21408 * hash-table.h: Ditto.
21409 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21411 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21412 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21414 * params.h (struct param_info): Make struct members non-const.
21415 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21416 with default initialization.
21417 * vec.h (vec_copy_construct, vec_default_construct): New helper
21419 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21420 with vec_copy_construct.
21421 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21422 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21423 * doc/invoke.texi (-Wclass-memaccess): Document.
21425 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21427 * emit-rtl.h (is_leaf): Update comment about local
21428 register allocator.
21430 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21433 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21434 for a variable to have a section before checking if the section has a
21436 Set section to.persistent if persistent attribute is set.
21437 Warn if .persistent attribute is used on an automatic variable.
21439 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21441 PR rtl-optimization/80474
21442 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21444 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21446 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21449 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21451 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21452 (casesi): Emit bounds checking as RTL.
21453 (casesi_internal_mips16_<mode>): Remove bounds checking.
21455 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21457 * config/xtensa/xtensa.c (xtensa_option_override): Append
21458 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21459 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21460 xtensa_doloop_hooks): Define unconditionally.
21461 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21463 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21464 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21465 for it in xtensa_option_override.
21466 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21467 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21469 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21471 * doc/cppopts.texi: Document '-' special value to -MF.
21473 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21475 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21476 (cortex_a53_fconst): Likewise.
21477 (cortex_a53_fpmul): Likewise.
21478 (cortex_a53_f_load_64): Likewise.
21479 (cortex_a53_f_load_many): Likewise.
21480 (cortex_a53_advsimd_alu): Likewise.
21481 (cortex_a53_advsimd_alu_q): Likewise.
21482 (cortex_a53_advsimd_mul): Likewise.
21483 (cortex_a53_advsimd_mul_q): Likewise.
21484 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21485 Add missing fmul, r2f_cvt and fconst cases.
21487 2017-06-14 Richard Biener <rguenther@suse.de>
21489 PR middle-end/81088
21490 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21492 (fold_binary_loc): When associating do not treat pre-existing
21493 TREE_OVERFLOW on literal constants as a reason to allow
21494 TREE_OVERFLOW on associated literal constants.
21496 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21498 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21499 (MASK_FEATURES): New macro.
21500 * config/sparc/sparc.c (sparc_option_override): Remove the special
21501 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21503 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21505 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21506 a division of 0 if non-call exceptions are enabled.
21508 2017-06-14 Andrew Pinski <apinski@cavium.com>
21509 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21512 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21513 Improve vector initialization code gen for only variable case.
21515 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21517 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21519 2017-06-14 Richard Biener <rguenther@suse.de>
21521 PR tree-optimization/81083
21522 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21525 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21527 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21528 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21529 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21530 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21531 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21532 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21534 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21536 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21537 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21539 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21541 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21543 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21545 * config/rs6000/t-rtems: Don't handle SPE.
21547 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21549 * config/rs6000/t-linux: Don't handle SPE.
21551 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21553 * config/rs6000/eabispe.h: Delete file.
21555 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21557 * config/rs6000/t-spe: Delete file.
21559 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21561 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21562 (rs6000_legitimate_offset_address_p): Return false for anything in
21563 V2SImode or V2SFmode.
21565 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21567 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21568 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21569 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21571 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21572 (rs6000_legitimate_offset_address_p): Ditto.
21573 (rs6000_emit_move): Ditto.
21574 (rs6000_init_builtins): Remove V4HI_type_node.
21576 2017-06-13 Martin Liska <mliska@suse.cz>
21579 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21580 (gate_asan): Likewise.
21581 * asan.h (asan_no_sanitize_address_p): Remove the function.
21582 (sanitize_flags_p): New function.
21583 * builtins.def: Fix coding style.
21584 * common.opt: Use renamed enum value.
21585 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21586 * doc/extend.texi: Document no_sanitize attribute.
21587 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21588 to SANITIZE_UNDEFINED_NONDEFAULT.
21589 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21590 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21591 Use sanitize_flags_p.
21592 * gimplify.c (gimplify_function_tree): Likewise.
21593 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21594 * opts.c (parse_no_sanitize_attribute): New function.
21595 (common_handle_option): Use renamed enum value.
21596 * opts.h (parse_no_sanitize_attribute): Declare.
21597 * tree.c (sanitize_flags_p): New function.
21598 * tree.h: Declared here.
21599 * tsan.c: Use sanitize_flags_p.
21600 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21601 (instrument_mem_ref): Likewise.
21602 (instrument_bool_enum_load): Likewise.
21603 (do_ubsan_in_current_function): Remove the function.
21604 (pass_ubsan::execute): Use sanitize_flags_p.
21605 * ubsan.h: Remove do_ubsan_in_current_function
21606 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21607 (dump_function_to_file): Use it here.
21609 2017-06-13 Martin Jambor <mjambor@suse.cz>
21611 PR tree-optimization/80803
21612 PR tree-optimization/81063
21613 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21614 (propagate_subaccesses_across_link): Enqueue subtree whenever
21615 necessary instead of relying on the caller.
21617 2017-06-13 Martin Jambor <mjambor@suse.cz>
21619 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21620 that have a first_link.
21621 (sort_and_splice_var_accesses): Do not check first_link before
21623 (subtree_mark_written_and_enqueue): Likewise.
21624 (propagate_all_subaccesses): Likewise and do not stop at first
21625 parent with a first_link.
21627 2017-06-13 Martin Jambor <mjambor@suse.cz>
21629 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21632 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21634 * match.pd: New pattern.
21636 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21638 * tree-vrp.c (is_masked_range_test): New function.
21639 (register_edge_assert_for): Determine ranges for
21642 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21644 PR tree-optimization/67328
21645 * fold-const.c (maskable_range_p): New function.
21646 (build_range_check): Generate bittests if possible.
21648 2017-06-13 Martin Liska <mliska@suse.cz>
21650 * gimple-pretty-print.c (dump_probability): Add new argument.
21651 (dump_edge_probability): Dump both probability and count.
21652 (dump_gimple_label): Likewise.
21653 (dump_gimple_bb_header): Likewise.
21655 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21658 * config/avr/avr-devices.c: Fix indentation.
21659 * config/avr/gen-avr-mmcu-specs.c: Dito.
21661 2017-06-13 Richard Biener <rguenther@suse.de>
21663 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21664 instead get vector type from stmt_info.
21665 (vectorizable_reduction): Adjust. Remove dead code.
21667 2017-06-13 Richard Biener <rguenther@suse.de>
21669 PR middle-end/81065
21670 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21671 case of C * (x * C2 + C3).
21672 (fold_addr_of_array_ref_difference): Properly fold index difference.
21674 2017-06-12 David S. Miller <davem@davemloft.net>
21677 * config/sparc/sparc.md (return expander): Emit frame blockage if
21678 function uses alloca.
21680 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21682 * combine.c (make_field_assignment): Check len rather than the mode
21683 precision when calling force_to_mode.
21685 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21687 Support multilibs and devices that see flash in RAM address range.
21690 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21691 (avr_mcu_t) <flash_pm_offset>: New field.
21692 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21693 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21694 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21695 (AVR_TINY_PM_OFFSET): Remove macro.
21696 * config/avr/avr.opt (-mshort-calls): New option.
21697 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21698 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21699 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21700 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21701 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21702 instead of avr_arch->have_jmp_call.
21703 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21704 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21705 avr_arch->flash_pm_offset to define.
21706 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21707 new field flash_pm_offset. Add entry for avrxmega3.
21708 (avr_texinfo): Add entry for avrxmega3.
21709 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21710 attiny212, attiny214,
21711 attiny412, attiny414, attiny416, attiny417,
21712 attiny814, attiny816, attiny817,
21713 attiny1614, attiny1616, attiny1617,
21714 attiny3214, attiny3216, attiny3217.
21715 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21716 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21717 (avr_print_operand_address) [AVR_TINY]: Same.
21718 (avr_asm_init_sections) <readonly_data_section>: Only patch
21719 callback if avr_arch->flash_pm_offset = 0.
21720 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21721 for rodata if avr_arch->flash_pm_offset != 0.
21722 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21723 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21724 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21725 (m_options): Append opt_rcall.
21726 (m_dirnames): Append dir_rcall.
21727 * config/avr/t-multilib: Regenerate.
21729 * configure.ac [target=avr]: Check whether avrxmega3 default
21730 linker description file works as needed.
21731 * configure: Regenerate.
21732 * doc/avr-mmcu.texi: Regenerate.
21733 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21734 <__AVR_ARCH__>: Document avrxmega3 and 103.
21735 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21736 <__AVR_SHORT_CALLS__>: Document it.
21737 <__AVR_PM_BASE_ADDRESS__>: Document it.
21738 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21739 (AVR Variable Attributes) <progmem>: Document this is
21740 not needed for avrxmega3.
21741 (AVR Named Address Spaces) <__flash>: Dito.
21743 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21745 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21747 2017-06-12 Doug Rupp <rupp@adacore.com>
21749 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21750 Append vxworks-stdint.h to the tm_file list.
21751 * config/vxworks-stdint.h: New file.
21753 2017-06-12 Martin Liska <mliska@suse.cz>
21755 PR tree-optimization/81041
21756 * tree-profile.c (gimple_gen_ic_func_profiler):
21757 Create an extra BB in profile-generate
21758 (gimple_gen_time_profiler): Likewise.
21760 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21762 PR tree-optimization/81003
21763 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21764 (update_range_test): Use it instead of force_gimple_operand_gsi.
21766 2017-06-12 Richard Biener <rguenther@suse.de>
21768 PR tree-optimization/81053
21769 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21770 with backedge value not defined in loop. Simplify def stmt
21773 2017-06-11 Tom de Vries <tom@codesourcery.com>
21776 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21778 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21779 nvptx_cannot_force_const_mem.
21781 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21783 * opts.c (finish_options): Move test for flag_split_stack after
21784 it has been initialized.
21786 2017-06-11 Jason Merrill <jason@redhat.com>
21788 * tree.h (id_equal): New.
21789 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21790 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21791 instead of strcmp of IDENTIFIER_POINTER.
21793 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21795 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21796 (mark_all_inlined_calls_cdtor): Fix formating.
21797 (inline_transform): Rescale profile before inlining.
21799 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21801 * cgraph.h (cgraph_edge::clone): Update prototype.
21802 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21803 (cgraph_node::create_clone): Update.
21804 (cgraph_node::create_version_clone): Update.
21805 * tree-inline.c (copy_bb): Update.
21806 (expand_call_inline): Update.
21808 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21810 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21811 factored out from ...
21812 (rs6000_emit_prologue): ... here.
21814 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21816 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21817 factored out from ...
21818 (rs6000_emit_prologue): ... here.
21820 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21822 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21824 (handle_missing_profiles): Fix computation of tp_first_run.
21825 (counts_to_freqs): Do not touch freqs when count is 0.
21827 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21829 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21832 2017-06-10 Tom de Vries <tom@codesourcery.com>
21834 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21835 attributes): Document signal effective target.
21837 2017-06-10 Tom de Vries <tom@codesourcery.com>
21839 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21840 Document effective target stack_size.
21842 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21844 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21845 to the edit_context if they can be auto-applied.
21847 2017-06-9 Ian Lance Taylor <iant@golang.org>
21849 * opts.c (finish_options): If -fsplit-stack, disable implicit
21850 -forder-blocks-and-partition.
21851 * doc/invoke.texi (Optimize Options): Document that when using
21852 -fsplit-stack -forder-blocks-and-partition is not implicitly
21855 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21857 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21858 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21859 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21860 * builtins.def (abort, trap, unreachable): Declare cold.
21861 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21862 * tree-core.h (ECF_COLD): New.
21863 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21864 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21866 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21868 * predict.c (unlikely_executed_stmt_p): Cleanup.
21870 2017-06-09 Richard Biener <rguenther@suse.de>
21872 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21873 model if the ref is always written to.
21875 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21877 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21879 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21881 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21884 2017-06-09 Tom de Vries <tom@codesourcery.com>
21887 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21888 "target cannot support label values" when encountering LABEL_REF.
21890 2017-06-09 Martin Liska <mliska@suse.cz>
21892 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21893 (gimple_gen_ic_func_profiler): Emit direct comparison
21894 of __gcov_indirect_call_callee with NULL.
21895 (gimple_gen_time_profiler): Change probability from
21896 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21898 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21900 * profile.c (edge_gcov_counts): Turn to pointer.
21901 (compute_branch_probabilities, compute_branch_probabilities): Update.
21902 (branch_prob): Do not clear edge_gcov_count.
21903 * profile.h (edge_gcov_counts): Turn to pointer.
21904 (edge_gcov_count): Update.
21906 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21908 * gimple.h (gimple_check_failed): Mark cold.
21910 2017-06-09 Richard Biener <rguenther@suse.de>
21912 PR tree-optimization/66623
21913 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21914 refactor check_reduction into two parts, properly computing
21915 whether we have to check reduction validity for outer loop
21918 2017-06-09 Richard Biener <rguenther@suse.de>
21920 PR tree-optimization/79483
21921 * graphite-scop-detection.c (order): New global.
21922 (get_order): Compute bb to order mapping that satisfies code
21923 generation constraints.
21924 (cmp_pbbs): New helper.
21925 (build_scops): Start domwalk at entry block, sort generated
21928 2017-06-09 Richard Biener <rguenther@suse.de>
21930 PR middle-end/81007
21931 * ipa-polymorphic-call.c
21932 (ipa_polymorphic_call_context::restrict_to_inner_class):
21933 Skip FIELD_DECLs with error_mark_node type.
21934 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21937 2017-06-09 Martin Liska <mliska@suse.cz>
21939 * predict.c (struct branch_predictor): New struct.
21940 (test_prediction_value_range): New test.
21941 (predict_c_tests): New function.
21942 * selftest-run-tests.c (selftest::run_tests): Run the function.
21943 * selftest.h: Declare new tests.
21945 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21948 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21949 gen_add3_insn did not fail.
21950 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21951 r0, construct that number in a temporary reg and add that reg to r0.
21952 If asked to put the result in r0 as well, fail.
21954 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21956 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21957 for early expansion of vec_eqv.
21959 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21961 PR middle-end/81005
21962 * ubsan.c (instrument_null): Avoid pointless code temporary.
21963 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21965 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21969 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21972 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21975 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21977 * predict.c (unlikely_executed_edge_p): Move ahead.
21978 (probably_never_executed_edge_p): Use it.
21980 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21982 PR middle-end/79988
21983 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21984 gimple_call_builtin_p call.
21986 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21988 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21989 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21990 rtl_check_failed_type2, rtl_check_failed_code1,
21991 rtl_check_failed_code2, rtl_check_failed_code_mode,
21992 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21993 rtvec_check_failed_bounds, rtl_check_failed_flag,
21994 _fatal_insn_not_found, _fatal_insn): Likewise.
21995 * tree.h (tree_contains_struct_check_failed,
21996 tree_check_failed, tree_not_check_failed,
21997 tree_class_check_failed, tree_range_check_failed,
21998 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21999 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22000 tree_operand_check_failed, omp_clause_check_failed,
22001 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22004 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22006 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22007 flag_branch_probabilities.
22008 * ipa-inline.c (edge_badness): Likewise.
22009 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22010 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22011 * predict.c (maybe_hot_frequency_p): Likewise.
22012 (probably_never_executed): Likewise.
22013 * sched-ebb.c (schedule_ebbs): Likewise.
22014 * sched-rgn.c (find_single_block_region): Likewise.
22015 * tracer.c (tail_duplicate): Likewise.
22017 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22019 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22020 longer requires x_flag_profile_use.
22022 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22024 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22025 instead of flag_reorder_blocks_and_partition.
22026 * dbxout.c (dbxout_function_end): Likewise.
22027 * dwarf2out.c (gen_subprogram_die): Likewise.
22028 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22029 * hw-doloop.c (reorg_loops): Likewise.
22030 * varasm.c (assemble_start_function,
22031 assemble_end_function): Likewise.
22032 (decide_function_section): Do not check for
22033 flag_reorder_blocks_and_partition.
22035 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22037 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22039 (chkp_get_hard_register_fake_addr_expr): Ditto.
22040 (chkp_build_addr_expr): Add check for hard reg case.
22041 (chkp_parse_array_and_component_ref): Ditto.
22042 (chkp_find_bounds_1): Ditto.
22043 (chkp_process_stmt): Don't generate bounds store for
22046 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22048 * predict.c (maybe_hot_bb_p): Do not check profile status.
22049 (maybe_hot_edge_p): Likewise.
22050 (probably_never_executed): Check for zero counts even if profile
22052 (unlikely_executed_edge_p): New function.
22053 (unlikely_executed_stmt_p): New function.
22054 (unlikely_executed_bb_p): New function.
22055 (set_even_probabilities): Use unlikely predicates.
22056 (combine_predictions_for_bb): Likewise.
22057 (predict_paths_for_bb): Likewise.
22058 (predict_paths_leading_to_edge): Likewise.
22059 (determine_unlikely_bbs): New function.
22060 (estimate_bb_frequencies): Use it.
22061 (compute_function_frequency): Use zero counts even if profile is
22063 * profile-count.h: Fix typo.
22065 2017-08-08 Julia Koval <julia.koval@intel.com>
22067 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22068 _mm512_mask_cvtsepi16_storeu_epi8,
22069 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22070 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22071 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22072 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22073 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22074 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22075 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22076 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22077 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22078 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22079 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22080 __builtin_ia32_pmovuswb256mem_mask,
22081 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22082 __builtin_ia32_pmovwb512mem_mask): New builtins.
22084 2017-08-08 Julia Koval <julia.koval@intel.com>
22086 PR target/73350,80862
22087 * config/i386/subst.md (round): Fix round pattern.
22088 * config/i386/i386.c (ix86_erase_embedded_rounding):
22089 Fix erasing rounding for the fixed pattern.
22091 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22093 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22095 2017-06-08 Martin Liska <mliska@suse.cz>
22097 PR gcov-profile/80911
22098 * gcov.c (block_info::block_info): New constructor.
22100 2017-06-07 Carl Love <cel@us.ibm.com>
22102 * config/rs6000/rs6000-c: The return type of the following
22103 built-in functions was implemented as int not long long. Fix sign
22104 of return value for the unsigned version of vec_mulo and vec_mule.
22105 vector unsigned long long vec_bperm (vector unsigned long long,
22106 vector unsigned char)
22107 vector signed long long vec_mule (vector signed int,
22109 vector unsigned long long vec_mule (vector unsigned int,
22110 vector unsigned int)
22111 vector signed long long vec_mulo (vector signed int,
22113 vector unsigned long long vec_mulo (vector unsigned int,
22114 vector unsigned int)
22115 * doc/extend.texi: Fix the documentation for the built-in
22118 2017-06-07 Carl Love <cel@us.ibm.com>
22121 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22124 2017-06-07 Carl Love <cel@us.ibm.com>
22126 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22127 support, Generate doublehv for signed int/float for BE case only.
22129 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22131 * doc/invoke.texi (mcx16): Rewrite.
22133 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22135 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22136 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22137 *mov<mode>_softfloat, and an anonymous splitter): Use
22138 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22140 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22142 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22144 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22145 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22146 (rs6000_debug_reg_global): Adjust.
22147 (rs6000_init_hard_regno_mode_ok): Adjust.
22148 (rs6000_dbx_register_number): Adjust.
22149 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22150 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22151 Remove SPE_ACC and SPEFSCR.
22152 (REG_ALLOC_ORDER): Ditto.
22153 (FRAME_POINTER_REGNUM): Change to 111.
22154 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22155 (REG_CLASS_NAMES): Ditto.
22156 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22157 (REGISTER_NAMES): Ditto.
22158 (ADDITIONAL_REG_NAMES): Ditto.
22159 (rs6000_reg_names): Ditto.
22160 * config/rs6000/rs6000.md: Renumber some register number
22163 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22165 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22167 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22168 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22170 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22171 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22172 Delete the SPE high registers.
22173 (REG_ALLOC_ORDER): Ditto.
22174 (enum reg_class): Remove SPE_HIGH_REGS.
22175 (REG_CLASS_NAMES): Ditto.
22176 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22177 (REGISTER_NAMES): Ditto.
22178 (rs6000_reg_names): Ditto.
22179 * doc/tm.texi.in: Remove SPE as example.
22180 * doc/tm.texi: Regenerate.
22182 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22184 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22185 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22186 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22187 * config/rs6000/rs6000.md (type): Remove "brinc".
22189 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22191 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22192 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22193 * config/rs6000/linuxspe.h: Delete file.
22194 * config/rs6000/rs6000.md: Don't include spe.md.
22195 * config/rs6000/spe.h: Delete file.
22196 * config/rs6000/spe.md: Delete file.
22197 * config/rs6000/t-rs6000: Remove spe.md.
22199 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22201 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22202 (reg_or_none500mem_operand): Delete.
22203 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22204 instead of reg_or_none500mem_operand.
22206 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22208 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22209 handling of SPE flags.
22210 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22212 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22214 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22216 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22217 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22218 paired_divv2sf3): Similar.
22219 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22220 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22221 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22223 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22224 Rename the paired_* instruction patterns.
22225 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22227 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22228 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22229 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22230 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22231 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22232 PAIRED_VECTOR_MODE.
22233 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22234 (spe_func_has_64bit_regs_p): Delete.
22235 (spe_expand_predicate_builtin): Delete.
22236 (spe_expand_evsel_builtin): Delete.
22237 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22238 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22239 (invalid_e500_subreg): Delete.
22240 (rs6000_legitimize_address): Always force_reg op2 as well, for
22241 paired single memory accesses.
22242 (rs6000_member_type_forces_blk): Delete.
22243 (rs6000_spe_function_arg): Delete.
22244 (rs6000_expand_unop_builtin): Delete SPE handling.
22245 (rs6000_expand_binop_builtin): Ditto.
22246 (spe_expand_stv_builtin): Delete.
22247 (bdesc_2arg_spe): Delete.
22248 (spe_expand_builtin): Delete.
22249 (spe_expand_predicate_builtin): Delete.
22250 (spe_expand_evsel_builtin): Delete.
22251 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22252 (spe_init_builtins): Delete.
22253 (spe_func_has_64bit_regs_p): Delete.
22254 (savres_routine_name): Delete "info" parameter. Adjust callers.
22255 (rs6000_emit_stack_reset): Ditto.
22256 (rs6000_dwarf_register_span): Delete.
22257 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22258 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22259 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22261 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22263 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22264 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22265 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22266 mulv2sf3, divv2sf3): Delete expanders.
22268 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22270 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22274 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22275 * config/rs6000/rs6000.c: Ditto.
22277 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22279 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22280 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22281 comparison_operator.
22283 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22285 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22286 * config/rs6000/rs6000.opt: Ditto.
22287 * config/rs6000/t-rtems: Ditto.
22289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22291 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22292 TARGET_E500_SINGLE by 0, simplify.
22293 * config/rs6000/rs6000.c: Ditto.
22294 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22295 (spe_build_register_parallel): Delete.
22296 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22297 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22298 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22299 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22300 (E500_CONVERT): Delete.
22301 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22303 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22305 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22306 * config/rs6000/dfp.md: Ditto.
22307 (negdd2, *negdd2_fpr): Merge.
22308 (absdd2, *absdd2_fpr): Merge.
22309 (negtd2, *negtd2_fpr): Merge.
22310 (abstd2, *abstd2_fpr): Merge.
22311 * config/rs6000/e500.h: Delete file.
22312 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22313 TARGET_FPRS by 1 and simplify.
22314 * config/rs6000/rs6000-c.c: Ditto.
22315 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22316 TARGET_DF_SPE by 0.
22317 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22319 * config/rs6000/rs6000.md: Ditto.
22320 (floatdidf2, *floatdidf2_fpr): Merge.
22321 (move_from_CR_gt_bit): Delete.
22322 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22323 (E500_CR_IOR_COMPARE): Delete.
22324 (All patterns that require !TARGET_FPRS): Delete.
22325 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22327 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22329 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22331 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22333 * graphds.c (add_edge): Intitialize edge's attached data.
22334 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22335 pointer parameter. Call pointed function on each edge during
22336 graph traversing. Skip traversing the edge when the function
22338 (graphds_dfs, graphds_scc): Ditto.
22339 (for_each_edge): New parameter. Pass the new parameter to callback
22341 * graphds.h (skip_edge_callback): New function pointer type.
22342 (graphds_dfs, graphds_scc): New function pointer parameter.
22343 (graphds_edge_callback, for_each_edge): New parameter.
22345 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22347 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22348 out code checking if runtime alias check is possible to below ...
22349 Call the new function.
22350 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22351 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22353 2017-06-07 Marek Polacek <polacek@redhat.com>
22356 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22357 TYPE_OVERFLOW_WRAPS check.
22359 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22361 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22362 if versioning is required.
22363 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22364 peeling with the check for versioning.
22366 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22368 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22369 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22370 Set true to new parameter if new ssa variable is defined.
22371 (vect_gen_vector_loop_niters): Refactor. Set range information
22372 for the new vector loop bound variable.
22373 (vect_do_peeling): Ditto.
22375 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22377 * tree-affine.c (ssa.h): Include header file.
22378 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22379 has wrapping overflow behavior.
22381 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22383 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22385 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22387 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22388 (tree_to_aff_combination): ... here.
22390 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22392 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22393 reg_pressure model function.
22394 (ivopts_global_cost_for_size): Delete.
22395 (determine_set_costs, iv_ca_recount_cost): Call new model function
22396 ivopts_estimate_reg_pressure.
22398 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22400 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22401 expensive than udiv. Remove floating point cases from mod.
22403 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22405 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22406 Increase idiv cost.
22408 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22410 * config/aarch64/aarch64.md
22411 (copysignsf3): Fix mask generation.
22413 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22415 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22417 (class dump_manager): Add register_dumps method.
22418 * dumpfile.c: Include langhooks.h.
22419 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22420 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22421 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22422 (dump_manager::dump_register): Start with 512 entries instead of 32.
22423 (dump_manager::register_dumps): New method.
22424 * toplev.c (general_init): Instead of invoking register_dumps
22425 langhook, invoke register_dumps method on the dump manager.
22426 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22429 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22431 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22432 Say that # requires an associated define_split to exist, and that
22433 the define_split must be suitable for use after register allocation.
22435 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22437 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22438 (compute_outgoing_frequencies): Also initialize zero counts.
22439 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22440 around loops; preserve more of profile when nothing changes.
22442 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22444 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22446 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22447 * config/arm/arm-cpu-cdata.h: Regenerate.
22448 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22449 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22450 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22451 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22452 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22454 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22456 * config/arm/t-rmprofile: Likewise.
22457 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22459 2017-06-06 David S. Miller <davem@davemloft.net>
22462 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22463 blockage if function uses alloca.
22465 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22467 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22468 New "uid" fields to hold pretty-print IDs of group and ref.
22469 Memory references are now identified as <group_id>:<ref_id>
22470 instead of using [random] addresses.
22471 (dump_mem_details): Simplify, no functional change.
22472 (dump_mem_ref): Simplify and make output more concise.
22473 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22474 (find_or_create_group): Initialize group uid.
22475 (record_ref): Initialize ref uid. Improve debug output.
22476 (prune_group_by_reuse, should_issue_prefetch_p,)
22477 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22478 (mark_nontemporal_store, determine_loop_nest_reuse):
22479 Improve debug output.
22481 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22483 * dbgcnt.def (prefetch): New debug counter.
22484 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22485 (schedule_prefetches): Stop issueing prefetches if debug counter
22488 2017-06-06 Tom de Vries <tom@codesourcery.com>
22490 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22491 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22493 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22495 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22496 Use aarch64_reg_or_zero predicate for operand 4.
22497 (aarch64_compare_and_swap<mode> define_insn_and_split):
22498 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22499 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22501 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22503 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22504 (arm_compute_save_core_reg_mask): This.
22505 (thumb1_compute_save_reg_mask): Rename into ...
22506 (thumb1_compute_save_core_reg_mask): This.
22507 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22508 (arm_compute_frame_layout): Likewise.
22510 2017-06-06 Richard Biener <rguenther@suse.de>
22512 PR tree-optimization/80974
22513 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22514 keep or clear leaders SSA info.
22516 2017-06-06 Tom de Vries <tom@codesourcery.com>
22518 * config/nvptx/nvptx.c (split_mode_p): New function.
22519 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22521 2017-06-06 Tom de Vries <tom@codesourcery.com>
22523 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22525 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22528 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22531 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22533 * shrink-wrap.c (handle_simple_exit): Update profile.
22534 (try_shrink_wrapping): Upate profile.
22536 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22538 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22539 (tree_guess_outgoing_edge_probabilities): New.
22540 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22541 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22543 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22545 * ipa-split.c (split_function): Initialize return bb profile.
22547 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22549 * profile.c (compute_branch_probabilities): Also initialize
22550 EXIT_BLOCK profile.
22552 2017-06-06 Richard Biener <rguenther@suse.de>
22554 PR tree-optimization/80928
22555 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22556 (vect_analyze_loop_operations): Properly guard analysis for
22558 (vect_transform_loop): Likewise.
22559 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22560 (vect_model_induction_cost): Do not cost for pure SLP.
22561 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22562 of induction in inner loop vectorization.
22563 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22564 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22565 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22567 (vect_analyze_slp_cost_1): Cost induction.
22568 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22569 (vect_get_slp_vect_defs): Likewise.
22570 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22571 (vect_transform_stmt): Handle SLP reductions.
22572 * tree-vectorizer.h (vectorizable_induction): Adjust.
22574 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22576 * config/rs6000/rs6000.c (make_resolver_func): Update
22577 init_lowered_empty_function call.
22579 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22581 * doc/invoke.texi: Document the -fprofile-abs-path option.
22582 * common.opt (fprofile-abs-path): New option.
22583 * gcov-io.h (gcov_write_filename): Declare.
22584 * gcov-io.c (gcov_write_filename): New function.
22585 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22586 * profile.c (output_location): Likewise.
22588 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22590 * shring-wrap.c: Revert accidental commit.
22592 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22594 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22596 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22598 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22600 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22601 profile in callgraph edge.
22602 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22603 (apply_scale): Likewise.
22604 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22605 Also scale profile when inlining function with zero profile.
22606 (initialize_cfun): Update exit block profile even when it is zero.
22607 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22608 when profile is read.
22610 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22612 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22613 (CLONE_*): New constants to define the processors we can generate
22614 code for with the target_clone attribute.
22615 (rs6000_clone_map): New array to identify which clone processors
22616 the current program is running on.
22617 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22618 target_clone attribute.
22619 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22620 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22621 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22622 (cpu_expand_builtin): Add support for target_clone attribute.
22623 (rs6000_valid_attribute_p): Allow "default" attribute.
22624 (get_decl_name): New debug function to simplify printing the
22625 current function name in debugging statements.
22626 (rs6000_clone_priority): New functions to support the target_clone
22627 attribute, and be able to generate code to switch between ISA 2.05
22628 through ISA 3.0 (power6 through power9).
22629 (rs6000_compare_version_priority): Likewise.
22630 (rs6000_get_function_versions_dispatcher): Likewise.
22631 (make_resolver_func): Likewise.
22632 (add_condition_to_bb): Likewise.
22633 (dispatch_function_versions): Likewise.
22634 (rs6000_generate_version_dispatcher_body): Likewise.
22635 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22636 (fusion_gpr_load_p): Fix a spacing issue.
22637 * doc/extend.texi (Common Function Attributes): Document that the
22638 PowerPC supports the target_clone attribute.
22640 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22642 * config/arm/arm.h: explain F symbol found in description of ARM
22643 register allocation in its legend.
22645 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22647 * config/mips/frame-header-opt.c: Include profile-count.h.
22648 * config/riscv/riscv.c: Include profile-count.h
22650 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22652 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22654 (sm_set_flag_if_changed): Add bbs field.
22655 (execute_sm_if_changed_flag_set): Pass BBS.
22656 (execute_sm): Update.
22658 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22660 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22663 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22665 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22666 (peephole2): New peephole2 to emit the above.
22667 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22669 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22671 * config/aarch64/aarch64.c (define_peephole2 above
22672 *sub_<shift>_<mode>): New peephole.
22674 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22676 * config/i386/i386.c (make_resolver_func): Update.
22677 * Makefile.in: Add profile-count.h and profile-count.o
22678 * auto-profile.c (afdo_indirect_call): Update to new API.
22679 (afdo_set_bb_count): Update.
22680 (afdo_propagate_edge): Update.
22681 (afdo_propagate_circuit): Update.
22682 (afdo_calculate_branch_prob): Update.
22683 (afdo_annotate_cfg): Update.
22684 * basic-block.h: Include profile-count.h
22685 (struct edge_def): Turn count to profile_count.
22686 (struct basic_block_def): Likewie.
22687 (REG_BR_PROB_BASE): Move to profile-count.h
22688 (RDIV): Move to profile-count.h
22689 * bb-reorder.c (max_entry_count): Turn to profile_count.
22690 (find_traces): Update.
22691 (rotate_loop):Update.
22692 (connect_traces):Update.
22693 (sanitize_hot_paths):Update.
22694 * bt-load.c (migrate_btr_defs): Update.
22695 * cfg.c (RDIV): Remove.
22696 (init_flow): Use alloc_block.
22697 (alloc_block): Uninitialize count.
22698 (unchecked_make_edge): Uninitialize count.
22699 (check_bb_profile): Update.
22700 (dump_edge_info): Update.
22701 (dump_bb_info): Update.
22702 (update_bb_profile_for_threading): Update.
22703 (scale_bbs_frequencies_int): Update.
22704 (scale_bbs_frequencies_gcov_type): Update.
22705 (scale_bbs_frequencies_profile_count): New.
22706 * cfg.h (update_bb_profile_for_threading): Update.
22707 (scale_bbs_frequencies_profile_count): Declare.
22708 * cfgbuild.c (compute_outgoing_frequencies): Update.
22709 (find_many_sub_basic_blocks): Update.
22710 * cfgcleanup.c (try_forward_edges): Update.
22711 (try_crossjump_to_edge): Update.
22712 * cfgexpand.c (expand_gimple_tailcall): Update.
22713 (construct_exit_block): Update.
22714 * cfghooks.c (verify_flow_info): Update.
22715 (dump_bb_for_graph): Update.
22716 (split_edge): Update.
22717 (make_forwarder_block): Update.
22718 (duplicate_block): Update.
22719 (account_profile_record): Update.
22720 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22721 (get_estimated_loop_iterations): Update.
22722 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22723 (single_likely_exit): Update.
22724 * cfgloopmanip.c (scale_loop_profile): Update.
22726 (set_zero_probability): Update.
22727 (lv_adjust_loop_entry_edge): Update.
22728 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22729 (purge_dead_edges): Update.
22730 (rtl_account_profile_record): Update.
22731 * cgraph.c (cgraph_node::create): Uninitialize count.
22732 (symbol_table::create_edge): Uninitialize count.
22733 (cgraph_update_edges_for_call_stmt_node): Update.
22734 (cgraph_edge::dump_edge_flags): Update.
22735 (cgraph_node::dump): Update.
22736 (cgraph_edge::maybe_hot_p): Update.
22737 * cgraph.h: Include profile-count.h
22738 (create_clone), create_edge, create_indirect_edge): Update.
22739 (cgraph_node): Turn count to profile_count.
22740 (cgraph_edge0: Likewise.
22741 (make_speculative, clone): Update.
22742 (create_edge): Update.
22743 (init_lowered_empty_function): Update.
22744 * cgraphclones.c (cgraph_edge::clone): Update.
22745 (duplicate_thunk_for_node): Update.
22746 (cgraph_node::create_clone): Update.
22747 * cgraphunit.c (cgraph_node::analyze): Update.
22748 (cgraph_node::expand_thunk): Update.
22749 * final.c (dump_basic_block_info): Update.
22750 * gimple-streamer-in.c (input_bb): Update.
22751 * gimple-streamer-out.c (output_bb): Update.
22752 * graphite.c (print_global_statistics): Update.
22753 (print_graphite_scop_statistics): Update.
22754 * hsa-brig.c: Include basic-block.h.
22755 * hsa-dump.c: Include basic-block.h.
22756 * hsa-gen.c (T sum_slice): Update.
22757 (convert_switch_statements):Update.
22758 * hsa-regalloc.c: Include basic-block.h.
22759 * ipa-chkp.c (chkp_produce_thunks): Update.
22760 * ipa-cp.c (struct caller_statistics): Update.
22761 (init_caller_stats): Update.
22762 (gather_caller_stats): Update.
22763 (ipcp_cloning_candidate_p): Update.
22764 (good_cloning_opportunity_p): Update.
22765 (get_info_about_necessary_edges): Update.
22766 (dump_profile_updates): Update.
22767 (update_profiling_info): Update.
22768 (update_specialized_profile): Update.
22769 (perhaps_add_new_callers): Update.
22770 (decide_about_value): Update.
22771 (ipa_cp_c_finalize): Update.
22772 * ipa-devirt.c (struct odr_type_warn_count): Update.
22773 (struct decl_warn_count): Update.
22774 (struct final_warning_record): Update.
22775 (possible_polymorphic_call_targets): Update.
22776 (ipa_devirt): Update.
22777 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22778 * ipa-icf.c (sem_function::merge): Update.
22779 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22780 * ipa-inline.c (compute_uninlined_call_time): Update.
22781 (compute_inlined_call_time): Update.
22782 (want_inline_small_function_p): Update.
22783 (want_inline_self_recursive_call_p): Update.
22784 (edge_badness): Update.
22785 (lookup_recursive_calls): Update.
22786 (recursive_inlining): Update.
22787 (inline_small_functions): Update.
22788 (dump_overall_stats): Update.
22789 (dump_inline_stats): Update.
22790 * ipa-profile.c (ipa_profile_generate_summary): Update.
22791 (ipa_propagate_frequency): Update.
22792 (ipa_profile): Update.
22793 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22794 * ipa-utils.c (ipa_merge_profiles): Update.
22795 * loop-doloop.c (doloop_modify): Update.
22796 * loop-unroll.c (report_unroll): Update.
22797 (unroll_loop_runtime_iterations): Update.
22798 * lto-cgraph.c (lto_output_edge): Update.
22799 (lto_output_node): Update.
22800 (input_node): Update.
22801 (input_edge): Update.
22802 (merge_profile_summaries): Update.
22803 * lto-streamer-in.c (input_cfg): Update.
22804 * lto-streamer-out.c (output_cfg): Update.
22805 * mcf.c (create_fixup_graph): Update.
22806 (adjust_cfg_counts): Update.
22807 (sum_edge_counts): Update.
22808 * modulo-sched.c (sms_schedule): Update.
22809 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22810 * predict.c (maybe_hot_count_p): Update.
22811 (probably_never_executed): Update.
22812 (dump_prediction): Update.
22813 (combine_predictions_for_bb): Update.
22814 (propagate_freq): Update.
22815 (handle_missing_profiles): Update.
22816 (counts_to_freqs): Update.
22817 (rebuild_frequencies): Update.
22818 (force_edge_cold): Update.
22819 * predict.h: Include profile-count.h
22820 (maybe_hot_count_p, counts_to_freqs): UPdate.
22821 * print-rtl-function.c: Do not include cfg.h
22822 * print-rtl.c: Include basic-block.h
22823 * profile-count.c: New file.
22824 * profile-count.h: New file.
22825 * profile.c (is_edge_inconsistent): Update.
22826 (correct_negative_edge_counts): Update.
22827 (is_inconsistent): Update.
22828 (set_bb_counts): Update.
22829 (read_profile_edge_counts): Update.
22830 (compute_frequency_overlap): Update.
22831 (compute_branch_probabilities): Update; Initialize and deinitialize
22833 (branch_prob): Update.
22834 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22835 (edge_gcov_count): New.
22836 (bb_gcov_count): New.
22837 * shrink-wrap.c (try_shrink_wrapping): Update.
22838 * tracer.c (better_p): Update.
22839 * trans-mem.c (expand_transaction): Update.
22840 (ipa_tm_insert_irr_call): Update.
22841 (ipa_tm_insert_gettmclone_call): Update.
22842 * tree-call-cdce.c: Update.
22843 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22844 (gimple_duplicate_sese_tail): Update.
22845 (gimple_account_profile_record): Update.
22846 (execute_fixup_cfg): Update.
22847 * tree-inline.c (copy_bb): Update.
22848 (copy_edges_for_bb): Update.
22849 (initialize_cfun): Update.
22850 (freqs_to_counts): Update.
22851 (copy_cfg_body): Update.
22852 (expand_call_inline): Update.
22853 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22854 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22855 (try_unroll_loop_completely): Update.
22856 (try_peel_loop): Update.
22857 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22858 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22859 * tree-ssa-loop-split.c (connect_loops): Update.
22860 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22861 * tree-ssa-reassoc.c (branch_fixup): Update.
22862 * tree-ssa-tail-merge.c (replace_block_by): Update.
22863 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22864 (compute_path_counts): Update.
22865 (update_profile): Update.
22866 (recompute_probabilities): Update.
22867 (update_joiner_offpath_counts): Update.
22868 (estimated_freqs_path): Update.
22869 (freqs_to_counts_path): Update.
22870 (clear_counts_path): Update.
22871 (ssa_fix_duplicate_block_edges): Update.
22872 (duplicate_thread_path): Update.
22873 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22874 (struct switch_conv_info): Update.
22875 * tree-tailcall.c (decrease_profile): Update.
22876 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22877 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22878 * value-prof.c (check_counter): Update.
22879 (gimple_divmod_fixed_value): Update.
22880 (gimple_mod_pow2): Update.
22881 (gimple_mod_subtract): Update.
22882 (gimple_ic_transform): Update.
22883 (gimple_stringop_fixed_value): Update.
22884 * value-prof.h (gimple_ic): Update.
22886 2017-06-02 Carl Love <cel@us.ibm.com>
22888 * config/rs6000/rs6000-c: Add support for built-in functions
22889 vector double vec_doublee (vector signed int);
22890 vector double vec_doublee (vector unsigned int);
22891 vector double vec_doublee (vector float);
22892 vector double vec_doubleh (vector signed int);
22893 vector double vec_doubleh (vector unsigned int);
22894 vector double vec_doubleh (vector float);
22895 vector double vec_doublel (vector signed int);
22896 vector double vec_doublel (vector unsigned int);
22897 vector double vec_doublel (vector float);
22898 vector double vec_doubleo (vector signed int);
22899 vector double vec_doubleo (vector unsigned int);
22900 vector double vec_doubleo (vector float);.
22901 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22902 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22904 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22905 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22906 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22908 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22909 vec_doublel, vec_doubleh.
22910 * doc/extend.texi: Update the built-in documentation file for the
22911 new built-in functions.
22913 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22916 * ipa-inline-analysis.c (free_growth_caches): Set
22917 edge_removal_hook_holder to NULL after removing it.
22919 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22921 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22922 comparision with zero.
22924 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22925 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22926 for early expansion of vec_min and vec_max builtins.
22927 (builtin_function_type): Add min/max unsigned variants to those
22928 identified as having unsigned arguments.
22930 2017-06-02 Olivier Hainque <hainque@adacore.com>
22932 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22934 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22936 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22937 Use VALL_F16 iterator rather than VALL.
22939 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22941 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22942 Emit CBNZ inside loop when doing a strong exchange and comparing
22943 against zero. Generate the CC flags after the loop.
22945 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22947 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22948 (dl_section_ref): New.
22949 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22950 On AIX, append an expression to subtract the size of the
22951 section length to dl_section_ref.
22953 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22955 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22956 for early expansion of vector absolute builtins.
22958 2017-06-02 Richard Biener <rguenther@suse.de>
22960 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22961 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22963 2017-06-02 Richard Biener <rguenther@suse.de>
22965 PR tree-optimization/80948
22966 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22967 stmt order as well.
22969 2017-06-02 Richard Biener <rguenther@suse.de>
22971 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22973 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22974 for inductions as relevant.
22976 2017-06-02 Richard Biener <rguenther@suse.de>
22978 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22979 (vectorizable_induction): ... this. Remove dead code.
22981 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22983 * builtins. (expand_builtin_alloca): Remove second parameter and
22984 infer its value from the first parameter instead.
22985 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22987 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22989 PR rtl-optimization/80903
22990 * loop-doloop.c (add_test): Unshare sequence.
22992 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22994 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22996 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22998 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23000 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23001 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23002 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23003 (xlogue_layout::get_stub_rtx): Make static.
23004 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23005 (xlogue_layout::compute_stub_managed_regs): Rename to...
23006 (xlogue_layout::count_stub_managed_regs): ...this.
23007 (xlogue_layout::is_stub_managed_reg): New function.
23008 (xlogue_layout::m_stub_names): Rename to...
23009 (xlogue_layout::s_stub_names): ...this, make static.
23010 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23011 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23012 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23013 xlogue_layout::s_stub_names): Instantiate statics.
23014 (stub_managed_regs): Remove.
23015 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23016 (disable_call_ms2sysv_xlogues): Rename to...
23017 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23018 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23020 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23021 change after reload_completed.
23022 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23024 (ix86_expand_prologue): Likewise.
23025 (ix86_expand_epilogue): Likewise.
23026 (ix86_expand_split_stack_prologue): Likewise.
23027 (ix86_compute_frame_layout): Remove frame parameter ...
23028 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23029 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23031 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23032 (ix86_frame): Move from here ...
23033 * config/i386/i386.h (ix86_frame): ... to here.
23034 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23035 complete ix86_frame data structure instead. Remove some_ld_name.
23037 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23039 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23040 symbols that hold a DECL_VALUE_EXPR.
23042 2017-06-01 Martin Jambor <mjambor@suse.cz>
23044 PR tree-optimization/80898
23045 * tree-sra.c (process_subtree_disqualification): Removed.
23046 (disqualify_candidate): Do not acll
23047 process_subtree_disqualification.
23048 (subtree_mark_written_and_enqueue): New function.
23049 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23050 RHS has been disqualified and re-queue LHS if necessary. Apart
23051 from that, ignore disqualified RHS.
23053 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23055 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23056 address fetch for z10 or later.
23058 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23060 * config/arc/arc.md (tst_movb): Add guard when splitting.
23062 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23064 * config/arc/arc.c (arc_can_eliminate): Test against
23065 arc_frame_pointer_needed.
23067 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23069 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23070 to prevent store reordering.
23071 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23072 (type): Add block type.
23073 (stack_tie): Define special instruction to be used in
23076 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23078 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23079 constraint. It is not valid for the pattern.
23080 (noncommutative_binary_comparison): Likewise.
23082 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23084 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23087 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23089 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23090 be used by the reg-alloc.
23092 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23094 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23095 reg-alloc when having mul64 or mul32x16 instructions.
23096 (mulsidi3): Likewise.
23097 (umulsidi3): Likewise.
23098 (mulsi32x16): New pattern.
23099 (mulsi64): Likewise.
23100 (mulsidi64): Likewise.
23101 (umulsidi64): Likewise.
23102 (MUL32x16_REG): Define.
23103 (mul64_600): Use MUL32x16_REG.
23104 (mac64_600): Likewise.
23105 (umul64_600): Likewise.
23106 (umac64_600): Likewise.
23108 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23110 * config/arc/arc.md (mulsi3_700): Make it commutative.
23112 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23114 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23116 (*sign_extendsidi2_insn): Likewise for movstosw.
23118 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23120 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23121 the type of the input discriminant value. Convert the
23122 discriminant value of signedness vary.
23124 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23126 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23127 Add to -Wall section.
23129 2017-06-01 Richard Biener <rguenther@suse.de>
23131 PR middle-end/66313
23132 * fold-const.c (fold_plusminus_mult_expr): If the factored
23133 factor may be zero use a wrapping type for the inner operation.
23134 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23135 and handle moved defs.
23136 (process_assignment): Properly guard the unary op case. Return a
23137 tri-state indicating that moving the stmt before the call may allow
23138 to continue. Pass through to_move.
23139 (find_tail_calls): Handle moving unrelated defs before
23142 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23145 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23146 splitter result in the canonical way.
23148 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23150 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23151 also for 32bit target. Update insn attributes.
23152 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23154 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23156 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23157 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23158 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23159 (_mm_maskz_min_ss): New intrinsics.
23161 2017-05-31 Martin Liska <mliska@suse.cz>
23163 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23164 Change comment style to one we normally use.
23165 (vectorizable_reduction): Likewise.
23166 (vectorizable_induction): Likewise.
23167 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23168 (vectorizable_call): Likewise.
23169 (vectorizable_simd_clone_call): Likewise.
23170 (vectorizable_conversion): Likewise.
23171 (vectorizable_assignment): Likewise.
23172 (vectorizable_shift): Likewise.
23173 (vectorizable_operation): Likewise.
23174 (vectorizable_store): Likewise.
23175 (vectorizable_load): Likewise.
23176 * tree-vectorizer.h: Likewise.
23178 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23180 * passes.c (emergency_dump_function): New.
23181 * tree-pass.h (emergency_dump_function): Declare.
23182 * plugin.c (plugins_internal_error_function): Remove.
23183 * plugin.h (plugins_internal_error_function): Remove declaration.
23184 * toplev.c (internal_error_function): New static function. Use it...
23185 (general_init): ...here.
23187 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23189 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23190 (arc_rtx_costs): Add costs for new patterns.
23191 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23192 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23194 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23196 * tree-ssa-strlen.c (get_next_strinfo): New function.
23197 (get_stridx_plus_constant): Use it.
23198 (zero_length_string): Likewise.
23199 (adjust_related_strinfos): Likewise.
23200 (adjust_last_stmt): Likewise.
23202 2017-05-31 Richard Biener <rguenther@suse.de>
23205 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23206 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23208 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23210 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23211 loop_vinfo argument and use of dependence distance vectors.
23212 Check instead whether the two references differ only in their
23213 initial value and assume that they have the same alignment if the
23214 difference is a multiple of the vector alignment.
23215 (vect_analyze_data_refs_alignment): Update call accordingly.
23217 2017-05-31 Martin Liska <mliska@suse.cz>
23220 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23222 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23224 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23225 (create_intersect_range_checks): Move from ...
23226 * tree-data-ref.c (create_intersect_range_checks_index)
23227 (create_intersect_range_checks): ... to here.
23228 (create_runtime_alias_checks): New function factored from ...
23229 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23230 here. Call above function.
23231 * tree-data-ref.h (create_runtime_alias_checks): New function.
23233 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23235 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23236 segment length for dr_b and compute it in wide_int.
23238 2017-05-31 Richard Biener <rguenther@suse.de>
23240 PR tree-optimization/80906
23241 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23242 and pass through iv_map.
23243 (copy_bb_and_scalar_dependences): Adjust.
23244 (translate_pending_phi_nodes): Likewise.
23245 (copy_loop_close_phi_args): Handle code-generating IVs instead
23248 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23250 * diagnostic-color.c (color_dict): Add "type-diff".
23251 (parse_gcc_colors): Update comment.
23252 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23253 -fdiagnostics-show-template-tree and -fno-elide-type.
23254 (GCC_COLORS): Add type-diff to example.
23256 (-fdiagnostics-show-template-tree): New.
23257 (-fno-elide-type): New.
23258 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23259 the pp_format_decoder callback. Call any m_format_postprocessor's
23261 (pretty_printer::pretty_printer): Initialize
23262 m_format_postprocessor.
23263 (pretty_printer::~pretty_printer): Delete any
23264 m_format_postprocessor.
23265 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23266 (class format_postprocessor): New class.
23267 (struct pretty_printer::format_decoder): Document the new parameters.
23268 (struct pretty_printer::m_format_postprocessor): New field.
23269 * tree-diagnostic.c (default_tree_printer): Update for new
23270 bool and const char ** params.
23271 * tree-diagnostic.h (default_tree_printer): Likewise.
23273 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23275 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23276 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23277 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23278 rs6000_gen_cell_microcode code.
23279 (rs6000_final_prescan_insn): Delete.
23280 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23281 "warn-cell-microcode" entries.
23282 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23283 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23284 throughout. Change cc_reg_not_micro_cr0_operand to
23285 cc_reg_not_cr0_operand throughout.
23286 (*extendhi<mode>2_noload): Delete.
23287 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23288 (mwarn-cell-microcode): Delete.
23289 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23290 -mgen-cell-microcode and -mwarn-cell-microcode.
23292 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23295 * config/i386/constraints.md (Yd): New constraint.
23297 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23298 and (?Yd, r) alternatives. Update insn attributes.
23299 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23300 and (?*Yd, r) alternatives. Update insn attributes.
23301 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23303 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23305 * gimplify.c (gimplify_modify_expr): Don't create a
23306 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23309 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23311 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23313 2017-05-30 Richard Biener <rguenther@suse.de>
23315 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23316 and reduc_def fields.
23317 (STMT_VINFO_REDUC_TYPE): New define.
23318 (STMT_VINFO_REDUC_DEF): Likewise.
23319 (vect_force_simple_reduction): Adjust prototype.
23320 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23321 (vect_is_simple_reduction): Remove check_reduction argument.
23322 (vect_force_simple_reduction): Adjust and set
23323 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23324 (vectorizable_reduction): Do not re-do reduction analysis
23325 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23326 * tree-parloops.c (gather_scalar_reductions): Adjust.
23328 2017-05-30 Richard Biener <rguenther@suse.de>
23330 PR middle-end/80901
23331 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23334 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23336 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23337 Introduce unknown_misalignment parameter and remove vf.
23338 (vect_peeling_hash_get_lowest_cost):
23339 Pass unknown_misalignment parameter.
23340 (vect_enhance_data_refs_alignment):
23341 Fix unsupportable data ref treatment.
23343 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23345 * tree-vect-data-refs.c (vect_get_data_access_cost):
23346 Workaround for SLP handling.
23347 (vect_enhance_data_refs_alignment):
23348 Compute costs for doing no peeling at all, compare to the best
23349 peeling costs so far and avoid peeling if cheaper.
23351 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23353 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23354 Return peeling info and set costs to zero for unlimited cost
23356 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23357 with unknown misalignment. Compute and costs for unknown
23358 misalignment, compare them to the costs for known misalignment
23359 and choose the cheapest for peeling.
23361 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23363 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23364 (vect_get_peeling_costs_all_drs): Create function.
23365 (vect_peeling_hash_get_lowest_cost):
23366 Use vect_get_peeling_costs_all_drs.
23367 (vect_peeling_supportable): Create function.
23368 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23370 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23372 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23373 DR_HAS_NEGATIVE_STEP.
23374 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23375 (vect_enhance_data_refs_alignment): Use.
23376 (vect_duplicate_ssa_name_ptr_info): Use.
23377 * tree-vectorizer.h (dr_misalignment): Use.
23378 (known_alignment_for_access_p): Use.
23380 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23383 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23385 (has_section_name): New function.
23387 2017-05-30 Martin Liska <mliska@suse.cz>
23390 * auto-profile.c (get_function_decl_from_block): Fix
23393 2017-05-30 Richard Biener <rguenther@suse.de>
23395 PR middle-end/80876
23396 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23398 2017-05-30 Martin Liska <mliska@suse.cz>
23400 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23401 * dumpfile.h (struct dump_file_info): Remove ctors.
23403 2017-05-30 Martin Liska <mliska@suse.cz>
23405 * predict.def: Fix GNU coding style.
23407 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23409 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23410 Mark 'to' argument with ATTRIBUTE_UNUSED.
23412 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23414 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23415 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23416 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23419 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23421 * doc/install.texi (Options specification): Restore entry of
23422 --enable-sjlj-exceptions.
23424 2017-05-27 Michael Eager <eager@eagercon.com>
23427 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23429 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23431 * config/microblaze/microblaze.h
23432 (FIXED_REGISTERS): Update in macro.
23433 (CALL_USED_REGISTERS): Update in macro.
23435 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23437 * doc/install.texi: Add links to macOS binary distributions.
23439 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23443 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23445 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23447 2017-05-26 Martin Liska <mliska@suse.cz>
23449 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23451 2017-05-26 Martin Liska <mliska@suse.cz>
23453 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23454 always leading ';; '.
23455 (dump_bb_info): Likewise.
23456 (brief_dump_cfg): Likewise.
23457 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23458 * dumpfile.c: Remove usage of TDF_VERBOSE.
23459 * dumpfile.h (enum dump_kind): Likewise.
23460 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23461 * print-tree.c (debug_verbose): Remove.
23462 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23463 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23464 * tree-diagnostic.c (default_tree_printer): Replace
23465 TDF_DIAGNOSTIC with TDF_SLIM.
23467 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23469 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23470 in parameter loop, rather than loop_vinfo.
23471 (create_intersect_range_checks): Ditto.
23472 (vect_create_cond_for_alias_checks): Update call to above functions.
23474 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23476 PR tree-optimization/80815
23477 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23478 for merging runtime alias checks. Handle negative DR_STEPs.
23480 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23482 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23484 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23485 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23486 out code pruning runtime alias checks.
23487 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23488 factored out from above.
23489 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23491 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23493 (prune_runtime_alias_test_list): New decalaration.
23495 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23497 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23498 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23499 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23500 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23501 (operator==, comp_dr_with_seg_len_pair): Ditto.
23502 (vect_prune_runtime_alias_test_list): Ditto.
23504 2017-05-26 Martin Liska <mliska@suse.cz>
23507 * params.def: Bound partial-inlining-entry-probability param.
23509 2017-05-26 Marek Polacek <polacek@redhat.com>
23512 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23515 2017-05-26 Richard Biener <rguenther@suse.de>
23517 PR tree-optimization/80842
23518 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23521 2017-05-26 Richard Biener <rguenther@suse.de>
23523 PR tree-optimization/80844
23524 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23526 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23528 * doc/md.texi (Machine Constraints): Update x86 family
23529 machine constraints section to match 'config/i386/constraints.md'.
23531 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23533 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23535 2017-05-25 Nathan Sidwell <nathan@acm.org>
23537 * doc/invoke.texi (--enable-languages): Update documentation.
23539 2017-05-25 Martin Liska <mliska@suse.cz>
23541 * dumpfile.c: Add TDF_FOLDING.
23542 * dumpfile.h (enum dump_kind): Likewise.
23543 * genmatch.c (dt_simplify::gen_1): Use it.
23545 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23547 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23549 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23551 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23552 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23554 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23556 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23557 * match.pd (X == C): Rewrite it here.
23558 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23559 with_certain_nonzero_bits2): New predicates.
23560 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23562 2017-05-24 Nathan Sidwell <nathan@acm.org>
23564 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23567 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23570 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23572 * config/powerpcspe: New port. Files are copied from the rs6000
23573 port, with "rs6000" in filenames replaced by "powerpcspe".
23575 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23577 PR rtl-optimization/80754
23578 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23580 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23582 * config/sparc/sparc.md (length): Return the correct value for -mflat
23583 sibcalls to match output_sibcall.
23585 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23589 * config/rs6000/rs6000.c (struct machine_function): Add new field
23591 (rs6000_get_separate_components): Init that field, use it.
23592 (rs6000_components_for_bb): Use the field.
23594 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23596 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23598 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23600 PR middle-end/80823
23601 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23603 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23606 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23607 address against address_operand predicate.
23608 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23610 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23612 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23613 parameters passed indirectly.
23615 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23617 * config/i386/i386.md (*movdi_internal): Remove SSE4
23618 alternative 18 (?r, *v). Update insn attributes.
23619 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23620 Update insn attributes.
23621 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23622 Update insn attributes.
23623 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23624 alternative 1 (r, v). Remove isa attribute.
23625 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23626 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23627 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23629 2017-05-23 Tom de Vries <tom@codesourcery.com>
23631 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23634 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23636 * cgraphunit.c (symbol_table::process_new_functions): Update.
23637 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23638 (inline_generate_summary): Rename to ...
23639 (ipa_fn_summary_generate): ... this one.
23640 (inline_read_summary): Rename to ...
23641 (ipa_fn_summary_read): ... this one.
23642 (inline_write_summary): Rename to ...
23643 (ipa_fn_summary_write): ... this one.
23644 (inline_free_summary): Rename to ...
23645 (ipa_free_fn_summary): ... this one.
23646 (pass_data_local_fn_summary, pass_local_fn_summary,
23647 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23648 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23649 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23650 make_pass_ipa_fn_summary): New.
23651 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23652 inline_write_summary, inline_free_summary): Remove.
23653 (ipa_free_fn_summary) : New.
23654 * ipa-inline.c (ipa_inline): Update.
23655 (pass_ipa_inline): Do not generate summaries.
23656 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23658 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23659 and add pass_ipa_fn_summary.
23660 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23662 (make_pass_inline_parameters): Remove.
23664 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23666 * omp-low.c (struct omp_context): Remove "default_kind" member.
23669 * omp-offload.c (execute_oacc_device_lower): Remove the
23670 parallelism dimensions function attributes for unparallelized
23671 OpenACC kernels constructs.
23673 2017-05-23 Martin Liska <mliska@suse.cz>
23675 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23677 (cgraph_edge::make_speculative): Likewise.
23678 (cgraph_edge::resolve_speculation): Likewise.
23679 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23680 (cgraph_node::dump): Likewise.
23681 * cgraph.h: Likewise.
23682 * cgraphunit.c (analyze_functions): Likewise.
23683 (symbol_table::compile): Likewise.
23684 * ipa-cp.c (print_all_lattices): Likewise.
23685 (determine_versionability): Likewise.
23686 (initialize_node_lattices): Likewise.
23687 (ipcp_verify_propagated_values): Likewise.
23688 (estimate_local_effects): Likewise.
23689 (update_profiling_info): Likewise.
23690 (create_specialized_node): Likewise.
23691 (perhaps_add_new_callers): Likewise.
23692 (decide_about_value): Likewise.
23693 (decide_whether_version_node): Likewise.
23694 (identify_dead_nodes): Likewise.
23695 (ipcp_store_bits_results): Likewise.
23696 * ipa-devirt.c (dump_targets): Likewise.
23697 (ipa_devirt): Likewise.
23698 * ipa-icf.c (sem_item::dump): Likewise.
23699 (sem_function::equals): Likewise.
23700 (sem_variable::equals): Likewise.
23701 (sem_item_optimizer::read_section): Likewise.
23702 (sem_item_optimizer::execute): Likewise.
23703 (congruence_class::dump): Likewise.
23704 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23705 (dump_inline_summary): Likewise.
23706 (estimate_node_size_and_time): Likewise.
23707 (inline_analyze_function): Likewise.
23708 * ipa-inline-transform.c (inline_call): Likewise.
23709 * ipa-inline.c (report_inline_failed_reason): Likewise.
23710 (want_early_inline_function_p): Likewise.
23711 (edge_badness): Likewise.
23712 (update_edge_key): Likewise.
23713 (inline_small_functions): Likewise.
23714 * ipa-profile.c (ipa_profile): Likewise.
23715 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23716 (ipa_make_edge_direct_to_target): Likewise.
23717 (remove_described_reference): Likewise.
23718 (ipa_impossible_devirt_target): Likewise.
23719 (propagate_controlled_uses): Likewise.
23720 (ipa_print_node_params): Likewise.
23721 (ipcp_transform_function): Likewise.
23722 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23723 (propagate_pure_const): Likewise.
23724 * ipa-reference.c (generate_summary): Likewise.
23725 (read_write_all_from_decl): Likewise.
23726 (propagate): Likewise.
23727 (ipa_reference_read_optimization_summary): Likewise.
23728 * ipa-utils.c (ipa_merge_profiles): Likewise.
23729 * ipa.c (walk_polymorphic_call_targets): Likewise.
23730 (symbol_table::remove_unreachable_nodes): Likewise.
23731 (ipa_single_use): Likewise.
23732 * passes.c (execute_todo): Likewise.
23733 * predict.c (drop_profile): Likewise.
23734 * symtab.c (symtab_node::get_dump_name): New function.
23735 (symtab_node::dump_name): Likewise.
23736 (symtab_node::dump_asm_name): Likewise.
23737 (symtab_node::dump_references): Likewise.
23738 (symtab_node::dump_referring): Likewise.
23739 (symtab_node::dump_base): Likewise.
23740 (symtab_node::debug_symtab): Likewise.
23741 * tree-sra.c (convert_callers_for_node): Likewise.
23742 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23743 * value-prof.c (init_node_map): Likewise.
23745 2017-05-23 Martin Liska <mliska@suse.cz>
23747 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23748 and symtab_node::debug_symtab to symbol_table::debug.
23749 * cgraphunit.c (analyze_functions): Use the renamed function.
23750 (symbol_table::compile): Likewise.
23751 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23752 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23753 * passes.c (execute_todo): Likewise.
23754 * symtab.c (symbol_table::dump): New function.
23755 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23757 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23759 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23760 that nonconst implies exec.
23762 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23764 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23765 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23766 (inline_edge_summary_vec): Turn into ...
23767 (ipa_call_summaries): ... this one.
23768 (redirect_to_unreachable, edge_set_predicate,
23769 evaluate_properties_for_edge, inline_summary_alloc,
23770 reset_ipa_call_summary, reset_inline_summary,
23771 inline_summary_t::duplicate): Update.
23772 (inline_edge_duplication_hook): Turn to ...
23773 (ipa_call_summary_t::duplicate): ... this one.
23774 (inline_edge_removal_hook): Turn to ...
23775 (ipa_call_summary_t::remove): ... this one.
23776 (dump_inline_edge_summary): Turn to ...
23777 (dump_ipa_call_summary): ... this one.
23778 (estimate_function_body_sizes): Update.
23779 (inline_update_callee_summaries): Update.
23780 (remap_edge_change_prob): Update.
23781 (remap_edge_summaries): Update.
23782 (inline_merge_summary): Update.
23783 (do_estimate_edge_time): Update.
23784 (inline_generate_summary): Update.
23785 (inline_read_section): Update.
23786 (inline_read_summary): Update.
23787 (inline_free_summary): Update.
23788 * ipa-inline.c (can_inline_edge_p): Update.
23789 (compute_inlined_call_time): Update.
23790 (want_inline_small_function_p): Update.
23791 (edge_badness): Update.
23792 (early_inliner): Update.
23793 * ipa-inline.h (inline_edge_summary): Turn to ...
23794 (ipa_call_summary): ... this one.
23795 (ipa_call_summary_t): New class.
23796 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23797 (ipa_call_summaries): New.
23798 (inline_edge_summary): Remove.
23799 (estimate_edge_growth): Update.
23800 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23801 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23802 * ipa-split.c (execute_split_functions): Update.
23803 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23805 2017-05-23 Tom de Vries <tom@codesourcery.com>
23807 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23808 attributes): Document rdrand effective target.
23810 2017-05-23 Tom de Vries <tom@codesourcery.com>
23812 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23813 attributes): Sort alphabetically.
23815 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23817 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23819 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23822 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23823 V2DF/V2DI splat into two separate patterns, one that handles
23824 registers, and the other that only handles memory. Drop support
23825 for splatting from a GPR on ISA 2.07 and then splitting the
23826 splat into direct move and splat.
23827 (vsx_splat_<mode>_reg): Likewise.
23828 (vsx_splat_<mode>_mem): Likewise.
23830 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23832 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23834 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23836 PR middle-end/80809
23837 * omp-low.c (finish_taskreg_remap): New function.
23838 (finish_taskreg_scan): If unit size of ctx->record_type
23839 is non-constant, unshare the size expression and replace
23840 decls in it with possible outer var refs.
23842 PR middle-end/80809
23843 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23844 GOVD_SHARED rather than GOVD_PRIVATE with it.
23845 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23846 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23848 PR middle-end/80853
23849 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23850 as last argument to build_outer_var_ref for pointer bases of array
23851 section reductions.
23853 2017-05-19 Martin Sebor <msebor@redhat.com>
23855 * print-tree.c (print_node): Print DECL_READ_P flag.
23857 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23859 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23860 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23861 * cgraph.c: Likewise.
23862 * cgraphunit.c: Likewise.
23863 * gengtype.c: Likewise.
23864 * ipa-cp.c: Likewise.
23865 * ipa-devirt.c: Likewise.
23866 * ipa-icf.c: Likewise.
23867 * ipa-predicate.c: Likewise.
23868 * ipa-profile.c: Likewise.
23869 * ipa-prop.c: Likewise.
23870 * ipa-split.c: Likewise.
23872 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23873 edge_predicate_pool, dump_inline_hints,
23874 inline_summary::account_size_time, redirect_to_unreachable,
23875 edge_set_predicate, set_hint_predicate,
23876 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23877 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23878 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23879 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23880 ipa_call_summary_t::remove, initialize_growth_caches,
23881 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23882 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23883 mark_modified, unmodified_parm_1, unmodified_parm,
23884 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23885 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23886 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23887 will_be_nonconstant_predicate, record_modified_bb_info,
23888 get_minimal_bb, record_modified, param_change_prob,
23889 phi_result_unknown_predicate, predicate_for_phi_result,
23890 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23891 estimate_function_body_sizes, compute_inline_parameters,
23892 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23893 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23894 inline_update_callee_summaries, remap_edge_change_prob,
23895 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23896 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23897 inline_analyze_function, inline_summary_t::insert,
23898 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23899 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23900 inline_free_summary): Move to ipa-fnsummary.h
23901 (predicate_t): Remove.
23902 * ipa-fnsummary.c: New file.
23903 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23904 (enum inline_hints_vals, inline_hints, agg_position_info,
23905 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23906 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23907 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23908 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23909 inline_read_summary, inline_write_summary, inline_free_summary,
23910 inline_analyze_function, initialize_inline_failed,
23911 inline_merge_summary, inline_update_overall_summary,
23912 compute_inline_parameters): Move to ipa-fnsummary.h
23913 * ipa-fnsummary.h: New file.
23914 * ipa-inline-transform.h: Include ipa-inline.h.
23915 * ipa-inline.c: LIkewise.
23917 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23919 * ipa-inline.c (edge_badness): Use inlined_time instead of
23920 inline_summaries->get.
23922 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23924 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23926 2017-05-22 Nathan Sidwell <nathan@acm.org>
23928 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23929 (fdump-lang): Document 'raw' option.
23930 * dumpfile.h (TDI_tu): Delete.
23931 * dumpfile.c (dump_files): Remove translation-unit.
23932 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23934 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23936 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23937 command option from $(AWK) call.
23938 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23939 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23940 [FORMAT]: Remove handling of variable.
23941 * config/avr/t-multilib: Regenerate.
23943 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23945 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23947 (dump_inline_summary): Do not print self_time.
23948 (estimate_function_body_sizes): Do not set self_time.
23949 (compute_inline_parameters): Likewise.
23950 (inline_read_section, inline_write_summary): Do not stream self_time.
23951 * ipa-inline.h (inline_summary): Drop self_time.
23953 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23955 * ipa-inline-analysis.c (account_size_time): Rename to ...
23956 (inline_summary::account_size_time): ... this one.
23957 (reset_ipa_call_summary): Turn to ...
23958 (ipa_call_summary::reset): ... this one.
23959 (reset_inline_summary): Turn to ...
23960 (inline_summary::reset): ... this one.
23961 (inline_summary_t::remove): Update.
23962 (inline_summary_t::duplicate): Update.
23963 (ipa_call_summary_t::remove): Update.
23964 (dump_inline_summary): Update.
23965 (estimate_function_body_sizes): Update.
23966 (compute_inline_parameters): Update.
23967 (estimate_node_size_and_time): Update.
23968 (inline_merge_summary): Update.
23969 (inline_update_overall_summary): Update.
23970 (inline_read_section): Update.
23971 (inline_write_summary): Update.
23972 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23973 add account_size_time and reset member functions.
23974 (ipa_call_summary): Add reset function.
23975 * ipa-predicate.h (predicate::operator &): Constify.
23977 2017-05-22 Richard Biener <rguenther@suse.de>
23979 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23981 2017-05-19 Jason Merrill <jason@redhat.com>
23983 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23985 2017-05-19 Marek Polacek <polacek@redhat.com>
23988 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23989 TYPE_OVERFLOW_WRAPS checks.
23991 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23993 * tree-core.h (enum omp_clause_default_kind): Add
23994 "OMP_CLAUSE_DEFAULT_PRESENT".
23995 * tree-pretty-print.c (dump_omp_clause): Handle it.
23996 * gimplify.c (enum gimplify_omp_var_data): Add
23997 "GOVD_MAP_FORCE_PRESENT".
23998 (gimplify_adjust_omp_clauses_1): Map it to
23999 "GOMP_MAP_FORCE_PRESENT".
24000 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24002 * gimplify.c (oacc_default_clause): Clarify.
24004 2017-05-19 Nathan Sidwell <nathan@acm.org>
24006 LANG_HOOK_REGISTER_DUMPS
24007 * toplev.c (general_init): Call register dump lang hook.
24008 * doc/invoke.texi: Document -fdump-lang option family.
24009 * dumpfile.c (dump_files): Remove class dump here.
24010 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24011 * dumpfile.h (tree_dump_index): Remove TDI_class.
24012 * langhooks-def.h (lhd_register_dumps): Declare.
24013 (LANG_HOOKS_REGISTER_DUMPS): Define.
24014 (LANG_HOOKS_INITIALIZER): Add it.
24015 * langhooks.c (lhd_register_dumps): Define.
24016 * langhooks.h (struct lang_hooks): Add register_dumps.
24018 2017-05-19 Nathan Sidwell <nathan@acm.org>
24020 * context.h (context::set_passes): New.
24021 * context.c (context::context): Do not create pass manager.
24022 * toplev.c (general_init): Create pass manager here.
24024 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24026 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24027 use this splitter if two add or or instructions would also work for
24028 the constant we want to generate.
24030 2017-05-19 Richard Biener <rguenther@suse.de>
24033 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24034 predicate evaluation.
24036 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24038 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24040 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24043 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24045 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24046 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24047 (inline_edge_summary_vec): Turn into ...
24048 (ipa_call_summaries): ... this one.
24049 (redirect_to_unreachable, edge_set_predicate,
24050 evaluate_properties_for_edge, inline_summary_alloc,
24051 reset_ipa_call_summary, reset_inline_summary,
24052 inline_summary_t::duplicate): Update.
24053 (inline_edge_duplication_hook): Turn to ...
24054 (ipa_call_summary_t::duplicate): ... this one.
24055 (inline_edge_removal_hook): Turn to ...
24056 (ipa_call_summary_t::remove): ... this one.
24057 (dump_inline_edge_summary): Turn to ...
24058 (dump_ipa_call_summary): ... this one.
24059 (estimate_function_body_sizes): Update.
24060 (inline_update_callee_summaries): Update.
24061 (remap_edge_change_prob): Update.
24062 (remap_edge_summaries): Update.
24063 (inline_merge_summary): Update.
24064 (do_estimate_edge_time): Update.
24065 (inline_generate_summary): Update.
24066 (inline_read_section): Update.
24067 (inline_read_summary): Update.
24068 (inline_free_summary): Update.
24069 * ipa-inline.c (can_inline_edge_p): Update.
24070 (compute_inlined_call_time): Update.
24071 (want_inline_small_function_p): Update.
24072 (edge_badness): Update.
24073 (early_inliner): Update.
24074 * ipa-inline.h (inline_edge_summary): Turn to ...
24075 (ipa_call_summary): ... this one.
24076 (ipa_call_summary_t): New class.
24077 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24078 (ipa_call_summaries): New.
24079 (inline_edge_summary): Remove.
24080 (estimate_edge_growth): Update.
24081 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24082 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24083 * ipa-split.c (execute_split_functions): Update.
24084 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24086 2017-05-19 Richard Biener <rguenther@suse.de>
24088 PR middle-end/80764
24089 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24091 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24093 * config/rs6000/rs6000.c (struct machine_function): Add field
24094 fpr_is_wrapped_separately.
24095 (rs6000_get_separate_components): Use 64 components. Handle the
24096 new FPR components.
24097 (rs6000_components_for_bb): Handle the FPR components.
24098 (rs6000_emit_prologue_components): Handle the FPR components.
24099 (rs6000_emit_epilogue_components): Handle the FPR components.
24100 (rs6000_set_handled_components): Handle the FPR components.
24101 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24102 that are already separately shrink-wrapped.
24103 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24104 that are already separately shrink-wrapped.
24106 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24109 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24112 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24113 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24114 cases where the register allocator uses a move and an offsettable
24115 memory operation to/from a FPR register on ISA 2.06/2.07.
24116 (define_peephole2 for Altivec d-form store): Likewise.
24118 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24121 * config/i386/mmx.md (*mov<mode>_internal): Enable
24122 alternatives 11, 12, 13 and 14 also for 32bit targets.
24123 Remove alternatives 15, 16, 17 and 18.
24124 * config/i386/sse.md (vec_concatv2di): Change
24125 alternative (!x, *y) to (x, ?!*Yn).
24127 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24129 * dumpfile.h (enum dump_kind): Remove stray comma.
24131 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24133 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24134 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24135 predicate::num_conditions
24136 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24137 (CHANGED): turn into predicate::changed.
24138 (agg_position_info): Move to ipa-predicate.h
24139 (add_condition, predicate::add_clause, predicate::operator &=,
24140 predicate::or_with, predicate::evaluate, predicate::probability,
24141 dump_condition, dump_clause, predicate::dump,
24142 predicate::remap_after_duplication, predicate::remap_after_inlining,
24143 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24144 (evaluate_conditions_for_known_args): Update.
24145 (set_cond_stmt_execution_predicate): Update.
24146 * ipa-inline.h: Include ipa-predicate.h
24147 (condition, inline_param_summary, conditions, agg_position_info,
24148 predicate): Move to ipa-predicate.h
24149 * ipa-predicate.c: New file.
24150 * ipa-predicate.h: New file.
24152 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24154 * final.c (leaf_function_p): Check we are not in a sequence.
24156 2017-05-18 Martin Liska <mliska@suse.cz>
24158 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24159 * dumpfile.c (dump_register): Use new enum dump_kind.
24160 (get_dump_file_name): Likewise.
24161 (dump_enable_all): Likewise.
24162 (dump_switch_p_1): Likewise.
24163 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24164 * dumpfile.h (enum dump_kind): New enum type.
24165 (struct dump_file_info): Create constructor and
24166 format fields and comments.
24167 * passes.c (pass_manager::register_one_dump_file):
24169 * statistics.c (statistics_early_init): Likewise.
24170 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24171 TDF_TREE with TDF_SLIM.
24172 (gather_memory_references_ref): Likewise.
24174 2017-05-18 Martin Liska <mliska@suse.cz>
24176 * vec.h (struct vnull): Use it.
24178 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24180 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24181 (true_predicate, false_predicate, true_predicate_p,
24182 false_predicate_p): Remove.
24183 (single_cond_predicate, not_inlined_predicate): Turn to member function
24185 (add_condition): Update.
24186 (add_clause): Turn to...
24187 (predicate::add_clause): ... this one; update; allow passing NULL
24189 (and_predicates): Turn to ...
24190 (predicate::operator &=): ... this one.
24191 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24192 (or_predicates): Turn to ...
24193 (predicate::or_with): ... this one.
24194 (evaluate_predicate): Turn to ...
24195 (predicate::evaluate): ... this one.
24196 (predicate_probability): Turn to ...
24197 (predicate::probability): ... this one.
24198 (dump_condition): Update.
24199 (dump_predicate): Turn to ...
24200 (predicate::dump): ... this one.
24201 (account_size_time): Update.
24202 (edge_set_predicate): Update.
24203 (set_hint_predicate): UPdate.
24204 (evaluate_conditions_for_known_args): Update.
24205 (evaluate_properties_for_edge): Update.
24206 (remap_predicate_after_duplication): Turn to...
24207 (predicate::remap_after_duplication): ... this one.
24208 (remap_hint_predicate_after_duplication): Update.
24209 (inline_summary_t::duplicate): UPdate.
24210 (dump_inline_edge_summary): Update.
24211 (dump_inline_summary): Update.
24212 (set_cond_stmt_execution_predicate): Update.
24213 (set_switch_stmt_execution_predicate): Update.
24214 (compute_bb_predicates): Update.
24215 (will_be_nonconstant_expr_predicate): Update.
24216 (will_be_nonconstant_predicate): Update.
24217 (phi_result_unknown_predicate): Update.
24218 (predicate_for_phi_result): Update.
24219 (array_index_predicate): Update.
24220 (estimate_function_body_sizes): Update.
24221 (estimate_node_size_and_time): Update.
24222 (estimate_ipcp_clone_size_and_time): Update.
24223 (remap_predicate): Rename to ...
24224 (predicate::remap_after_inlining): ... this one.
24225 (remap_hint_predicate): Update.
24226 (inline_merge_summary): Update.
24227 (inline_update_overall_summary): Update.
24228 (estimate_size_after_inlining): Update.
24229 (read_predicate): Rename to ...
24230 (predicate::stream_in): ... this one.
24231 (read_inline_edge_summary): Update.
24232 (write_predicate): Rename to ...
24233 (predicate::stream_out): ... this one.
24234 (write_inline_edge_summary): Update.
24235 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24236 (clause_t): Turn to uint32_t
24237 (predicate): Turn to class; implement constructor and operators
24239 (size_time_entry): Update.
24240 (inline_summary): Update.
24241 (inline_edge_summary): Update.
24243 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24245 * fold-const.c (fold_binary_loc): Move transformation...
24246 * match.pd (C - X CMP X): ... here.
24248 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24250 * config/sparc/sparc.c (sparc_option_override): Set function
24251 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24252 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24254 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24256 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24258 2017-05-18 Marek Polacek <polacek@redhat.com>
24261 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24262 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24263 gimple_assign_load_p.
24265 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24267 PR middle-end/80692
24268 * real.c (do_compare): Give decimal_do_compare preference over
24269 comparing just the signs.
24271 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24273 * doc/md.texi (Canonicalization of Instructions): Describe the
24274 canonical form of instructions that inherently set a condition
24277 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24279 PR middle-end/80775
24280 * tree-cfg.c: Move deletion of unreachable case statements to after
24281 the merging of consecutive case labels.
24283 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24285 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24286 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24287 restoring of callee-saved registers.
24289 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24291 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24292 * config/visium/visium.c (single_set_and_flags): Likewise.
24293 * config/visium/visium.md (Substitutions): Likewise.
24295 2017-05-17 Martin Liska <mliska@suse.cz>
24297 * cfg.c: Introduce dump_flags_t type and
24298 use it instead of int type.
24300 * cfghooks.c: Likewise.
24301 * cfghooks.h (struct cfg_hooks): Likewise.
24302 * cfgrtl.c: Likewise.
24303 * cfgrtl.h: Likewise.
24304 * cgraph.c (cgraph_node::get_body): Likewise.
24305 * coretypes.h: Likewise.
24306 * domwalk.c: Likewise.
24307 * domwalk.h: Likewise.
24308 * dumpfile.c (struct dump_option_value_info): Likewise.
24309 (dump_enable_all): Likewise.
24310 (dump_switch_p_1): Likewise.
24311 (opt_info_switch_p): Likewise.
24312 * dumpfile.h (enum tree_dump_index): Likewise.
24313 (struct dump_file_info): Likewise.
24314 * genemit.c: Likewise.
24315 * generic-match-head.c: Likewise.
24316 * gengtype.c (open_base_files): Likewise.
24317 * gimple-pretty-print.c: Likewise.
24318 * gimple-pretty-print.h: Likewise.
24319 * graph.c (print_graph_cfg): Likewise.
24320 * graphite-scop-detection.c (dot_all_sese): Likewise.
24321 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24322 * loop-unroll.c (report_unroll): Likewise.
24323 * passes.c (pass_manager::register_one_dump_file): Likewise.
24324 * print-tree.c: Likewise.
24325 * statistics.c: Likewise.
24326 * tree-cfg.c: Likewise.
24327 * tree-cfg.h: Likewise.
24328 * tree-dfa.c: Likewise.
24329 * tree-dfa.h: Likewise.
24330 * tree-dump.c (dump_function): Likewise.
24331 * tree-dump.h (struct dump_info): Likewise.
24332 * tree-pretty-print.c: Likewise.
24333 * tree-pretty-print.h: Likewise.
24334 * tree-ssa-live.c: Likewise.
24335 * tree-ssa-live.h: Likewise.
24336 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24337 * tree-vect-loop.c: Likewise.
24338 * tree-vect-slp.c: Likewise.
24340 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24341 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24343 PR tree-optimization/80457
24344 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24345 of all arguments to a statement as scalar_to_vec operations.
24346 (vectorizable_call): Adjust call to vect_model_simple_cost for
24348 (vectorizable_conversion): Likewise.
24349 (vectorizable_assignment): Likewise.
24350 (vectorizable_shift): Likewise.
24351 (vectorizable_operation): Likewise.
24352 (vectorizable_comparison): Likewise.
24353 (vect_is_simple_cond): Record the def types for operands.
24354 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24355 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24356 for statement argument count.
24358 2017-05-16 Carl Love <cel@us.ibm.com>
24360 * config/rs6000/rs6000-c: Add support for built-in functions
24361 vector unsigned long long vec_bperm (vector unsigned long long,
24362 vector unsigned char)
24363 vector signed long long vec_mule (vector signed int,
24365 vector unsigned long long vec_mule (vector unsigned int,
24366 vector unsigned int)
24367 vector signed long long vec_mulo (vector signed int,
24369 vector unsigned long long vec_mulo (vector unsigned int,
24370 vector unsigned int)
24371 vector signed char vec_sldw (vector signed char,
24372 vector signed char,
24374 vector unsigned char vec_sldw (vector unsigned char,
24375 vector unsigned char,
24377 vector signed short vec_sldw (vector signed short,
24378 vector signed short,
24380 vector unsigned short vec_sldw (vector unsigned short,
24381 vector unsigned short,
24383 vector signed int vec_sldw (vector signed int,
24386 vector unsigned int vec_sldw (vector unsigned int,
24387 vector unsigned int,
24389 vector signed long long vec_sldw (vector signed long long,
24390 vector signed long long,
24392 vector unsigned long long vec_sldw (vector unsigned long long,
24393 vector unsigned long long,
24395 * config/rs6000/rs6000-c: Add support for built-in functions
24396 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24397 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24398 * doc/extend.texi: Update the built-in documentation for the
24399 new built-in functions.
24401 2017-05-16 Marek Polacek <polacek@redhat.com>
24405 * tree.c (save_expr): Don't fold the expression.
24407 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24409 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24410 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24411 and (?*y,m). Update insn attributes.
24413 2017-05-16 Martin Liska <mliska@suse.cz>
24415 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24416 flags argument of print_gimple_stmt, print_gimple_expr,
24417 print_generic_stmt and print_generic_expr.
24418 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24419 * coretypes.h: Likewise.
24420 * except.c (dump_eh_tree): Likewise.
24421 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24422 * gimple-pretty-print.h: Likewise.
24423 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24424 (backprop::push_to_worklist): Likewise.
24425 (backprop::pop_from_worklist): Likewise.
24426 (backprop::process_use): Likewise.
24427 (backprop::intersect_uses): Likewise.
24428 (note_replacement): Likewise.
24429 * gimple-ssa-store-merging.c
24430 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24431 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24432 (pass_store_merging::execute): Likewise.
24433 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24434 (ssa_base_cand_dump_callback): Likewise.
24435 (dump_incr_vec): Likewise.
24436 (replace_refs): Likewise.
24437 (replace_mult_candidate): Likewise.
24438 (create_add_on_incoming_edge): Likewise.
24439 (create_phi_basis): Likewise.
24440 (insert_initializers): Likewise.
24441 (all_phi_incrs_profitable): Likewise.
24442 (introduce_cast_before_cand): Likewise.
24443 (replace_one_candidate): Likewise.
24444 * gimplify.c (gimplify_expr): Likewise.
24445 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24446 (set_rename): Likewise.
24447 (rename_uses): Likewise.
24448 (copy_loop_phi_nodes): Likewise.
24449 (add_close_phis_to_merge_points): Likewise.
24450 (copy_loop_close_phi_args): Likewise.
24451 (copy_cond_phi_args): Likewise.
24452 (graphite_copy_stmts_from_block): Likewise.
24453 (translate_pending_phi_nodes): Likewise.
24454 * graphite-poly.c (print_pdr): Likewise.
24455 (dump_gbb_cases): Likewise.
24456 (dump_gbb_conditions): Likewise.
24457 (print_scop_params): Likewise.
24458 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24459 (build_cross_bb_scalars_use): Likewise.
24460 (gather_bbs::before_dom_children): Likewise.
24461 * hsa-dump.c (dump_hsa_immed): Likewise.
24462 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24463 (get_replacement_map): Likewise.
24464 * ipa-inline-analysis.c (dump_condition): Likewise.
24465 (estimate_function_body_sizes): Likewise.
24466 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24467 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24468 * ipa-prop.c (ipa_dump_param): Likewise.
24469 (ipa_print_node_jump_functions_for_edge): Likewise.
24470 (ipa_modify_call_arguments): Likewise.
24471 (ipa_modify_expr): Likewise.
24472 (ipa_dump_param_adjustments): Likewise.
24473 (ipa_dump_agg_replacement_values): Likewise.
24474 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24475 * ipa-pure-const.c (check_stmt): Likewise.
24476 (pass_nothrow::execute): Likewise.
24477 * ipa-split.c (execute_split_functions): Likewise.
24478 * omp-offload.c (dump_oacc_loop_part): Likewise.
24479 (dump_oacc_loop): Likewise.
24480 * trans-mem.c (tm_log_emit): Likewise.
24481 (tm_memopt_accumulate_memops): Likewise.
24482 (dump_tm_memopt_set): Likewise.
24483 (dump_tm_memopt_transform): Likewise.
24484 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24485 (print_loop): Likewise.
24486 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24487 (chkp_gather_checks_info): Likewise.
24488 (chkp_get_check_result): Likewise.
24489 (chkp_remove_check_if_pass): Likewise.
24490 (chkp_use_outer_bounds_if_possible): Likewise.
24491 (chkp_reduce_bounds_lifetime): Likewise.
24492 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24493 (chkp_mark_completed_bounds): Likewise.
24494 (chkp_register_incomplete_bounds): Likewise.
24495 (chkp_mark_invalid_bounds): Likewise.
24496 (chkp_maybe_copy_and_register_bounds): Likewise.
24497 (chkp_build_returned_bound): Likewise.
24498 (chkp_get_bound_for_parm): Likewise.
24499 (chkp_build_bndldx): Likewise.
24500 (chkp_get_bounds_by_definition): Likewise.
24501 (chkp_generate_extern_var_bounds): Likewise.
24502 (chkp_get_bounds_for_decl_addr): Likewise.
24503 * tree-chrec.c (chrec_apply): Likewise.
24504 * tree-data-ref.c (dump_data_reference): Likewise.
24505 (dump_subscript): Likewise.
24506 (dump_data_dependence_relation): Likewise.
24507 (analyze_overlapping_iterations): Likewise.
24508 * tree-inline.c (expand_call_inline): Likewise.
24509 (tree_function_versioning): Likewise.
24510 * tree-into-ssa.c (dump_defs_stack): Likewise.
24511 (dump_currdefs): Likewise.
24512 (dump_names_replaced_by): Likewise.
24513 (dump_update_ssa): Likewise.
24514 (update_ssa): Likewise.
24515 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24516 * tree-parloops.c (build_new_reduction): Likewise.
24517 (try_create_reduction_list): Likewise.
24518 (ref_conflicts_with_region): Likewise.
24519 (oacc_entry_exit_ok_1): Likewise.
24520 (oacc_entry_exit_single_gang): Likewise.
24521 * tree-pretty-print.h: Likewise.
24522 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24523 (get_scalar_evolution): Likewise.
24524 (add_to_evolution): Likewise.
24525 (get_loop_exit_condition): Likewise.
24526 (analyze_evolution_in_loop): Likewise.
24527 (analyze_initial_condition): Likewise.
24528 (analyze_scalar_evolution): Likewise.
24529 (instantiate_scev): Likewise.
24530 (number_of_latch_executions): Likewise.
24531 (gather_chrec_stats): Likewise.
24532 (final_value_replacement_loop): Likewise.
24533 (scev_const_prop): Likewise.
24534 * tree-sra.c (dump_access): Likewise.
24535 (disqualify_candidate): Likewise.
24536 (create_access): Likewise.
24537 (reject): Likewise.
24538 (maybe_add_sra_candidate): Likewise.
24539 (create_access_replacement): Likewise.
24540 (analyze_access_subtree): Likewise.
24541 (analyze_all_variable_accesses): Likewise.
24542 (sra_modify_assign): Likewise.
24543 (initialize_constant_pool_replacements): Likewise.
24544 (find_param_candidates): Likewise.
24545 (decide_one_param_reduction): Likewise.
24546 (replace_removed_params_ssa_names): Likewise.
24547 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24548 * tree-ssa-copy.c (dump_copy_of): Likewise.
24549 (copy_prop_visit_cond_stmt): Likewise.
24550 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24551 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24552 (record_equivalences_from_stmt): Likewise.
24553 * tree-ssa-dse.c (compute_trims): Likewise.
24554 (delete_dead_call): Likewise.
24555 (delete_dead_assignment): Likewise.
24556 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24557 (forward_propagate_into_cond): Likewise.
24558 (pass_forwprop::execute): Likewise.
24559 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24560 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24562 (move_computations_worker): Likewise.
24563 (execute_sm): Likewise.
24564 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24565 (remove_exits_and_undefined_stmts): Likewise.
24566 (remove_redundant_iv_tests): Likewise.
24567 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24568 (adjust_iv_update_pos): Likewise.
24569 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24570 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24571 (value_replacement): Likewise.
24572 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24573 * tree-ssa-pre.c (print_pre_expr): Likewise.
24574 (get_representative_for): Likewise.
24575 (create_expression_by_pieces): Likewise.
24576 (insert_into_preds_of_block): Likewise.
24577 (eliminate_insert): Likewise.
24578 (eliminate_dom_walker::before_dom_children): Likewise.
24579 (eliminate): Likewise.
24580 (remove_dead_inserted_code): Likewise.
24581 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24582 * tree-ssa-reassoc.c (get_rank): Likewise.
24583 (eliminate_duplicate_pair): Likewise.
24584 (eliminate_plus_minus_pair): Likewise.
24585 (eliminate_not_pairs): Likewise.
24586 (undistribute_ops_list): Likewise.
24587 (eliminate_redundant_comparison): Likewise.
24588 (update_range_test): Likewise.
24589 (optimize_range_tests_var_bound): Likewise.
24590 (optimize_vec_cond_expr): Likewise.
24591 (rewrite_expr_tree): Likewise.
24592 (rewrite_expr_tree_parallel): Likewise.
24593 (linearize_expr): Likewise.
24594 (break_up_subtract): Likewise.
24595 (linearize_expr_tree): Likewise.
24596 (attempt_builtin_powi): Likewise.
24597 (attempt_builtin_copysign): Likewise.
24598 (transform_stmt_to_copy): Likewise.
24599 (transform_stmt_to_multiply): Likewise.
24600 (dump_ops_vector): Likewise.
24601 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24602 (print_scc): Likewise.
24603 (set_ssa_val_to): Likewise.
24604 (visit_reference_op_store): Likewise.
24605 (visit_use): Likewise.
24606 (sccvn_dom_walker::before_dom_children): Likewise.
24607 (run_scc_vn): Likewise.
24608 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24610 (expr_hash_elt::print): Likewise.
24611 (const_and_copies::pop_to_marker): Likewise.
24612 (const_and_copies::record_const_or_copy_raw): Likewise.
24613 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24614 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24615 (dump_predicates): Likewise.
24616 (find_uninit_use): Likewise.
24617 (warn_uninitialized_phi): Likewise.
24618 (pass_late_warn_uninitialized::execute): Likewise.
24619 * tree-ssa.c (verify_vssa): Likewise.
24620 (verify_ssa): Likewise.
24621 (maybe_optimize_var): Likewise.
24622 * tree-vrp.c (dump_value_range): Likewise.
24623 (dump_all_value_ranges): Likewise.
24624 (dump_asserts_for): Likewise.
24625 (register_edge_assert_for_2): Likewise.
24626 (vrp_visit_cond_stmt): Likewise.
24627 (vrp_visit_switch_stmt): Likewise.
24628 (vrp_visit_stmt): Likewise.
24629 (vrp_visit_phi_node): Likewise.
24630 (simplify_cond_using_ranges_1): Likewise.
24631 (fold_predicate_in): Likewise.
24632 (evrp_dom_walker::before_dom_children): Likewise.
24633 (evrp_dom_walker::push_value_range): Likewise.
24634 (evrp_dom_walker::pop_value_range): Likewise.
24635 (execute_early_vrp): Likewise.
24637 2017-05-16 Richard Biener <rguenther@suse.de>
24639 * dwarf2out.c (loc_list_from_tree_1): Do not create
24640 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24642 2017-05-16 Richard Biener <rguenther@suse.de>
24644 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24646 (note_variable_value_in_expr): If we resolved the decl ref
24647 do not push to the stack.
24649 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24651 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24652 operations in fast-math mode.
24653 (vaddq_f16): Likewise.
24654 (vmul_f16): Likewise.
24655 (vmulq_f16): Likewise.
24656 (vsub_f16): Likewise.
24657 (vsubq_f16): Likewise.
24658 * config/arm/neon.md (add<mode>3): New.
24660 (fma:<VH:mode>3): New. Also remove outdated comment.
24663 2017-05-16 Martin Liska <mliska@suse.cz>
24667 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24668 (odr_types_equivalent_p): Likewise.
24670 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24672 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24674 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24677 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24678 non-interunit SSE move alternatives with '?'.
24679 (zero-extendsidi peephole2): New peephole to skip intermediate
24680 general register in SSE zero-extend sequence.
24682 2017-05-15 Jeff Law <law@redhat.com>
24684 * reorg.c (relax_delay_slots): Create a new variable to hold
24685 the temporary target rather than clobbering TARGET_LABEL.
24687 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24688 missing argument to extract_bit_field call.
24689 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24691 2017-05-15 Martin Liska <mliska@suse.cz>
24694 * gcc.c (process_command): Do not allow empty argument of -o option.
24696 2017-05-15 Renlin Li <renlin.li@arm.com>
24698 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24699 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24700 * config/aarch64/constraints.md (Usf): Add long call check.
24701 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24702 (call_value): Likewise.
24703 (sibcall): Likewise.
24704 (sibcall_value): Likewise.
24706 (call_value_insn): New.
24707 (sibcall_insn): Update rtx pattern.
24708 (sibcall_value_insn): Likewise.
24709 (call_internal): Remove.
24710 (call_value_internal): Likewise.
24711 (sibcall_internal): Likewise.
24712 (sibcall_value_internal): Likewise.
24713 (call_reg): Likewise.
24714 (call_symbol): Likewise.
24715 (call_value_reg): Likewise.
24716 (call_value_symbol): Likewise.
24718 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24721 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24723 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24725 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24726 compatible with CCGOCmode and with CCZmode.
24728 2017-05-14 Martin Sebor <msebor@redhat.com>
24730 PR middle-end/77671
24731 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24732 (gimple_fold_builtin_snprintf): Same.
24733 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24734 (gimple_fold_builtin_snprintf): Same.
24735 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24736 of character types.
24737 (is_call_safe): New function.
24738 (try_substitute_return_value): Call it.
24739 (try_simplify_call): New function.
24740 (pass_sprintf_length::handle_gimple_call): Call it.
24742 2017-05-14 Martin Sebor <msebor@redhat.com>
24744 PR middle-end/80669
24745 * builtins.c (expand_builtin_stpncpy): Simplify.
24747 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24749 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24750 * config/i386/i386.h
24751 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24752 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24753 (struct machine_function): Add new members call_ms2sysv,
24754 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24755 (struct machine_frame_state): New fields sp_realigned and
24756 sp_realigned_offset.
24757 * config/i386/i386.c
24758 (enum xlogue_stub): New enum.
24759 (enum xlogue_stub_sets): New enum.
24760 (class xlogue_layout): New class.
24761 (struct ix86_frame): New fields stack_realign_allocate_offset,
24762 stack_realign_offset and outlined_save_offset. Modify comments to
24763 detail stack layout when using out-of-line stubs.
24764 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24765 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24766 -mcall-ms2sysv-xlogues.
24767 (stub_managed_regs): New static variable.
24768 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24769 registers managed by out-of-line stub.
24770 (disable_call_ms2sysv_xlogues): New function.
24771 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24772 m->call_ms2sysv when appropriate and compute frame layout for
24774 (sp_valid_at, fp_valid_at): New inline functions.
24775 (choose_basereg): New function.
24776 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24778 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24779 Use align parameter of choose_baseaddr to generated aligned SSE movs
24781 (pro_epilogue_adjust_stack): Modify to track
24782 machine_frame_state::sp_realigned.
24783 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24784 (ix86_nsaved_sseregs): Likewise.
24785 (ix86_emit_save_regs): Likewise.
24786 (ix86_emit_save_regs_using_mov): Likewise.
24787 (ix86_emit_save_sse_regs_using_mov): Likewise.
24788 (get_scratch_register_on_entry): Likewise.
24789 (gen_frame_set): New function.
24790 (gen_frame_load): Likewise.
24791 (gen_frame_store): Likewise.
24792 (emit_outlined_ms2sysv_save): Likewise.
24793 (emit_outlined_ms2sysv_restore): Likewise.
24794 (ix86_expand_prologue): Modify stack re-alignment code and call
24795 emit_outlined_ms2sysv_save when appropriate.
24796 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24797 parameter rtx_insn *insn, which allows the function to be used to only
24798 generate the notes.
24799 (ix86_expand_epilogue): Modify validity checks of frame and stack
24800 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24801 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24802 * config/i386/predicates.md
24803 (save_multiple): New predicate.
24804 (restore_multiple): Likewise.
24805 * config/i386/sse.md
24806 (save_multiple<mode>): New pattern.
24807 (save_multiple_realign<mode>): Likewise.
24808 (restore_multiple<mode>): Likewise.
24809 (restore_multiple_and_return<mode>): Likewise.
24810 (restore_multiple_leave_return<mode>): Likewise.
24811 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24813 2017-05-14 Julia Koval <julia.koval@intel.com>
24815 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24816 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24817 (__builtin_ia32_xsetbv): New builtins.
24818 * config/i386/i386.c (ix86_expand_special_args_builtin):
24820 (ix86_expand_builtin): Special expand for new intrinsics.
24821 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24822 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24823 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24825 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24827 * cfganal.c (inverted_post_order_compute): Change argument type
24829 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24830 * df-core.c (rest_of_handle_df_initialize): Adjust.
24831 (rest_of_handle_df_finish): Likewise.
24832 (df_analyze_1): Likewise.
24833 (df_analyze): Likewise.
24834 (loop_inverted_post_order_compute): Change argument to be a vec *.
24835 (df_analyze_loop): Adjust.
24836 (df_get_n_blocks): Likewise.
24837 (df_get_postorder): Likewise.
24838 * df.h (struct df_d): Change field to be a vec.
24839 * lcm.c (compute_laterin): Adjust.
24840 (compute_available): Likewise.
24841 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24842 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24843 * tree-ssa-pre.c (compute_antic): Likewise.
24845 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24847 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24848 (depth_first_search::depth_first_search): Change structure init
24849 function to this constructor.
24850 (depth_first_search::add_bb): Rename function to this member.
24851 (depth_first_search::execute): Likewise.
24852 (flow_dfs_compute_reverse_finish): Adjust.
24854 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24856 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24857 (longest_simple_path): Likewise.
24858 * shrink-wrap.c (spread_components): Likewise.
24859 (disqualify_problematic_components): Likewise.
24860 (emit_common_heads_for_components): Likewise.
24861 (emit_common_tails_for_components): Likewise.
24862 (insert_prologue_epilogue_for_components): Likewise.
24864 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24866 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24869 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24871 * df-core.c (df_set_blocks): Start using auto_bitmap.
24872 (df_compact_blocks): Likewise.
24873 * df-problems.c (df_rd_confluence_n): Likewise.
24874 * df-scan.c (df_insn_rescan_all): Likewise.
24875 (df_process_deferred_rescans): Likewise.
24876 (df_update_entry_block_defs): Likewise.
24877 (df_update_exit_block_uses): Likewise.
24878 (df_entry_block_bitmap_verify): Likewise.
24879 (df_exit_block_bitmap_verify): Likewise.
24880 (df_scan_verify): Likewise.
24881 * lra-constraints.c (lra_constraints): Likewise.
24882 (undo_optional_reloads): Likewise.
24883 (lra_undo_inheritance): Likewise.
24884 * lra-remat.c (calculate_gen_cands): Likewise.
24885 (do_remat): Likewise.
24886 * lra-spills.c (assign_spill_hard_regs): Likewise.
24887 (spill_pseudos): Likewise.
24888 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24889 (bitmap_set_subtract_values): Likewise.
24891 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24893 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24894 management with auto_bitmap.
24895 (fix_inter_tick): Likewise.
24896 (fix_recovery_deps): Likewise.
24897 * ira.c (add_store_equivs): Likewise.
24898 (find_moveable_pseudos): Likewise.
24899 (split_live_ranges_for_shrink_wrap): Likewise.
24900 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24901 (rtx_reuse_manager::seen_def_p): Likewise.
24902 (rtx_reuse_manager::set_seen_def): Likewise.
24903 * print-rtl.h (class rtx_reuse_manager): Likewise.
24905 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24907 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24909 (migrate_btr_def): Likewise.
24910 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24911 * df-core.c (loop_post_order_compute): Likewise.
24912 (loop_inverted_post_order_compute): Likewise.
24913 * hsa-common.h: Likewise.
24914 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24915 * init-regs.c (initialize_uninitialized_regs): Likewise.
24916 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24917 (inline_small_functions): Likewise.
24918 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24919 * ira.c (combine_and_move_insns): Likewise.
24920 (build_insn_chain): Likewise.
24921 * loop-invariant.c (find_invariants): Likewise.
24922 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24923 * predict.c (tree_predict_by_opcode): Likewise.
24924 (predict_paths_leading_to): Likewise.
24925 (predict_paths_leading_to_edge): Likewise.
24926 (estimate_loops_at_level): Likewise.
24927 (estimate_loops): Likewise.
24928 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24929 (spread_components): Likewise.
24930 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24931 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24932 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24933 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24934 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24935 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24936 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24937 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24938 (mark_threaded_blocks): Likewise.
24939 (thread_through_all_blocks): Likewise.
24940 * tree-ssa.c (verify_ssa): Likewise.
24941 (execute_update_addresses_taken): Likewise.
24942 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24944 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24946 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24948 (post_order_compute): Likewise.
24949 (inverted_post_order_compute): Likewise.
24950 (pre_and_rev_post_order_compute_fn): Likewise.
24952 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24954 * genrecog.c (int_set::int_set): Explicitly construct our
24955 auto_vec base class.
24956 * vec.h (auto_vec::auto_vec): New constructor.
24958 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24960 * bitmap.h (class auto_bitmap): New constructor taking
24961 bitmap_obstack * argument.
24963 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24965 * bitmap.h (class auto_bitmap): Change type of m_bits to
24966 bitmap_head, and adjust ctor / dtor and member operators.
24968 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24970 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24971 when returned register mode doesn't match original mode.
24973 2017-05-12 Jeff Law <law@redhat.com>
24974 Jakub Jelinek <jakub@redhat.com>
24976 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24977 we look for cc setter after the compare-elim changes.
24978 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24979 within the vector to match what compare-elim now expects.
24980 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24981 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24983 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24984 after the compare-elim changes.
24985 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24986 the vector to match what compare-elim now expects.
24987 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24988 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24989 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24990 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24991 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24993 * config/visium/visium.c (single_set_and_flags): Fix where
24994 we look for cc setter after the compare-elim changes.
24995 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24996 with the vector to match what compare-elim now expects.
24997 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24998 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24999 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25000 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25001 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25002 (neg<mode>2_insn_set_overflow): Likewise.
25004 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25006 PR middle-end/79794
25007 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25008 maybe_expand_insn call, set ops[0].target. If still set after call,
25009 set alt_rtl. Add extra arg to recursive calls.
25010 (extract_bit_field): Add alt_rtl argument. Pass to
25012 * expmed.h (extract_bit_field): Fix prototype.
25013 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25014 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25015 to extract_bit_field_calls.
25016 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25017 Pass alt_rtl to extract_bit_field calls.
25018 * calls.c (store_unaligned_arguments_into_psuedos)
25019 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25020 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25022 * optabs.h (struct expand_operand): Add target bitfield.
25024 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25026 * compare-elim.c (try_eliminate_compare): Canonicalize
25027 operation with embedded compare to
25028 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25029 (set (reg) (operation)].
25031 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25033 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25036 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25037 cost of adding a carry flag for ADC instruction.
25038 [case MINUS]: Ignore the cost of subtracting a carry flag
25039 for SBB instruction.
25041 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25043 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25045 * config/rs6000/bmiintrin.h: New file.
25046 * config/rs6000/bmi2intrin.h: New file.
25047 * config/rs6000/x86intrin.h: New file.
25049 2017-05-12 Jeff Law <law@redhat.com>
25051 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25054 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25056 PR middle-end/80707
25057 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25059 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25061 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25062 early expansion of vector divide builtins.
25063 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25064 builtins identified as having unsigned arguments.
25066 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25068 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25069 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25070 expansion of vector logical operations (and, andc, or, xor,
25073 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25075 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25076 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25078 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25080 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25081 early GIMPLE expansion of vector multiplies.
25083 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25085 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25086 TARGET_HAVE_MOVT conditional.
25087 (movt splitter): Likewise.
25089 2017-05-12 Richard Biener <rguenther@suse.de>
25091 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25092 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25093 Fold all stmts not inplace.
25095 2017-05-12 Richard Biener <rguenther@suse.de>
25097 PR tree-optimization/80713
25098 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25099 inserted_exprs bit for not removed stmts.
25101 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25103 PR middle-end/69921
25104 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25105 parallelized" attribute for parallelized OpenACC kernels.
25106 * omp-offload.c (execute_oacc_device_lower): Use it.
25108 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25109 Set "oacc kernels" attribute.
25110 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25111 parameter. Adjust all users.
25112 (oacc_fn_attrib_kernels_p): Remove function.
25113 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25114 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25115 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25116 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25117 assert "oacc kernels" attribute is set.
25119 2017-05-11 Carl Love <cel@us.ibm.com>
25121 * config/rs6000/rs6000-c: Add support for built-in functions
25122 vector unsigned char vec_popcnt (vector signed char)
25123 vector unsigned char vec_popcnt (vector unsigned char)
25124 vector unsigned short vec_popcnt (vector signed short)
25125 vector unsigned short vec_popcnt (vector unsigned short)
25126 vector unsigned int vec_popcnt (vector signed int)
25127 vector unsigned int vec_popcnt (vector unsigned int)
25128 vector unsigned long long vec_popcnt (vector signed long long)
25129 vector unsigned long long vec_popcnt (vector unsigned long long)
25130 vector signed long long vec_slo (vector signed long long,
25131 vector signed char)
25132 vector signed long long vec_slo (vector signed long long,
25133 vector unsigned char)
25134 vector unsigned long long vec_slo (vector unsigned long long,
25135 vector signed char)
25136 vector unsigned long long vec_slo (vector unsigned long long,
25137 vector unsigned char)
25138 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25139 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25140 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25141 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25142 * doc/extend.texi: Update the built-in documentation file for the
25143 new built-in functions.
25145 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25147 * attribs.h (sorted_attr_string): Move machine independent
25148 functions for target clone support from the i386 port to common
25149 code. Rename ix86_function_versions to common_function_versions.
25150 Rename make_name to make_unique_name.
25151 (common_function_versions): Likewise.
25152 (make_unique_name): Likewise.
25153 (make_dispatcher_decl): Likewise.
25154 (is_function_default_version): Likewise.
25155 * attribs.c (attr_strcmp): Likewise.
25156 (sorted_attr_string): Likewise.
25157 (common_function_versions): Likewise.
25158 (make_unique_name): Likewise.
25159 (make_dispatcher_decl): Likewise.
25160 (is_function_default_version): Likewise.
25161 * config/i386/i386.c (attr_strcmp): Likewise.
25162 (sorted_attr_string): Likewise.
25163 (ix86_function_versions): Likewise.
25164 (make_name): Likewise.
25165 (make_dispatcher_decl): Likewise.
25166 (is_function_default_version): Likewise.
25167 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25169 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25172 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25173 Account for direct move costs for vec_construct of integer
25176 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25179 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25180 (UNSPEC_STX_ATOMIC): Ditto.
25181 (loaddi_via_sse): New insn.
25182 (storedi_via_sse): Ditto.
25183 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25184 Update corresponding peephole2 patterns.
25185 (atomic_storedi_fpu): Ditto.
25187 2017-05-11 Julia Koval <julia.koval@intel.com>
25189 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25190 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25192 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25193 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25194 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25196 2017-05-11 Nathan Sidwell <nathan@acm.org>
25198 * graphite-poly.c: Include dumpfile.h.
25200 * dumpfle.h (dump_function): Declare here ...
25201 * tree-dump.h (dump_function): ... not here.
25202 * dumpfile.c: #include tree-cfg.h.
25203 (dump_function): Move here from ...
25204 * tree-dump.c (dump_function): ... here.
25205 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25206 * graphite-poly.c: Don't include tree-dump.h.
25207 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25208 * print-tree.c: Likewise.
25209 * stor-layout.c: Likewise.
25210 * tree-nested.c: Likewise.
25212 * dumpfile.c (dump_start): Use TDF_FLAGS.
25213 (dump_enable_all): Fix TDF_KIND check thinko.
25215 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25217 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25218 array entries to represent two legal parameterizations of the
25219 overloaded __builtin_cmpb function, as represented by the
25220 P6_OV_BUILTIN_CMPB constant.
25221 (altivec_resolve_overloaded_builtin): Add special case handling
25222 for the __builtin_cmpb function, as represented by the
25223 P6_OV_BUILTIN_CMPB constant.
25224 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25225 (BU_P6_64BIT_2): New macro.
25226 (BU_P6_OVERLOAD_2): New macro
25227 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25228 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25229 (CMPB): Add overload support to represent both 32-bit and 64-bit
25230 compare-bytes function.
25231 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25232 support for TARGET_CMPB.
25233 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25234 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25235 documentation of the __builtin_cmpb overloaded built-in function.
25237 2017-05-11 Richard Biener <rguenther@suse.de>
25239 PR tree-optimization/80705
25240 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25241 bases are not vectorizable.
25243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25245 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25246 when counting register pressure.
25248 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25250 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25251 (iv_ca_more_deps): Renamed to ...
25252 (iv_ca_compare_deps): ... this.
25253 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25255 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25257 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25259 (determine_group_iv_costs): ... here.
25260 (find_inv_vars_cb): Record inv var if it's not recorded before.
25262 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25264 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25265 (get_shiftadd_cost): Ditto.
25267 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25269 * tree-ssa-address.c: Include header file.
25270 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25272 (add_to_parts): Refactor.
25273 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25274 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25277 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25279 PR tree-optimization/53090
25280 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25282 (extract_cond_operands): Detect condition with IV on both sides
25283 and return COMP_IV_EXPR_2.
25284 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25285 (rewrite_use_compare): Simplify by removing call to function
25286 extract_cond_operands.
25288 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25290 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25291 (extract_cond_operands): Detect condition comparing against non-
25292 invariant bound and return appropriate enum value.
25293 (find_interesting_uses_cond): Update use of extract_cond_operands.
25294 Handle its return value accordingly.
25295 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25297 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25299 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25300 nonlinear iv_use computation in loop invariant sensitive way.
25302 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25304 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25305 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25307 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25309 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25310 (dump_cand): Support iv_cand.inv_exprs.
25311 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25313 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25316 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25318 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25320 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25321 as local function. Include necessary header files.
25322 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25326 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25328 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25330 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25331 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25332 RSHIFT_EXPR and BIT_NOT_EXPR.
25334 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25336 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25337 (adjust_setup_cost): New parameter supporting round up adjustment.
25338 (struct address_cost_data): Delete.
25339 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25340 (split_address_cost, ptr_difference_cost): Delete.
25341 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25342 (struct ainc_cost_data): New struct.
25343 (get_address_cost_ainc): New function.
25344 (get_address_cost, get_computation_cost): Reimplement.
25345 (determine_group_iv_cost_address): Record inv_expr for all uses of
25347 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25348 (iv_ca_has_deps): Reimplemented to ...
25349 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25351 (iv_ca_extend): Call iv_ca_more_deps.
25353 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25355 * tree-ssa-address.c (struct mem_address): Move to header file.
25356 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25357 * tree-ssa-address.h (struct mem_address): Move from C file.
25358 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25360 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25362 * tree-affine.h (aff_combination_type): New interface.
25363 (aff_combination_zero_p): Remove static.
25364 (aff_combination_const_p): New interface.
25365 (aff_combination_singleton_var_p): New interfaces.
25367 2017-05-11 Richard Biener <rguenther@suse.de>
25369 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25370 Skip unreachable blocks and destinations.
25371 (eliminate): Move stmt removal and fixup ...
25372 (fini_eliminate): ... here. Skip inserted exprs.
25373 (pass_pre::execute): Move fini_pre after fini_eliminate.
25374 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25375 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25376 PRE to get rid of dead code that has invalid SSA form and
25377 split critical edges again.
25379 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25381 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25383 2017-05-11 Richard Biener <rguenther@suse.de>
25385 * passes.c (execute_function_todo): Verify loops if they are
25386 said to be up-to-date.
25387 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25388 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25390 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25393 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25394 handle calling assemble_external ourself.
25397 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25398 modes with zero size. Enhance comment.
25400 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25402 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25403 built-ins for vec_xl and vec_xst with short and char pointer
25406 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25408 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25409 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25410 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25411 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25412 (_mm_maskz_min_round_ss): New intrinsics.
25413 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25414 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25415 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25416 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25417 (__builtin_ia32_minss_mask_round): New builtins.
25418 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25419 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25420 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25422 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25423 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25425 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25428 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25430 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25431 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25432 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25433 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25434 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25435 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25436 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25437 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25438 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25439 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25440 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25441 (__builtin_ia32_mulss_mask_round): New builtins.
25442 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25443 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25444 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25446 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25447 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25449 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25452 2017-05-10 Julia Koval <julia.koval@intel.com>
25454 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25455 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25456 (_mm256_setr_m128i): New intrinsics.
25458 2017-05-10 Julia Koval <julia.koval@intel.com>
25460 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25461 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25462 (_mm_maskz_rcp14_ss): New intrinsics.
25463 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25464 (__builtin_ia32_rcp14ss_mask): New builtins.
25465 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25467 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25469 PR tree-optimization/51513
25470 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25471 (assert_unreachable_fallthru_edge_p): Use it.
25472 (group_case_labels_stmt): Likewise.
25473 * tree-cfg.h: Prototype it.
25474 * stmt.c: Include cfghooks.h and tree-cfg.h.
25475 (emit_case_dispatch_table) <gap_label>: New local variable.
25476 Use it to fill dispatch table gaps.
25477 Test for default_label before updating probabilities.
25478 (expand_case) <default_label>: Remove unneeded initialization.
25479 Test for unreachable default case statement and remove its edge.
25480 Set default_label accordingly.
25481 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25483 2017-05-10 Carl Love <cel@us.ibm.com>
25485 * config/rs6000/rs6000-c: Add support for built-in functions
25486 vector signed char vec_neg (vector signed char)
25487 vector signed short int vec_neg (vector short int)
25488 vector signed int vec_neg (vector signed int)
25489 vector signed long long vec_neg (vector signed long long)
25490 vector float vec_neg (vector float)
25491 vector double vec_neg (vector double)
25492 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25494 * config/rs6000/altivec.h: Add define for vec_neg
25495 * doc/extend.texi: Update the built-in documentation for the
25496 new built-in functions.
25498 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25500 PR tree-optimization/77644
25501 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25503 2017-05-10 Nathan Sidwell <nathan@acm.org>
25505 * dumpfile.h (TDI_lang_all): New.
25506 (TDF_KIND): New. Renumber others
25507 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25509 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25511 (get_dump_file_name): Adjust suffix generation.
25512 (dump_enable_all): Use TDF_KIND.
25513 * doc/invoke.texi (-fdump-lang-all): Document.
25515 * dumpfile.h: Tabify.
25517 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25520 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25521 Move member access before delete.
25523 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25525 * tree-inline.c (expand_call_inline): Split block at stmt
25528 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25531 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25532 are now unused after splitting mov{sf,sd}_hardfloat.
25533 (f32_lr2): Likewise.
25534 (f32_lm): Likewise.
25535 (f32_lm2): Likewise.
25536 (f32_li): Likewise.
25537 (f32_li2): Likewise.
25538 (f32_lv): Likewise.
25539 (f32_sr): Likewise.
25540 (f32_sr2): Likewise.
25541 (f32_sm): Likewise.
25542 (f32_sm2): Likewise.
25543 (f32_si): Likewise.
25544 (f32_si2): Likewise.
25545 (f32_sv): Likewise.
25546 (f32_dm): Likewise.
25547 (f32_vsx): Likewise.
25548 (f32_av): Likewise.
25549 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25550 For movsf, order stores so the VSX stores occur before the GPR
25551 store which encourages the register allocator to use a traditional
25552 FPR instead of a GPR. For movsd, order the stores so that the GPR
25553 store comes before the VSX stores to allow the power6 to work.
25554 This is due to the power6 not having a 32-bit integer store
25555 instruction from a FPR.
25556 (movsf_hardfloat): Likewise.
25557 (movsd_hardfloat): Likewise.
25559 2017-05-09 Martin Sebor <msebor@redhat.com>
25561 PR translation/80280
25562 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25565 PR translation/80280
25566 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25567 data member added in r247778.
25568 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25570 2017-05-09 Nathan Sidwell <nathan@acm.org>
25572 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25574 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25577 2017-05-09 Marek Polacek <polacek@redhat.com>
25579 * doc/invoke.texi: Fix typo.
25581 2017-05-09 Richard Biener <rguenther@suse.de>
25583 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25584 (vrp_val_is_min): Likewise.
25585 (set_value_range_to_value): Likewise.
25586 (set_value_range_to_nonnegative): Likewise.
25587 (gimple_assign_nonzero_p): Likewise.
25588 (gimple_stmt_nonzero_p): Likewise.
25589 (vrp_int_const_binop): Likewise. Remove unreachable case.
25590 (adjust_range_with_scev): Adjust comments.
25591 (compare_range_with_value): Likewise.
25592 (extract_range_from_phi_node): Likewise.
25593 (test_for_singularity): Likewise.
25595 2017-05-09 Richard Biener <rguenther@suse.de>
25597 * tree-vrp.c (get_single_symbol): Add assert that we don't
25598 get overflowed constants as invariant part.
25599 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25600 checks. Use wi::cmp instead of recursing for integer constants.
25601 (compare_values): Just ignore whether we assumed undefined
25602 overflow instead of failing the compare.
25603 (extract_range_for_var_from_comparison_expr): Add comment before the
25604 TREE_NO_WARNING sets.
25605 (test_for_singularity): Likewise.
25606 (extract_range_from_comparison): Do not disable optimization
25607 when we assumed undefined overflow.
25608 (extract_range_basic): Remove init of unused var.
25610 2017-05-09 Richard Biener <rguenther@suse.de>
25612 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25613 (extract_range_from_multiplicative_op_1): Adjust.
25614 (extract_range_from_binary_expr_1): Use int_const_binop.
25616 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25619 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25620 rs6000_store_data_bypass_p in seven define_bypass directives and
25621 in several comments.
25622 * config/rs6000/rs6000-protos.h: Add prototype for
25623 rs6000_store_data_bypass_p function.
25624 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25625 function implements slightly different (rs6000-specific) semantics
25626 than store_data_bypass_p, returning false rather than aborting
25627 with assertion error when arguments do not satisfy the
25628 requirements of store data bypass.
25629 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25630 rs6000_store_data_bypass_p.
25632 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25634 * config/xtensa/xtensa-protos.h
25635 (xtensa_initial_elimination_offset): New declaration.
25636 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25637 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25638 macro definition, add case for FRAME_POINTER_REGNUM when
25639 FRAME_GROWS_DOWNWARD.
25640 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25641 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25642 xtensa_initial_elimination_offset.
25644 2017-05-08 Nathan Sidwell <nathan@acm.org>
25646 * doc/invoke.texi: Alphabetize -fdump options.
25648 2017-05-08 Martin Sebor <msebor@redhat.com>
25650 PR translation/80280
25651 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25653 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25655 * target.def (compute_frame_layout): New optional target hook.
25656 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25657 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25658 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25660 * reload1.c (verify_initial_elim_offsets): Likewise.
25661 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25662 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25663 (arm_get_frame_offsets): Split up into this ...
25664 (arm_compute_frame_layout): ... and this function.
25666 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25668 * config/aarch64/constraints.md (Usa): New constraint.
25669 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25671 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25673 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25674 with_multilib_list after it has been checked.
25676 2017-05-08 Richard Biener <rguenther@suse.de>
25678 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25679 (bitmap_set_subtract_values): Likewise.
25681 2017-05-08 Richard Biener <rguenther@suse.de>
25683 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25684 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25686 (gimple_stmt_nonzero_warnv_p): Rename to ...
25687 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25689 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25690 (extract_range_basic): Adjust, do not disable propagation on
25691 strict overflow sensitive simplification.
25692 (vrp_visit_cond_stmt): Likewise.
25694 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25696 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25697 body size unconditionally.
25699 2017-05-07 Jeff Law <law@redhat.com>
25702 2017-05-06 Jeff Law <law@redhat.com>
25703 PR tree-optimization/78496
25704 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25707 PR tree-optimization/78496
25708 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25709 (simplify_stmt_using_ranges): Call it.
25710 (vrp_dom_walker::before_dom_children): Extract equivalences
25711 from an ASSERT_EXPR with an equality comparison against a
25714 2017-05-06 Jeff Law <law@redhat.com>
25716 PR tree-optimization/78496
25717 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25720 PR tree-optimization/78496
25721 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25722 (simplify_stmt_using_ranges): Call it.
25723 (vrp_dom_walker::before_dom_children): Extract equivalences
25724 from an ASSERT_EXPR with an equality comparison against a
25727 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25729 * lra-constraints.c (lra_copy_reg_equiv): New function.
25730 (split_reg): Use it to copy equivalence information from the
25731 original register to the spill register.
25733 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25735 PR rtl-optimization/75964
25736 * simplify-rtx.c (simplify_const_relational_operation): Remove
25737 invalid handling of comparisons of integer ABS.
25739 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25741 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25742 initialize to zero.
25743 (init_regs): Remove declaration.
25744 (function_arg_advance_32): Initialize error_p as boolean variable.
25746 2017-05-05 Nathan Sidwell <nathan@acm.org>
25748 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25749 lines. Use for (;;).
25751 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25753 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25754 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25755 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25756 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25757 VF=2 that require versioning.
25759 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25761 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25764 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25766 * diagnostic.h (diagnostic_override_option_index): Convert from
25767 macro to inline function.
25769 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25771 * diagnostic.c (last_module_changed_p): New function.
25772 (set_last_module): New function.
25773 (diagnostic_report_current_module): Convert macro usage to
25774 the above functions.
25775 * diagnostic.h (diagnostic_context::last_module): Strengthen
25776 from const line_map * to const line_map_ordinary *.
25777 (diagnostic_last_module_changed): Delete macro.
25778 (diagnostic_set_last_module): Delete macro.
25780 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25782 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25783 with diagnostic_report_diagnostic.
25784 (diagnostic_n_impl_richloc): Likewise.
25785 * diagnostic.h (report_diagnostic): Delete macro.
25786 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25787 with diagnostic_report_diagnostic.
25788 * substring-locations.c (format_warning_va): Likewise.
25790 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25792 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25793 save/restor of format_spec. Move option-printing code to...
25794 (print_option_information): ...this new function, and
25795 reimplement by simply printing to the pretty_printer,
25796 rather than appending to the format string.
25798 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25800 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25801 handling logic into...
25802 (update_effective_level_from_pragmas): ...this new function.
25804 2017-05-04 Andrew Waterman <andrew@sifive.com>
25806 * config/riscv/riscv.opt (mstrict-align): New option.
25807 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25808 (SLOW_UNALIGNED_ACCESS): Define.
25809 (riscv_slow_unaligned_access): Declare.
25810 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25812 (riscv_slow_unaligned_access): New variable.
25813 (rocket_tune_info): Set slow_unaligned_access to true.
25814 (optimize_size_tune_info): Set slow_unaligned_access to false.
25815 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25816 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25817 (riscv_option_override): Set riscv_slow_unaligned_access.
25818 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25820 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25822 * config/riscv/riscv.md: Unify indentation.
25824 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25829 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25831 (extendsi<mode>2): Add support for doing sign extension via
25832 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25833 don't have ISA 3.0 instructions.
25834 (extendsi<mode>2 splitter): Likewise.
25835 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25836 generate the normal insns since SImode can now go in vector
25837 registers. Disallow the special UNSPECs needed for previous
25838 machines to hide SImode being used. Add new insns
25839 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25840 (fix_trunc<mode>si2_stfiwx): Likewise.
25841 (fix_trunc<mode>si2_internal): Likewise.
25842 (fixuns_trunc<mode>si2): Likewise.
25843 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25844 (fctiw<u>z_<mode>_smallint): Likewise.
25845 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25846 of floating point to 32-bit integer from doing a direct move to
25847 the GPR registers to do a store.
25848 (fctiwz_<mode>): Break long line.
25850 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25852 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25853 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25854 (addr_list, addr_offset_valid_p): New.
25855 (split_address_groups): Check offset validity with above function.
25856 (gt-tree-ssa-loop-ivopts.h): Include header file.
25858 2017-05-05 Nathan Sidwell <nathan@acm.org>
25860 * config.gcc (arm*-*-*): Add missing 'fi'.
25862 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25864 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25865 included in -fopt-info does not matter.
25866 * doc/optinfo.texi (-fopt-info): Fix description of default
25867 behavour. Explicitly say order of options included in -fopt-info
25870 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25872 * config.gcc: Allow combinations of aprofile and rmprofile values for
25873 --with-multilib-list.
25874 * config/arm/t-multilib: New file.
25875 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25876 variables. Remove setting of ISA and floating-point ABI in
25877 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25878 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25879 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25880 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25882 * config/arm/t-rmprofile: Likewise except for the matches changes.
25883 * doc/install.texi (--with-multilib-list): Document the combination of
25884 aprofile and rmprofile values and warn about pitfalls in doing that.
25886 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25888 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25889 (movdi_aarch64): Likewise.
25891 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25893 PR tree-optimization/80632
25894 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25896 (build_arrays): Initialize it for virtual phis.
25897 (fix_phi_nodes): Use it for virtual phis.
25899 PR tree-optimization/80558
25900 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25901 [x, y] op z into [x op, y op z] for op & or | if conditions
25904 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25905 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25908 * config/arm/arm.md (use_literal_pool): Remove.
25909 (64-bit immediate split): No longer takes cost into consideration
25910 if arm_disable_literal_pool is enabled.
25911 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25912 used when arm_disable_literal_pool is enabled.
25913 (arm_max_const_double_inline_cost): Remove use of
25914 arm_disable_literal_pool.
25915 (push_minipool_fix): Add assert.
25916 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25917 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25918 (no_literal_pool_sf_immediate): New.
25920 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25922 PR tree-optimization/80613
25923 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25924 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25926 2017-05-05 Richard Biener <rguenther@suse.de>
25928 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25930 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25932 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25933 of this flag from insn conditions due to removal from r247495.
25935 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25937 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25939 (arm_early_store_addr_dep_ptr): Likewise.
25940 * config/arm/aarch-common-protos.h
25941 (arm_early_load_addr_dep_ptr): Add prototype.
25942 (arm_early_store_addr_dep_ptr): Likewise.
25943 * config/arm/cortex-a53.md: Add new bypasses.
25945 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25947 * tree.c (next_type_uid): Change type to unsigned.
25948 (type_hash_canon): Decrement back next_type_uid if
25949 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25950 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25953 2017-05-04 Martin Sebor <msebor@redhat.com>
25955 * builtins.c: Fix a trivial typo in a comment.
25957 PR middle-end/79234
25958 * builtins.c (check_sizes): Adjust to handle reading past the end.
25959 Avoid printing excessive upper bound of ranges. Use %E to print
25960 tree nodes instead of converting them to %wu.
25961 (expand_builtin_memchr): New function.
25962 (compute_dest_size): Rename...
25963 (compute_objsize): ...to this.
25964 (expand_builtin_memcpy): Adjust.
25965 (expand_builtin_mempcpy): Adjust.
25966 (expand_builtin_strcat): Adjust.
25967 (expand_builtin_strcpy): Adjust.
25968 (check_strncat_sizes): Adjust.
25969 (expand_builtin_strncat): Adjust.
25970 (expand_builtin_strncpy): Adjust and simplify.
25971 (expand_builtin_memset): Adjust.
25972 (expand_builtin_bzero): Adjust.
25973 (expand_builtin_memcmp): Adjust.
25974 (expand_builtin): Handle memcmp.
25975 (maybe_emit_chk_warning): Check strncat just once.
25977 2017-05-04 Martin Sebor <msebor@redhat.com>
25979 PR preprocessor/79214
25980 PR middle-end/79222
25981 PR middle-end/79223
25982 * builtins.c (check_sizes): Add inlining context and issue
25983 warnings even when -Wno-system-headers is set.
25984 (check_strncat_sizes): Same.
25985 (expand_builtin_strncat): Same.
25986 (expand_builtin_memmove): New function.
25987 (expand_builtin_stpncpy): Same.
25988 (expand_builtin): Handle memmove and stpncpy.
25990 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25992 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25993 which is not used any more.
25995 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25997 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25999 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26001 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26002 (cortexa53_tunings): Likewise.
26003 (cortexa57_tunings): Likewise.
26004 (cortexa72_tunings): Likewise.
26005 (cortexa73_tunings): Likewise.
26007 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26009 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26010 Set loop alignment to 8.
26012 2017-05-04 Martin Sebor <msebor@redhat.com>
26014 PR translation/80280
26015 * builtins.c (expand_builtin_object_size): Add missing quoting to
26016 %D and like directives.
26017 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26018 (hsa_type_for_tree_type): Same.
26019 (verify_function_arguments): Same.
26020 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26021 * varasm.c (get_section): Same.
26024 2017-05-04 Martin Sebor <msebor@redhat.com>
26026 PR translation/80280
26027 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26029 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26031 * config/aarch64/aarch64.c (generic_addrcost_table):
26032 Change HI/TI mode setting.
26034 2017-05-04 Martin Jambor <mjambor@suse.cz>
26036 PR tree-optimization/80622
26037 * tree-sra.c (comes_initialized_p): New function.
26038 (build_accesses_from_assign): Only set write lazily when
26039 comes_initialized_p is false.
26040 (analyze_access_subtree): Use comes_initialized_p.
26041 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26042 instead of testing for PARM_DECL.
26044 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26046 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26047 constraint on operand 0 to allow more general addressing modes.
26048 Adjust output template.
26049 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26051 * config/aarch64/aarch64-protos.h
26052 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26053 * config/aarch64/constraints.md (Dp): New address constraint.
26054 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26057 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26059 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26060 update use of estimate_ipcp_clone_size_and_time.
26061 (estimate_local_effects): Update use of
26062 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26063 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26064 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26065 Return nonspecialized time.
26067 2017-05-04 Richard Biener <rguenther@suse.de>
26069 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26070 for the last VUSE which def dominates the PHI. Directly call
26072 (get_continuation_for_phi_1): Remove.
26074 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26076 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26077 to explain the use of truncating division. Cap the number of
26078 iterations to the maximum given by nb_iterations_upper_bound,
26081 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26083 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26084 * configure: Regenerate.
26085 * config.in: Regenerate.
26086 * config/i386/driver-mingw32.c: new file.
26087 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26088 * config.host: Link driver-mingw32.o on MinGW host.
26089 * doc/install.texi: Document new --enable-mingw-wildcard configure
26092 2017-05-04 Marek Polacek <polacek@redhat.com>
26094 PR tree-optimization/80612
26095 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26097 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26098 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26100 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26101 (movt splitter): Likewise.
26102 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26103 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26104 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26105 block for Thumb-1 with MOVT.
26106 (thumb2_legitimate_address_p): Move code block ...
26107 (can_avoid_literal_pool_for_label_p): ... into this new function.
26108 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26110 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26111 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26112 "M-profile targets with the MOVT instruction".
26114 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26116 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26117 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26118 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26120 2017-05-04 Martin Liska <mliska@suse.cz>
26122 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26123 variable cond_code.
26125 2017-05-04 Richard Biener <rguenther@suse.de>
26127 * tree.c (array_at_struct_end_p): Handle arrays at struct
26128 end with flexarrays more conservatively. Refactor and treat
26129 arrays of arrays or aggregates more strict. Fix
26130 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26131 * tree.c (array_at_struct_end_p): Adjust prototype.
26132 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26133 * gimple-fold.c (get_range_strlen): Likewise.
26134 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26136 2017-05-04 Richard Biener <rguenther@suse.de>
26138 PR tree-optimization/31130
26139 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26141 (supports_overflow_infinity): Likewise.
26142 (is_negative_overflow_infinity): Likewise.
26143 (is_positive_overflow_infinity): Likewise.
26144 (is_overflow_infinity): Likewise.
26145 (stmt_overflow_infinity): Likewise.
26146 (overflow_infinity_range_p): Likewise.
26147 (usable_range_p): Remove as always returning true.
26148 (make_overflow_infinity): Remove.
26149 (negative_overflow_infinity): Likewise.
26150 (positive_overflow_infinity): Likewise.
26151 (avoid_overflow_infinity): Likewise.
26152 (set_value_range): Adjust accordingly.
26153 (set_value_range_to_nonnegative): Likewise, remove now unused
26154 overflow_infinity arg.
26155 (vrp_operand_equal_p): Adjust.
26156 (update_value_range): Likewise.
26157 (range_int_cst_singleton_p): Likewise.
26158 (operand_less_p): Likewise.
26159 (compare_values_warnv): Likewise.
26160 (extract_range_for_var_from_comparison_expr): Likewise.
26161 (vrp_int_const_binop): Likewise.
26162 (zero_nonzero_bits_from_vr): Likewise.
26163 (extract_range_from_multiplicative_op_1): Likewise.
26164 (extract_range_from_binary_expr_1): Likewise.
26165 (extract_range_from_unary_expr): Likewise.
26166 (extract_range_from_comparison): Likewise.
26167 (extract_range_basic): Likewise.
26168 (adjust_range_with_scev): Likewise.
26169 (compare_ranges): Likewise.
26170 (compare_range_with_value): Likewise.
26171 (dump_value_range): Likewise.
26172 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26174 (simplify_cond_using_ranges): Adjust.
26176 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26178 * brig-builtins.def: Added a builtin for class_f64.
26179 * builtin-types.def: Added a builtin type needed by class_f64.
26181 2017-05-03 Jason Merrill <jason@redhat.com>
26183 * timevar.def: Add TV_CONSTEXPR.
26185 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26187 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26189 2017-05-03 Martin Jambor <mjambor@suse.cz>
26191 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26192 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26193 * ipa-cp.c (ipcp_propagate_stage): Do not call
26194 ipa_update_after_lto_read.
26195 * ipa-inline.c (ipa_inline): Likewise.
26197 2017-05-03 Martin Jambor <mjambor@suse.cz>
26199 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26200 tag. Added a default constructor and a destructor.
26201 (ipa_edge_args_sum_t): New class;
26202 (ipa_edge_args_sum): Declare.
26203 (ipa_edge_args_vector): Remove declaration.
26204 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26205 (ipa_free_edge_args_substructures): Remove declaration.
26206 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26207 (ipa_edge_args_info_available_for_edge_p): Likewise.
26208 * ipa-prop.c (ipa_edge_args_vector): Removed.
26209 (edge_removal_hook_holder): Likewise.
26210 (edge_duplication_hook_holder): Likewise.
26211 (ipa_edge_args_sum): New variable.
26212 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26213 ipa_edge_args_vector.
26214 (ipa_free_edge_args_substructures): Likewise.
26215 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26216 ipa_edge_args_vector.
26217 (ipa_edge_removal_hook): Turned into method
26218 ipa_edge_args_sum_t::remove.
26219 (ipa_edge_duplication_hook): Turned into method
26220 ipa_edge_args_sum_t::duplicate.
26221 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26222 registering edge hooks.
26223 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26224 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26225 ipa_edge_args_sum instead of ipa_edge_args_vector.
26226 * ipa-profile.c (ipa_profile): Likewise.
26228 2017-05-03 Martin Jambor <mjambor@suse.cz>
26230 * symbol-summary.h (function_summary): New method exists.
26231 (function_summary::symtab_removal): Deallocate through release.
26232 (call_summary): New class.
26233 (gt_ggc_mx): New overload.
26234 (gt_pch_nx): Likewise.
26235 (gt_pch_nx): Likewise.
26237 2017-05-03 Jeff Law <law@redhat.com>
26239 PR tree-optimization/78496
26240 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26241 from simplify_cond_using_ranges. Split off code to walk
26242 backwards through casts into ...
26243 (simplify_cond_using_ranges_2): New function.
26244 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26245 (execute_vrp): After identifying jump threads, call
26246 simplify_cond_using_ranges_2.
26248 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26251 * ipa-inline.h (inline_summary): Add ctor.
26252 (create_ggc): Do not use ggc_cleared_alloc.
26254 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26255 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26257 * gcc.c (handle_braces): Support escaping in switch matching
26259 * doc/invoke.texi (Spec Files): Document it.
26260 Remove superfluous @code markup in items.
26262 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26264 * diagnostic-show-locus.c (struct column_range): New struct.
26265 (get_affected_columns): New function.
26266 (get_printed_columns): New function.
26267 (struct correction): New struct.
26268 (correction::ensure_capacity): New function.
26269 (correction::ensure_terminated): New function.
26270 (struct line_corrections): New struct.
26271 (line_corrections::~line_corrections): New dtor.
26272 (line_corrections::add_hint): New function.
26273 (layout::print_trailing_fixits): Reimplement in terms of the new
26275 (selftest::test_overlapped_fixit_printing): New function.
26276 (selftest::diagnostic_show_locus_c_tests): Call it.
26278 2017-05-03 Nathan Sidwell <nathan@acm.org>
26280 Canonicalize canonical type hashing
26281 * tree.h (type_hash_canon_hash): Declare.
26282 * tree.c (type_hash_list, attribute_hash_list): Move into
26283 type_hash_canon_hash.
26284 (build_type_attribute_qual_variant): Break out hash code calc into
26285 type_hash_canon_hash.
26286 (type_hash_canon_hash): New. Generic type hash computation.
26287 (build_range_type_1, build_array_type_1, build_function_type,
26288 build_method_type_directly, build_offset_type, build_complex_type,
26289 make_vector_type): Call it.
26291 2017-05-03 Richard Biener <rguenther@suse.de>
26293 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26294 When all DRs have unknown misaligned do not always peel
26295 when there is a store but apply the same costing model as if
26296 there were only loads.
26298 2017-05-03 Richard Biener <rguenther@suse.de>
26301 PR tree-optimization/80492
26302 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26303 compare_base_decls returning dont-know properly.
26305 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26307 * config/arm/iterators.md (CCSI): New mode iterator.
26308 (arch): New mode attribute.
26309 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26310 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26311 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26312 code iterator for success result mode.
26313 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26314 the corresponding new insn generators.
26316 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26319 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26320 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26322 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26324 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26325 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26326 (DDR_A): Wrap DDR argument in brackets.
26327 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26328 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26329 (DDR_REVERSED_P): Likewise.
26331 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26333 PR tree-optimization/79472
26334 * tree-switch-conversion.c (struct switch_conv_info): Add
26335 contiguous_range and default_case_nonstandard fields.
26336 (collect_switch_conv_info): Compute contiguous_range and
26337 default_case_nonstandard fields, don't clear final_bb if
26338 contiguous_range and only the default case doesn't have the required
26340 (check_all_empty_except_final): Set default_case_nonstandard instead
26341 of failing if contiguous_range and the default case doesn't have empty
26343 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26344 and only the default case doesn't have the required constants. Skip
26346 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26347 if default_case_nonstandard.
26348 (build_constructors): Build constant 1 just once. Assert that default
26349 values aren't inserted in between cases if contiguous_range. Skip
26351 (build_arrays): Skip virtual phis.
26352 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26353 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26354 Handle virtual phis.
26355 (gen_inbound_check): Handle default_case_nonstandard case.
26356 (process_switch): Adjust check_final_bb caller. Call
26357 gather_default_values with the first non-default case instead of
26358 default case if default_case_nonstandard.
26360 2017-05-02 Nathan Sidwell <nathan@acm.org>
26362 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26363 check. Fix formatting.
26365 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26367 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26368 errors when comparing specialized and unspecialized times.
26370 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26372 * diagnostic-show-locus.c
26373 (layout::should_print_annotation_line_p): Make private.
26374 (layout::print_annotation_line): Make private.
26375 (layout::annotation_line_showed_range_p): Make private.
26376 (layout::show_ruler): Make private.
26377 (layout::print_source_line): Make private. Pass in line and
26378 line_width, rather than calling location_get_source_line. Drop
26380 (layout::print_leading_fixits): New method.
26381 (layout::print_any_fixits): Rename to...
26382 (layout::print_trailing_fixits): ...this, and make private.
26383 Don't print newline fixits.
26384 (diagnostic_show_locus): Move logic for printing one row into...
26385 (layout::print_line): ...this new function. Move the
26386 location_get_source_line call and error-handling from
26387 print_source_line to here. Call print_leading_fixits, and rename
26388 print_any_fixits to print_trailing_fixits.
26389 (selftest::test_fixit_insert_containing_newline): Update now that
26390 newlines are partially supported.
26391 (selftest::test_fixit_insert_containing_newline_2): New test.
26392 (selftest::test_fixit_replace_containing_newline): Update comments.
26393 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26394 * edit-context.c (class added_line): New class.
26395 (class edited_line): Describe newline handling in comment.
26396 (edited_line::actually_edited_p): New method.
26397 (edited_line::print_content): Delete redundant decl.
26398 (edited_line::m_predecessors): New field.
26399 (edited_file::print_content): Call edited_line::print_content.
26400 (edited_file::print_diff): Update to support newlines.
26401 (edited_file::print_diff_hunk): Likewise.
26402 (edited_file::print_run_of_changed_lines): New function.
26403 (edited_file::print_diff_line): Convert to...
26404 (print_diff_line): ...this.
26405 (edited_file::get_effective_line_count): New function.
26406 (edited_line::edited_line): Initialize new field m_predecessors.
26407 (edited_line::~edited_line): Clean up m_predecessors.
26408 (edited_line::apply_fixit): Handle newlines.
26409 (edited_line::get_effective_line_count): New function.
26410 (edited_line::print_content): New function.
26411 (edited_line::print_diff_lines): New function.
26412 (selftest::test_applying_fixits_insert_containing_newline): New
26414 (selftest::test_applying_fixits_replace_containing_newline): New
26416 (selftest::insert_line): New function.
26417 (selftest::test_applying_fixits_multiple_lines): Add example of
26419 (selftest::edit_context_c_tests): Call the new tests.
26421 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26423 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26424 parameter cand. Update dump information.
26425 (get_computation_cost): Update uses.
26427 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26429 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26430 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26431 (get_computation_at, rewrite_use_address): Update use of
26432 get_computation_aff.
26434 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26436 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26437 (get_computation): Delete.
26438 (get_computation_cost): Implement like get_computation_cost_at.
26439 Use get_computation_at.
26440 (get_computation_cost_at): Delete.
26441 (rewrite_use_nonlinear_expr): Use get_computation_at.
26442 (rewrite_use_compare, remove_unused_ivs): Ditto.
26444 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26446 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26448 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26450 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26451 (ivopts_global_cost_for_size): Rename parameter and update uses.
26452 (iv_ca_recount_cost): Update uses.
26453 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26454 candidates seperately in n_invs and n_cands.
26455 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26457 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26459 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26460 (find_inv_vars_cb): New.
26461 (find_depends): Renamed to ...
26462 (find_inv_vars): ... this.
26463 (add_candidate_1, force_var_cost): Call find_inv_vars.
26464 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26466 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26468 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26469 inv_vars. Add inv_exprs.
26470 (struct iv_cand): Rename depends_on to inv_vars.
26471 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26472 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26473 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26474 (dump_cand): Dump inv_vars.
26475 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26476 (record_invariant, find_depends, add_candidate_1): Ditto.
26477 (set_group_iv_cost, force_var_cost): Ditto.
26478 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26479 (get_computation_cost_at, get_computation_cost): Ditto.
26480 (determine_group_iv_cost_generic): Ditto.
26481 (determine_group_iv_cost_address): Ditto.
26482 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26483 (determine_group_iv_costs): Ditto.
26484 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26485 (iv_ca_set_remove_invariants): Renamed to ...
26486 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26487 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26488 (iv_ca_set_add_invariants): Renamed to ...
26489 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26490 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26491 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26492 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26493 (create_new_ivs): Remove useless dump.
26495 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26497 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26499 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26500 (iv_ca_set_no_cp, create_new_iv): Ditto.
26502 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26504 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26506 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26508 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26509 function tree_check2.
26511 2017-05-02 Martin Liska <mliska@suse.cz>
26513 * doc/gcov.texi: Add missing preposition.
26514 * gcov.c (function_info::function_info): Properly fill up
26515 all member variables.
26517 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26519 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26521 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26523 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26525 2017-05-02 Martin Liska <mliska@suse.cz>
26528 * lto-streamer-in.c (lto_read_tree_1): Remove
26529 LTO_STREAMER_DEBUG.
26530 * lto-streamer.c (struct tree_hash_entry): Likewise.
26531 (struct tree_entry_hasher): Likewise.
26532 (tree_entry_hasher::hash): Likewise.
26533 (tree_entry_hasher::equal): Likewise.
26534 (lto_streamer_init): Likewise.
26535 (lto_orig_address_map): Likewise.
26536 (lto_orig_address_get): Likewise.
26537 (lto_orig_address_remove): Likewise.
26538 * lto-streamer.h: Likewise.
26539 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26540 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26542 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26544 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26545 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26546 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26547 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26548 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26549 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26550 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26551 (mm_maskz_sub_ss): New intrinsics.
26552 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26553 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26554 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26555 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26556 (__builtin_ia32_subss_mask_round): New builtins.
26557 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26558 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26559 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26561 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26562 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26564 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26567 2017-05-02 Martin Jambor <mjambor@suse.cz>
26569 PR tree-optimization/78687
26570 * tree-sra.c (access): New field parent.
26571 (process_subtree_disqualification): New function.
26572 (disqualify_candidate): Call it.
26573 (build_accesses_from_assign): Reset write flag if creating an
26575 (build_access_subtree): Fill in parent field and also prpagate
26576 down grp_write flag.
26577 (create_artificial_child_access): New parameter set_grp_write, set
26578 grp_write to its value.
26579 (propagate_subaccesses_across_link): Also propagate grp_write flag
26581 (propagate_all_subaccesses): Push the closest parent back to work
26582 queue if add_access_to_work_queue returned true.
26584 2017-05-02 Richard Biener <rguenther@suse.de>
26586 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26587 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26588 -fstrict-overflow documentation.
26589 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26590 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26591 flag_strict_overflow.
26592 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26593 * lto-opts.c (lto_write_options): Do not stream it.
26594 * lto-wrapper.c (merge_and_complain): Do not handle it.
26595 * opts.c (default_options_table): Do not set -fstrict-overflow.
26596 (finish_options): Likewise do not clear it when sanitizing.
26597 * simplify-rtx.c (simplify_const_relational_operation): Do not
26598 test flag_strict_overflow.
26600 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26602 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26603 using enabled attribute.
26604 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26605 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26606 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26607 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26608 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26609 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26610 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26611 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26612 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26613 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26615 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26617 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26619 2017-05-02 Richard Biener <rguenther@suse.de>
26621 PR tree-optimization/80591
26623 2017-04-10 Richard Biener <rguenther@suse.de>
26625 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26628 2017-05-02 Richard Biener <rguenther@suse.de>
26630 PR tree-optimization/80549
26631 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26632 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26633 headers if they do not have a preheader.
26635 2017-05-02 Martin Liska <mliska@suse.cz>
26638 * common.opt: Fix typo.
26639 * doc/invoke.texi: Likewise.
26641 2017-05-01 Jan Beulich <jbeulich@suse.com>
26643 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26644 swapping, add (x,x,m,x,n) alternative.
26646 2017-05-01 Nathan Sidwell <nathan@acm.org>
26648 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26649 unnecessary unadjusted_alignment check.
26651 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26654 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26656 * gimplify.c (gimplify_cilk_detach): New function.
26657 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26658 * tree-core.h: Document EXPR_CILK_SPAWN.
26659 * tree.h (EXPR_CILK_SPAWN): Define.
26661 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26663 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26664 to use new fixit_hint representation, using the "replace" logic.
26665 (get_line_span_for_fixit_hint): Likewise.
26666 (layout::print_any_fixits): Likewise.
26667 (selftest::test_one_liner_many_fixits): Rename to...
26668 (selftest::test_one_liner_many_fixits_1): ...this, and update
26669 comment and expected output to reflect that the multiple fix-it
26670 hints are now consolidated into one insertion.
26671 (selftest::test_one_liner_many_fixits_2): New test.
26672 (selftest::test_diagnostic_show_locus_one_liner): Update for
26674 (selftest::test_fixit_consolidation): Update for fix-it API
26676 * diagnostic.c (print_parseable_fixits): Likewise.
26677 * edit-context.c (edited_line::m_line_events): Convert from
26678 auto_vec <line_event *> to auto_vec <line_event>.
26679 (class line_event): Convert from abstract base class to a concrete
26680 class, taking over the role of replace_event.
26681 (class insert_event): Delete.
26682 (class replace_event): Rename to class line_event. Convert to
26684 (edit_context::add_fixits): Reimplement.
26685 (edit_context::apply_insert): Delete.
26686 (edit_context::apply_replace): Rename to...
26687 (edit_context::apply_fixit): ...this. Convert to half-open range.
26688 (edited_file::apply_insert): Delete.
26689 (edited_file::apply_replace): Rename to...
26690 (edited_file::apply_fixit): ...this.
26691 (edited_line::~edited_line): Drop deletion of events.
26692 (edited_line::apply_insert): Delete.
26693 (edited_line::apply_replace): Rename to...
26694 (edited_line::apply_fixit): ...this. Convert to half-open range.
26695 Update for change to type of m_line_events.
26696 * edit-context.h (edit_context::apply_insert): Delete.
26697 (edit_context::apply_replace): Rename to...
26698 (edit_context::apply_fixit): ...this.
26700 2017-05-01 Martin Sebor <msebor@redhat.com>
26702 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26705 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26708 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26709 __get_cpuid_max returns 0.
26710 (__get_cpuid_count): Ditto.
26712 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26714 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26715 replacement expression is another instance of one of its arguments.
26717 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26720 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26721 check for stack push/pop autoinc.
26722 * config/i386/i386.c (ix86_agi_dependent): Return false
26723 if the only reason why modified_in_p returned true is that
26724 addr is SP based and set_insn is a push or pop.
26726 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26728 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26731 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26734 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26735 (account_size_time): Use two predicates - exec_pred and
26737 (evaluate_conditions_for_known_args): Compute both clause and
26739 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26740 (inline_summary_t::duplicate): Update.
26741 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26743 (compute_inline_parameters): Likewise.
26744 (estimate_edge_size_and_time): Update caluclation of time.
26745 (estimate_node_size_and_time): Compute both time and nonspecialized
26747 (estimate_ipcp_clone_size_and_time): Update.
26748 (inline_merge_summary): Update.
26749 (do_estimate_edge_time): Update.
26750 (do_estimate_edge_size): Update.
26751 (do_estimate_edge_hints): Update.
26752 (inline_read_section, inline_write_summary): Stream both new predicates.
26753 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26755 (compute_inlined_call_time): Cleanup.
26756 (big_speedup_p): Update.
26757 (edge_badness): Update.
26758 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26759 (size_time_entry): Replace predicate by exec_predicate and
26760 nonconst_predicate.
26761 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26762 (estimate_edge_time): Return also nonspec_time.
26763 (reset_edge_growth_cache): Update.
26765 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26767 PR rtl-optimization/80491
26768 * ifcvt.c (noce_process_if_block): When looking for x setter
26769 with missing else_bb, don't check only the insn right before
26770 cond_earliest, but look for the last insn that x is modified in
26771 within the same bb.
26773 PR rtl-optimization/80491
26774 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26776 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26778 PR tree-optimization/80487
26779 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26781 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26783 PR tree-optimization/79697
26784 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26785 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26786 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26788 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26789 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26791 2017-04-28 Martin Sebor <msebor@redhat.com>
26793 PR tree-optimization/80523
26794 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26795 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26797 (maybe_warn, format_directive, parse_directive): Use new functions.
26798 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26800 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26802 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26804 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26806 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26807 target_header_dir): Set correctly.
26808 * configure: Regenerated.
26809 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26810 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26811 instead of SYSTEM_HEADER_DIR.
26813 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26815 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26816 (estimate_local_effects): Likewise.
26817 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26818 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26819 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26820 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26821 do_estimate_edge_time, estimate_edge_time): Likewise.
26822 * ipa-inline-analysis.c (estimate_node_size_and_time,
26823 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26824 (estimate_time_after_inlining): Remove.
26826 2017-04-28 Martin Liska <mliska@suse.cz>
26828 * doc/gcov.texi: Enhance documentation of gcov.
26830 2017-04-28 Martin Liska <mliska@suse.cz>
26832 * doc/gcov.texi: Sort options in alphabetic order.
26833 * doc/gcov-dump.texi: Likewise.
26834 * doc/gcov-tool.texi: Likewise.
26835 * gcov.c (print_usage): Likewise.
26836 * gcov-dump.c (print_usage): Likewise.
26837 * gcov-tool.c (print_merge_usage_message): Likewise.
26838 (print_rewrite_usage_message): Likewise.
26839 (print_overlap_usage_message): Likewise.
26841 2017-04-28 Martin Liska <mliska@suse.cz>
26843 PR gcov-profile/53915
26844 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26846 2017-04-28 Martin Liska <mliska@suse.cz>
26848 PR gcov-profile/79891
26849 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26850 is marked by compiler as living on a line.
26851 (get_cycles_count): Remove usage of the union.
26852 (output_intermediate_file): Likewise.
26853 (find_source): Fix GNU coding style.
26854 (accumulate_line_counts): Remove old non-all block mode.
26855 (output_lines): Remove usage of the union.
26856 * profile.c (output_location): Include all BBs, even if
26857 belonging to a same line (and file) as a previous BB.
26859 2017-04-28 Martin Liska <mliska@suse.cz>
26861 * gcov.c (process_args): Handle new argument 'w'.
26862 (read_graph_file): Assign ID to BBs.
26863 (output_branch_count): Display BB # if verbose flag is set.
26864 (output_lines): Likewise for arcs.
26865 (print_usage): Add '--verbose' option help.
26866 * doc/gcov.texi: Document --verbose (-w) option.
26868 2017-04-28 Martin Liska <mliska@suse.cz>
26870 * gcov.c (struct block_location_info): New struct.
26871 (process_file): Fill up the new structure.
26872 (read_graph_file): Replace usage of encoding by the newly added
26874 (add_line_counts): Likewise.
26875 (accumulate_line_counts): Remove usage of the union.
26876 (function_info::function_info): New function.
26877 (function_info::~function_info): Likewise.
26878 (process_file): Call delete instead of release_function.
26879 (release_function): Release the function.
26880 (release_structures): Call delete instead of release_function.
26881 (solve_flow_graph): Replace usage of num_blocks.
26882 (find_exception_blocks): Likewise.
26883 (output_lines): Fix GNU coding style.
26885 2017-04-28 Martin Liska <mliska@suse.cz>
26888 * coverage.c (coverage_remove_note_file): New function.
26889 * coverage.h: Declare the function.
26890 * toplev.c (finalize): Clean if an error has been seen.
26892 2017-04-28 Martin Liska <mliska@suse.cz>
26894 PR gcov-profile/80031
26895 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26896 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26897 * gcov.c (read_graph_file): Read just number of blocks.
26898 * profile.c (branch_prob): Do not stream 0 flags per a basic
26901 2017-04-28 Martin Liska <mliska@suse.cz>
26903 * gcov-dump.c (tag_*): Add new argument to declarations.
26904 (dump_gcov_file): Likewise.
26905 (tag_blocks): Add and use new argument depth.
26906 (tag_arcs): Likewise.
26907 (tag_lines): Likewise.
26908 (tag_counters): Likewise.
26909 (tag_summary): Likewise.
26910 (dump_working_sets): Use depth to do a proper indentation.
26912 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26915 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26916 * symtab.c (symtab_node::debug_symtab): Move definition here.
26918 2017-04-28 Richard Biener <rguenther@suse.de>
26920 * lto-streamer.h (LTO_major_version): Bump to 7.
26922 2017-04-28 Richard Biener <rguenther@suse.de>
26924 * tree-vrp.c (assert_info): New struct.
26925 (add_assert_info): New helper.
26926 (register_edge_assert_for_2): Refactor to add asserts to a vector
26928 (register_edge_assert_for_1): Likewise.
26929 (register_edge_assert_for): Likewise.
26930 (finish_register_edge_assert_for): New helper actually registering
26931 asserts where live on edge.
26932 (find_conditional_asserts): Adjust.
26933 (find_switch_asserts): Likewise.
26934 (evrp_dom_walker::try_find_new_range): Generalize.
26935 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26937 2017-04-27 Marek Polacek <polacek@redhat.com>
26940 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26941 arg10 and arg11 to itype.
26943 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26945 * doc/extend.texi (Object Size Checking): Improve grammar.
26947 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26950 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26951 that the logic for permitting reciprocal estimates matches that
26954 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26957 * tree.c (type_cache_hasher::equal): Only compare
26958 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26959 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26960 non-aggregate element types.
26961 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26962 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26964 2017-04-27 Richard Biener <rguenther@suse.de>
26966 PR middle-end/80533
26967 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26968 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26969 keeping a reference to a trailing array.
26971 2017-04-27 Richard Biener <rguenther@suse.de>
26973 PR middle-end/80539
26974 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26975 being in loop-closed SSA form conservatively.
26976 (chrec_fold_multiply_poly_poly): Likewise.
26978 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26980 PR middle-end/79665
26981 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26982 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26984 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26987 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26988 (aarch64_function_arg_alignment): Return unsigned int again, but still
26989 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26990 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26991 Don't emit -Wpsabi note.
26992 (aarch64_function_arg_boundary): Likewise.
26993 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26996 2017-04-26 Nathan Sidwell <nathan@acm.org>
26998 * tree.h (crc32_unsigned_n): Declare.
26999 (crc32_unsigned, crc32_unsigned): Make inline.
27000 * tree.c (crc32_unsigned_bits): Replace with ...
27001 (crc32_unsigned_n): ... this.
27002 (crc32_unsigned, crc32_byte): Remove.
27003 (crc32_string): Remove unnecessary braces.
27005 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27007 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27008 * ipa-inline-analysis.c (MAX_TIME): Remove.
27009 (account_size_time): Use sreal for time.
27010 (dump_inline_summary): Update.
27011 (estimate_function_body_sizes): Update.
27012 (estimate_edge_size_and_time): Update.
27013 (estimate_calls_size_and_time): Update.
27014 (estimate_node_size_and_time): Update.
27015 (inline_merge_summary): Update.
27016 (inline_update_overall_summary): Update.
27017 (estimate_time_after_inlining): Update.
27018 (inline_read_section): Update.
27019 (inline_write_summary): Update.
27020 * ipa-inline.c (compute_uninlined_call_time): Update.
27021 (compute_inlined_call_time): Update.
27022 (recursive_inlining): Update.
27023 (inline_small_functions): Update.
27024 (dump_overall_stats): Update.
27025 * ipa-inline.h: Include sreal.h.
27026 (size_time_entry): Turn time to sreal.
27027 (inline_summary): Turn self_time nad time to sreal.
27029 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27031 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27033 (sreal::stream_out, sreal::stream_in): New.
27034 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27036 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27038 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27041 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27044 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27046 (dimode_scalar_chain::compute_convert_gain): Ditto.
27047 (dimode_scalar_chain::make_vector_copies): Ditto.
27048 (dimode_scalar_chain::convert_reg): Ditto.
27049 (dimode_scalar_chain::convert_insn): Ditto.
27050 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27051 (VI248_AVX512BW_1): New mode iterator.
27052 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27053 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27056 2017-04-25 Martin Sebor <msebor@redhat.com>
27058 PR tree-optimization/80497
27059 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27060 constants are representable in HOST_WIDE_INT.
27061 (parse_directive): Ditto.
27063 2017-04-25 Martin Sebor <msebor@redhat.com>
27066 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27067 (new_zero_array): Adjust signature.
27068 (dom_info::dom_init): Used unsigned rather that size_t.
27069 (dom_info::dom_info): Same.
27071 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27072 Jakub Jelinek <jakub@redhat.com>
27075 * config/arm/arm.c: Include gimple.h.
27076 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27077 returns negative, increment ncrn only if it returned positive.
27078 (arm_needs_doubleword_align): Return int instead of bool,
27079 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27080 members, but if there is any such non-FIELD_DECL
27081 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27082 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27083 returns negative, increment nregs only if it returned positive.
27084 (arm_setup_incoming_varargs): Likewise.
27085 (arm_function_arg_boundary): Emit -Wpsabi note if
27086 arm_needs_doubleword_align returns negative, return
27087 DOUBLEWORD_ALIGNMENT only if it returned positive.
27089 2017-04-25 Marek Polacek <polacek@redhat.com>
27092 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27093 first argument to type.
27095 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27098 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27099 type checks to test for compatibility instead of equality.
27101 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27102 Jakub Jelinek <jakub@redhat.com>
27105 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27107 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27108 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27109 the alignment computation, but return their maximum in warn_alignment.
27110 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27111 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27113 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27114 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27117 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27119 * config/arc/simdext.md (dmpyh): Fix typo.
27121 2017-04-25 Richard Biener <rguenther@suse.de>
27123 PR tree-optimization/80492
27124 * alias.c (compare_base_decls): Handle registers with asm
27125 specification conservatively.
27126 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27127 compare_base_decls returning dont-know properly.
27129 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27131 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27132 (legitimate_offset_address_p): New function.
27133 (arc_legitimate_address_p): Use above function.
27135 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27137 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27139 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27141 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27142 ACCH registers whenever they are available.
27144 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27146 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27147 double regs fix when not used.
27149 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27151 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27153 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27154 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27156 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27158 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27159 when emitting short ADD instructions.
27161 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27163 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27165 (cmpsi_cc_c_insn): Likewise.
27166 (cbranchsi4_scratch): Compute proper instruction length using
27167 compact_hreg_operand.
27168 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27170 2017-04-25 Richard Biener <rguenther@suse.de>
27172 PR middle-end/80509
27173 * passes.c (pass_manager::pass_manager): Initialize
27174 m_name_to_pass_map.
27176 2017-04-25 Richard Biener <rguenther@suse.de>
27178 PR tree-optimization/79201
27179 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27181 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27184 * config/s390/vector.md: Split MEM->GPR vector moves for
27185 non-s_operand addresses.
27187 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27190 * config/s390/predicates.md (reload_const_wide_int_operand): New
27192 * config/s390/s390.md ("movti"): Remove d/P alternative.
27193 ("movti_bigconst"): New pattern definition.
27195 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27198 * s390-protos.h (s390_expand_cs_hqi): Removed.
27199 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27200 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27201 modes as well as CCZ1mode and CCZmode.
27202 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27203 signature of s390_emit_compare_and_swap.
27204 (s390_expand_cs_hqi): Likewise, make static.
27205 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27206 compare-and-swap, in some cases.
27207 (s390_expand_cs): Wrapper function.
27208 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27210 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27211 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27212 patterns for small and large integers. Forbid symref memory operands.
27213 Move expander to s390.c. Require cc register.
27214 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27215 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27216 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27217 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27218 symref memory operands. Remove CC mode and call s390_match_ccmode
27220 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27222 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27224 * config/s390/s390.md (define_peephole2): New peephole to help
27225 combining the load-and-test pattern with volatile memory.
27227 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27229 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27230 with CCZmode for TARGET_Z196.
27232 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27234 PR rtl-optimization/80501
27235 * combine.c (make_compound_operation_int): Set subreg_code to SET
27236 even for AND with mask of the sign bit of mode.
27238 PR rtl-optimization/80500
27239 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27240 sum's initial value.
27242 2017-04-25 Julian Brown <julian@codesourcery.com>
27243 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27245 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27247 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27249 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27251 2017-04-25 Julian Brown <julian@codesourcery.com>
27252 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27254 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27255 (thunderx2t99_sha): New Reservation.
27257 2017-04-25 Julian Brown <julian@codesourcery.com>
27258 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27260 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27261 type for 1-element load.
27263 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27265 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27267 2017-04-24 Martin Jambor <mjambor@suse.cz>
27269 PR tree-optimization/80293
27270 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27271 char arrays not totally scalarizable if it is false.
27272 (analyze_all_variable_accesses): Pass correct value in the new
27273 parameter. Add a statistics counter.
27275 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27277 PR middle-end/79931
27278 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27280 2017-04-24 Richard Biener <rguenther@suse.de>
27282 PR tree-optimization/80494
27283 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27284 out for complex types.
27286 2017-04-24 Richard Biener <rguenther@suse.de>
27288 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27289 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27290 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27291 (DFS): Adjust and never fail.
27292 (sccvn_dom_walker::fail): Remove.
27293 (sccvn_dom_walker::before_dom_children): Adjust.
27294 (run_scc_vn): Likewise and never fail.
27295 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27296 (pass_fre::execute): Likewise.
27298 2017-04-24 Richard Biener <rguenther@suse.de>
27300 PR tree-optimization/79725
27301 * tree-ssa-sink.c (statement_sink_location): Return whether
27302 failure reason was zero uses. Move that check later.
27303 (sink_code_in_bb): Deal with zero uses by removing the stmt
27306 2017-04-24 Richard Biener <rguenther@suse.de>
27309 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27310 pointer-based references.
27312 2017-04-24 Richard Biener <rguenther@suse.de>
27315 * pass_manager.h (pass_manager::operator new): Remove.
27316 (pass_manager::operator delete): Likewise.
27317 * passes.c (pass_manager::operator new): Remove.
27318 (pass_manager::operator delete): Likewise.
27319 (pass_manager::pass_manager): Zero individual pass members.
27321 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27324 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27325 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27326 Check "XEXP (src, 1)" operand here.
27327 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27328 Check "XEXP (src, 1)" operand here.
27329 (dimode_scalar_chain::make_vector_copies): Detect count register
27330 of a shift instruction. Zero extend count register from QImode
27331 to DImode to satisfy vector shift pattern count operand predicate.
27332 Substitute vector shift count operand with a DImode copy.
27333 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27336 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27338 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27339 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27340 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27341 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27342 (UNSPEC_NOREX_MEM): Remove definition.
27344 2017-04-21 Richard Biener <rguenther@suse.de>
27346 PR tree-optimization/79547
27347 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27348 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27349 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27350 without any constraints.
27352 2017-04-21 Richard Biener <rguenther@suse.de>
27354 PR tree-optimization/78847
27355 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27357 2017-04-21 Richard Biener <rguenther@suse.de>
27359 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27360 (build_distinct_type_copy): Likewise.
27361 (build_variant_type_copy): Likewise.
27362 * tree.c (build_qualified_type): Pass down mem-stat info.
27363 (build_distinct_type_copy): Likewise.
27364 (build_variant_type_copy): Likewise.
27366 2017-04-21 Richard Biener <rguenther@suse.de>
27368 PR tree-optimization/80237
27369 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27371 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27372 for a simplified result.
27374 2016-04-21 Richard Biener <rguenther@suse.de>
27376 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27377 sth as strict as a simple_iv but a chrec without symbols and an
27378 operand defined in the loop we are peeling (and not some subloop).
27379 (propagate_constants_for_unrolling): Propagate all constants.
27381 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27384 * config/i386/i386.c (print_reg): Remove assert for disalowed
27385 regno values, call output_operand_lossage instead.
27387 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27390 * config/i386/constraints.md (Yc): New register constraint.
27391 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27392 Use Yc constraint for alternative 2 of operand 0. Remove
27393 preferred_for_speed attribute.
27395 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27397 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27398 lastprivate clauses in SIMT case.
27400 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27402 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27404 2017-04-20 Richard Biener <rguenther@suse.de>
27406 PR tree-optimization/57796
27407 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27408 as N scalar stores.
27409 (vect_model_load_cost): Cost gathers as N scalar loads.
27411 2017-04-20 Richard Biener <rguenther@suse.de>
27413 * ggc-page.c (ggc_allocated_p): Rename to ...
27414 (safe_lookup_page_table_entry): ... this and return the lookup
27416 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27418 2017-04-20 Richard Biener <rguenther@suse.de>
27420 PR tree-optimization/80453
27421 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27422 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27423 from the conditions.
27424 (vn_phi_eq): Pass them down.
27425 (vn_phi_lookup): Record them.
27426 (vn_phi_insert): Likewise.
27428 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27430 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27431 uninitialized variable warning to avoid buffer overrun.
27433 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27436 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27437 is suppressed for '{ 0 }' in C.
27439 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27441 * BASE-VER: Set to 8.0.0.
27443 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27445 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27446 priority .init_array and .fini_array section with SECTION_NOTYPE
27449 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27451 PR middle-end/80423
27452 * tree.h (build_array_type): Add typeless_storage default argument.
27453 * tree.c (type_cache_hasher::equal): Also compare
27454 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27455 (build_array_type): Add typeless_storage argument, set
27456 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27458 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27459 (build_array_type): Likewise. Add typeless_storage argument.
27461 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27462 Jakub Jelinek <jakub@redhat.com>
27464 PR tree-optimization/80426
27465 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27466 operation on symbolic operands, also compute the overflow for the
27467 invariant part when the operation degenerates into a negation.
27469 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27472 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27473 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27476 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27478 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27481 * config/avr/avr.c (tree.h): Include it.
27482 (cgraph.h): Include it.
27483 (avr_encode_section_info): Don't warn for uninitialized progmem
27484 variable if it's just an alias.
27486 2017-04-19 Richard Biener <rguenther@suse.de>
27489 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27490 when needed by AutoPGO.
27492 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27495 * doc/lto.texi: Remove an extra 'that'.
27497 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27499 PR rtl-optimization/80429
27500 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27501 are only used in debug insns.
27503 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27504 Vladimir Makarov <vmakarov@redhat.com>
27506 * config/sparc/predicates.md (input_operand): Add comment. Return
27507 true for any memory operand when LRA is in progress.
27508 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27510 2017-04-18 Jeff Law <law@redhat.com>
27513 * mips.md ({return,simple_return}_internal): Do not overwrite
27516 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27518 PR tree-optimization/80443
27519 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27520 instead of adding 1, subtract -1 and similarly instead of subtracting
27523 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27525 PR rtl-optimization/80357
27526 * haifa-sched.c (tmp_bitmap): New variable.
27527 (model_recompute): Handle duplicate use records.
27528 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27529 (free_global_sched_pressure_data): Free it.
27531 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27534 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27535 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27536 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27537 instead of SYSTEM_HEADER_DIR.
27539 2017-04-18 Jeff Law <law@redhat.com>
27541 PR middle-end/80422
27542 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27543 predecessors after walking up the insn chain.
27545 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27548 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27549 sizetype type into debug info.
27551 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27554 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27555 unneeded test for TARGET_UPPER_REGS_SF.
27556 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27558 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27561 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27562 instead of gsi_after_labels.
27564 2017-04-18 Jeff Law <law@redhat.com>
27566 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27570 2017-04-13 Jeff Law <law@redhat.com>
27571 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27572 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27574 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27577 * config/avr/avr.c (intl.h): Include it.
27578 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27580 2017-04-18 Martin Liska <mliska@suse.cz>
27582 PR gcov-profile/78783
27583 * gcov-tool.c (gcov_output_files): Validate that destination
27584 file is either removed by the tool or by a user.
27586 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27587 Guy Benyei <guybe@mellanox.com>
27589 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27590 block, and do not negate it, the stored id is already negative.
27592 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27594 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27596 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27599 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27600 masks of options that should be turned off if the VSX vector
27601 options are turned off.
27602 (OTHER_P8_VECTOR_MASKS): Likewise.
27603 (OTHER_VSX_VECTOR_MASKS): Likewise.
27604 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27605 rs6000_disable_incompatible_switches to validate no type switches
27607 (rs6000_incompatible_switch): New function to disallow turning on
27608 other vector options if -mno-vsx, -mno-power8-vector, or
27609 -mno-power9-vector are specified.
27611 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27613 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27615 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27617 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27618 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27619 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27620 (ARG_POINTER_CFA_OFFSET): Likewise.
27622 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27624 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27625 conditions to take advantage of various optimizations.
27627 2017-04-13 Jeff Law <law@redhat.com>
27629 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27630 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27631 (zero_extendsidi2_dext): Likewise.
27633 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27636 * fold-const.c (fold_ternary_loc): Revert
27637 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27640 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27642 PR rtl-optimization/80343
27643 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27644 new scratch pseudo.
27646 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27649 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27650 to ubsan_encode_value.
27652 2017-04-13 Jeff Law <law@redhat.com>
27654 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27655 appearing in DEBUG_INSNs.
27657 2017-04-13 Martin Liska <mliska@suse.cz>
27659 PR gcov-profile/80413
27660 * gcov-io.c (gcov_write_string): Copy to buffer just when
27661 allocated size is greater than zero.
27663 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27666 * dwarf2out.c (decls_for_scope): Ignore declarations of
27667 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27669 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27672 * ipa-visibility.c (non_local_p): Fix typos.
27673 (localize_node): When localizing symbol in same comdat group,
27674 dissolve the group only when we know external symbols are going
27676 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27678 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27680 PR tree-optimization/79390
27681 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27682 order does not result in usable sequence, retry with reversed operand
27688 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27689 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27690 op0 instead of fold_convert_loc (loc, type, arg0).
27692 2017-04-12 Jeff Law <law@redhat.com>
27694 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27695 has a delay slot in the generated code.
27697 * config/cris/cris.md (cris_preferred_reload_class): Return
27698 GENNONACR_REGS rather than GENERAL_REGS.
27700 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27703 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27704 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27705 signedness of the result type.
27707 2017-04-12 Richard Biener <rguenther@suse.de>
27708 Jeff Law <law@redhat.com>
27710 PR tree-optimization/80359
27711 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27712 trim stores to TARGET_MEM_REFs.
27714 2017-04-12 Richard Biener <rguenther@suse.de>
27716 PR tree-optimization/79390
27717 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27718 threading case even more.
27720 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27723 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27724 for quad_address_p for TImode, instead of just not indexed_address.
27726 2017-04-12 Richard Biener <rguenther@suse.de>
27727 Bernd Edlinger <bernd.edlinger@hotmail.de>
27729 PR middle-end/79671
27730 * alias.c (component_uses_parent_alias_set_from): Handle
27731 TYPE_TYPELESS_STORAGE.
27732 (get_alias_set): Likewise.
27733 * tree-core.h (tree_type_common): Add typeless_storage flag.
27734 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27735 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27736 for types containing members with TYPE_TYPELESS_STORAGE.
27737 (place_field): Likewise.
27738 (layout_type): Likewise for ARRAY_TYPE.
27739 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27740 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27741 TYPE_TYPELESS_STORAGE.
27742 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27744 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27747 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27748 first argument to type.
27750 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27754 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27755 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27756 (rs6000_expand_binop_builtin): Likewise.
27757 (rs6000_expand_ternop_builtin): Likewise; also add missing
27758 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27759 vshasigma built-ins.
27760 * doc/extend.texi: Document that vec_xxpermdi's third argument
27761 must be a constant.
27763 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27765 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27766 Use shift_const cost parameter when calculating gain of STV shifts.
27768 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27770 PR rtl-optimization/70478
27771 * lra-constraints.c (process_alt_operands): Check memory for
27772 disfavoring memory insn operand.
27774 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27776 PR middle-end/80100
27777 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27778 left shift in unsigned HOST_WIDE_INT type.
27780 PR rtl-optimization/80385
27781 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27782 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27785 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27786 if they have any depend clauses.
27788 2017-04-11 Martin Liska <mliska@suse.cz>
27791 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27792 * ipa-split.c (split_function): Create a local comdat symbol
27793 if caller is in a comdat group.
27795 2017-04-11 Martin Liska <mliska@suse.cz>
27798 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27801 2017-04-11 Martin Sebor <msebor@redhat.com>
27803 PR middle-end/80364
27804 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27805 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27807 (directive::set_width, directive::set_precision, format_character):
27809 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27812 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27815 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27816 conflict, set target->arch_name instead of target->cpu_name.
27818 2017-04-11 Richard Biener <rguenther@suse.de>
27820 PR tree-optimization/80374
27821 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27822 build_zero_cst, remove fold_convertible_p check again.
27824 2017-04-11 Martin Liska <mliska@suse.cz>
27827 * ubsan.c (instrument_object_size): Do not instrument register
27830 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27833 * config/i386/i386-builtin-types.def
27834 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27835 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27836 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27837 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27838 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27839 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27840 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27841 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27842 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27843 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27844 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27845 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27846 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27847 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27848 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27849 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27850 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27851 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27852 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27853 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27854 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27855 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27856 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27857 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27858 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27859 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27860 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27861 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27862 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27863 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27864 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27865 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27866 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27867 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27868 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27869 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27870 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27871 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27872 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27873 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27874 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27875 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27876 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27877 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27878 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27879 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27881 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27882 flag to second_arg_count, handle 4 argument function type _COUNT
27883 aliases, handle second_arg_count on second argument rather than last.
27885 2017-04-10 Jeff Law <law@redhat.com>
27887 PR tree-optimization/80374
27888 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27889 record anything if we can not convert integer_zero_node to the
27892 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27895 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27896 Enhance special handling given to the TARGET_P9_MINMAX option in
27897 relation to certain other options.
27899 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27901 PR tree-optimization/80153
27902 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27903 remove POINTER_PLUS_EXPR's base part directly, rather than through
27906 2017-04-10 Richard Biener <rguenther@suse.de>
27907 Bin Cheng <bin.cheng@arm.com>
27909 PR tree-optimization/80153
27910 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27911 the first element of pointer type aff_tree. Build result expr in
27913 (add_elt_to_tree): Convert to type unconditionally. Remove other
27914 fold_convert calls.
27915 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27916 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27918 2017-04-10 Richard Biener <rguenther@suse.de>
27920 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27923 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27925 PR rtl-optimization/70478
27926 * lra-constraints.c (curr_small_class_check): New.
27927 (update_and_check_small_class_inputs): New.
27928 (process_alt_operands): Update curr_small_class_check. Disfavor
27929 alternative insn memory operands. Check available regs for small
27932 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27935 * config/mips/mips.opt (-mvirt): Update description.
27936 * doc/invoke.texi (-mvirt): Likewise.
27938 2017-04-10 Richard Biener <rguenther@suse.de>
27940 PR middle-end/80362
27941 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27942 looking for NEGATE_EXPR in -A / -B to A / B folding.
27944 2017-04-10 Martin Liska <mliska@suse.cz>
27946 PR gcov-profile/80224
27947 * gcov.c (print_usage): Fix usage string.
27948 (get_gcov_intermediate_filename): Remove.
27949 (output_gcov_file): Use both for normal and intermediate format.
27950 (generate_results): Do not initialize special file for
27951 intermediate format.
27953 2017-04-10 Richard Biener <rguenther@suse.de>
27955 PR tree-optimization/80304
27956 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27959 2017-04-10 Nathan Sidwell <nathan@acm.org>
27962 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27963 (rs6000_init_builtins): Use it.
27965 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27967 * config/arm/arm.md (<mrc>): Add mode to SET source.
27968 (<mrrc>): Likewise.
27970 2017-04-10 Richard Biener <rguenther@suse.de>
27972 PR middle-end/80344
27973 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27975 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27978 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27979 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27980 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27981 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27982 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27983 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27984 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27985 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27986 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27987 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27988 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27989 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27990 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27991 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27992 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27993 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27994 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27995 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27996 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27997 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27998 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27999 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28000 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28002 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28004 PR rtl-optimization/70478
28005 * lra-constraints.c: Reverse the last patch.
28007 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28009 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28010 Add comment for WCHAR_T.
28012 2017-04-08 Martin Liska <mliska@suse.cz>
28015 2017-04-07 Martin Liska <mliska@suse.cz>
28018 * ipa-split.c (split_function): Add function part to a same comdat
28021 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28024 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28026 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28028 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28029 (vec_pairing): ...to this.
28030 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28031 (rs6000_sched_init): Adjust for name change.
28032 (struct rs6000_sched_context): Likewise.
28033 (rs6000_init_sched_context): Likewise.
28034 (rs6000_set_sched_context): Likewise.
28036 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28042 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28044 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28045 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28046 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28048 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28050 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28052 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28054 PR rtl-optimization/70703
28055 * ira-color.c (update_conflict_hard_regno_costs): Use
28056 int64_t instead of HOST_WIDE_INT.
28058 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28060 PR rtl-optimization/70478
28061 * lra-constraints.c (process_alt_operands): Disfavor alternative
28062 insn memory operands.
28064 2017-04-07 Jeff Law <law@redhat.com>
28066 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28067 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28069 2017-04-07 Martin Liska <mliska@suse.cz>
28072 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28073 Show error message instead of an ICE.
28075 2017-04-07 Martin Liska <mliska@suse.cz>
28078 * ipa-split.c (split_function): Add function part to a same comdat
28081 2017-04-07 Richard Biener <rguenther@suse.de>
28083 PR middle-end/80341
28084 * tree.c (get_unwidened): Also handle ! for_type case for
28086 * convert.c (do_narrow): Split out from ...
28087 (convert_to_integer_1): ... here. Do not pass final truncation
28088 type to get_unwidened for TRUNC_DIV_EXPR.
28090 2017-04-07 Richard Biener <rguenther@suse.de>
28092 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28094 (aff_combination_const): Adjust.
28095 (aff_combination_scale): Likewise.
28096 (aff_combination_add_elt): Likewise.
28097 (aff_combination_add_cst): Likewise.
28098 (aff_combination_convert): Likewise.
28099 (add_elt_to_tree): Likewise. Remove unused argument.
28100 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28102 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28104 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28106 * config/arm/arm.c (arm_default_short_enums): Use
28107 ARM_DEFAULT_SHORT_ENUMS.
28108 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28110 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28113 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28114 members with redundant out-of-class redeclaration.
28116 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28119 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28120 * config/i386/i386.md (*zero_extendsidi2):
28121 Add (?*x,*x) and (?*v,*v) alternatives.
28123 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28126 * config/i386/i386.c (ix86_expand_builtin)
28127 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28128 mode from insn data. Convert operands to insn operand mode.
28129 Copy operands that don't satisfy insn predicate to a register.
28131 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28133 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28136 2017-04-06 Richard Biener <rguenther@suse.de>
28138 PR tree-optimization/80334
28139 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28140 preserve alignment of accesses.
28142 2017-04-06 Richard Biener <rguenther@suse.de>
28144 PR tree-optimization/80262
28145 * tree-sra.c (build_ref_for_offset): Preserve address-space
28147 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28148 Drop useless address-space information on MEM_REF offsets.
28150 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28152 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28154 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28156 PR rtl-optimization/70703
28157 * ira-color.c (update_conflict_hard_regno_costs): Use
28158 HOST_WIDE_INT instead of long.
28160 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28163 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28164 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28166 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28167 for x86_64 target. Handle -m3dnowa option.
28169 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28171 PR rtl-optimization/70703
28172 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28173 (update_conflict_hard_regno_costs): Use long instead of unsigned
28174 arithmetic for cost calculation.
28176 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28177 Bernd Edlinger <bernd.edlinger@hotmail.de>
28180 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28183 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28186 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28187 ptr_mode with Pmode throughout.
28188 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28189 into probe_stack_range and use DImode.
28191 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28194 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28195 call_eh_return is true.
28197 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28199 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28200 Initialize last_match_fntype_index.
28202 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28205 * tree-nvr.c: Include internal-fn.h.
28206 (pass_return_slot::execute): Ignore internal calls without
28209 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28210 Richard Biener <rguenther@suse.de>
28213 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28214 captures used multiple times, except for the last use.
28215 * generic-match-head.c: Include gimplify.h.
28217 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28219 PR tree-optimization/79390
28220 * target.h (struct noce_if_info): Declare.
28221 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28222 * target.def (noce_conversion_profitable_p): New target hook.
28223 * ifcvt.h (struct noce_if_info): New type, moved from ...
28224 * ifcvt.c (struct noce_if_info): ... here.
28225 (noce_conversion_profitable_p): Renamed to ...
28226 (default_noce_conversion_profitable_p): ... this. No longer
28228 (noce_try_store_flag_constants, noce_try_addcc,
28229 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28230 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28231 instead of noce_conversion_profitable_p.
28232 * config/i386/i386.c: Include ifcvt.h.
28233 (ix86_option_override_internal): Don't override
28234 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28235 (ix86_noce_conversion_profitable_p): New function.
28236 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28237 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28238 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28239 * doc/tm.texi: Regenerated.
28241 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28243 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28246 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28249 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28250 instructions for small multiply cores.
28252 2017-04-04 Jeff Law <law@redhat.com>
28254 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28256 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28257 that are not set by the loop over the elements.
28259 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28262 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28263 int mode, convert_modes it to mode as unsigned, otherwise use
28264 lowpart_subreg to mode rather than SImode.
28265 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28266 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28267 Use DImode instead of SImode for the shift count operand.
28268 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28271 2017-04-04 Richard Biener <rguenther@suse.de>
28273 PR middle-end/80281
28274 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28275 arithmetic done for the negate or the plus. Simplify.
28276 (A - (-B) -> A + B): Likewise.
28277 * fold-const.c (split_tree): Make sure to not negate pointers.
28279 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28281 PR rtl-optimization/60818
28282 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28283 a compare of comparisons with the thing compared if this results
28284 in a different machine mode.
28286 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28288 * alias.c (base_alias_check): Fix typo in comment.
28289 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28290 * cgraphunit.c (symbol_table::compile): Likewise.
28291 * collect2.c (maybe_run_lto_and_relink): Likewise.
28292 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28293 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28294 * config/avr/avr.c (avr_map_op_t): Likewise.
28295 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28296 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28297 * config/epiphany/epiphany.md (movcc): Likewise.
28298 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28299 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28301 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28302 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28303 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28304 * config/sh/sh.c (sh_rtx_costs): Likewise.
28305 * fold-const.c (fold_truth_andor): Likewise.
28306 * genautomata.c (collapse_flag): Likewise.
28307 * gengtype.h (struct type::u::s): Likewise.
28308 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28309 * input.c (FORMAT_AMOUNT): Likewise.
28310 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28311 (known_aggs_to_agg_replacement_list): Likewise.
28312 * ipa-inline-analysis.c: Likewise.
28313 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28314 * ipa-polymorphic-call.c
28315 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28316 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28317 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28319 * modulo-sched.c (apply_reg_moves): Likewise.
28320 * omp-expand.c (build_omp_regions_1): Likewise.
28321 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28322 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28323 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28324 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28325 * value-prof.c: Likewise.
28326 * var-tracking.c (val_reset): Likewise.
28328 2017-04-03 Richard Biener <rguenther@suse.de>
28330 PR tree-optimization/80275
28331 * fold-const.c (split_address_to_core_and_offset): Handle
28334 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28336 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28337 descriptors is at least equal to that of functions.
28339 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28341 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28343 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28346 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28347 (mov<IMOD4:mode>): New expander.
28348 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28350 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28352 PR rtl-optimization/79405
28353 * fwprop.c (propagations_left): New variable.
28354 (forward_propagate_into): Decrement it.
28355 (fwprop_init): Initialize it.
28356 (fw_prop): If the variable has reached zero, stop propagating.
28357 (fwprop_addr): Ditto.
28359 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28362 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28363 a FUNCTION_DECL, pass it as decl instead of origin to
28366 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28368 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28371 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28374 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28375 TARGET_VSX_SMALL_INTEGER.
28377 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28379 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28380 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28382 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28384 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28385 extraction from odd-numbered MSA register.
28387 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28389 PR middle-end/80173
28390 * expmed.c (store_bit_field_1): Don't attempt to create
28391 a word subreg out of hard registers wider than word if they
28392 have HARD_REGNO_NREGS of 1 for their mode.
28394 PR middle-end/80163
28395 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28396 conversions to integer types wider than word and pointer.
28399 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28400 (rtx_equal_for_cselib_p): Pass 0 to it.
28401 * cselib.c (cselib_hasher::equal): Likewise.
28402 (rtx_equal_for_cselib_1): Add depth argument. If depth
28403 is 128, don't look up VALUE locs and punt. Increment
28404 depth in recursive calls when walking VALUE locs.
28406 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28408 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28409 (make_gcov_file_name): Use the canonical path name for generating
28411 (read_line): Fix handling of files with ascii null bytes.
28413 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28415 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28416 to initialise a vector register instead
28417 of using a const_int.
28419 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28421 PR translation/80189
28422 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28423 diagnostic messages.
28425 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28428 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28429 (dfp_diex_<mode>): Update mode of operand 1.
28430 * doc/extend.texi (dxex, dxexq): Document change to return type.
28431 (diex, diexq): Document change to argument type.
28433 2017-03-30 Martin Jambor <mjambor@suse.cz>
28436 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28437 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28438 it reflects the signature changes performed at the callee side.
28439 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28440 to cgraph_build_function_type_skip_args.
28441 (build_function_decl_skip_args): Adjust call to the above function.
28443 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28446 * config/i386/sse.md
28447 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28448 register as dest whenever it is a MEM not rtx_equal_p to the
28449 corresponding dup operand, and when forcing into reg move the
28450 reg into the memory afterwards.
28451 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28452 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28453 for the force_reg mode.
28454 (avx512vl_vextractf128<mode>): Use register as dest either
28455 always when a MEM, or when it is a MEM not rtx_equal_p to the
28456 corresponding dup operand, or even not when it is a CONST_VECTOR
28457 depending on the mode and lo vs. hi.
28458 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28460 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28461 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28462 Likewise. Require that operands[2] is even.
28463 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28464 Remove extraneous parens. Require that operands[2] is a multiple
28466 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28467 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28468 disallow memory then.
28470 2017-03-30 Richard Biener <rguenther@suse.de>
28472 PR tree-optimization/77498
28473 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28474 to non-constants over backedges.
28476 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28478 PR rtl-optimization/80233
28479 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28480 as last_combined_insn. Do not test for BARRIER_P separately.
28482 2017-03-29 Andreas Schwab <schwab@suse.de>
28485 * calls.c (prepare_call_address): Convert funexp to Pmode before
28486 copying to temp reg.
28488 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28490 PR tree-optimization/80158
28491 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28492 Handle possible future case of more than one alternate
28494 (replace_rhs_if_not_dup): Likewise.
28495 (replace_one_candidate): Likewise.
28497 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28499 PR rtl-optimization/80193
28500 * ira.c (ira): Do not check allocation for LRA.
28502 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28504 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28505 (nvptx_output_simt_exit): Declare.
28506 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28507 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28508 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28509 (nvptx_declare_function_name): ...here. Emit declaration of local
28510 memory space buffer for omp_simt_enter insn.
28511 (nvptx_output_unisimt_switch): New.
28512 (nvptx_output_softstack_switch): New.
28513 (nvptx_output_simt_enter): New.
28514 (nvptx_output_simt_exit): New.
28515 * config/nvptx/nvptx.h (struct machine_function): New fields
28516 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28517 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28518 (UNSPECV_SIMT_EXIT): Ditto.
28519 (omp_simt_enter_insn): New insn.
28520 (omp_simt_enter): New expansion.
28521 (omp_simt_exit): New insn.
28522 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28524 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28525 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28526 (expand_GOMP_SIMT_EXIT): New.
28527 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28528 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28529 (GOMP_SIMT_EXIT): Ditto.
28530 * target-insns.def (omp_simt_enter): New insn.
28531 (omp_simt_exit): Ditto.
28532 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28534 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28535 (lower_rec_input_clauses): Likewise.
28536 (lower_lastprivate_clauses): Handle SIMT privatization.
28538 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28539 (ompdevlow_adjust_simt_enter): New.
28540 (find_simtpriv_var_op): New.
28541 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28542 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28544 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28545 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28546 (copy_decl_for_dup_finish): Ditto.
28548 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28550 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28553 * config/i386/i386.c (ix86_option_override_internal): Always
28554 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28556 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28558 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28560 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28562 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28563 mark new edge's irreducible flag accordign to it.
28564 (vect_do_peeling): Check loop preheader edge's irreducible flag
28565 and pass it to function slpeel_add_loop_guard.
28567 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28569 PR tree-optimization/80218
28570 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28571 Update block frequencies and counts.
28573 2017-03-28 Richard Biener <rguenther@suse.de>
28575 PR tree-optimization/78644
28576 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28577 of a simplification result we may not use it at all.
28579 2017-03-28 Richard Biener <rguenther@suse.de>
28582 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28583 without arguments, generate default definition of a SSA name.
28585 2017-03-28 Richard Biener <rguenther@suse.de>
28587 PR middle-end/80222
28588 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28589 TYPE_REF_CAN_ALIAS_ALL references.
28590 * fold-const.c (fold_indirect_ref_1): Likewise.
28592 2017-03-28 Martin Liska <mliska@suse.cz>
28595 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28596 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28598 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28599 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28601 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28602 (EXTRA_SPECS): Define.
28603 (SUBTARGET_EXTRA_SPECS): Likewise.
28604 (SUBTARGET_CPP_SPEC): Likewise.
28605 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28606 SUBTARGET_EXTRA_SPECS.
28607 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28609 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28611 * config/arc/simdext.md (vst64_insn): Update pattern.
28612 (vld32wh_insn): Likewise.
28613 (vld32wl_insn): Likewise.
28614 (vld64_insn): Likewise.
28615 (vld32_insn): Likewise.
28617 2017-03-28 Marek Polacek <polacek@redhat.com>
28620 * fold-const.c (fold_comparison): Use protected_set_expr_location
28621 instead of SET_EXPR_LOCATION.
28623 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28625 * tree.c (add_expr): Avoid name lookup warning.
28627 2017-03-27 Jeff Law <law@redhat.com>
28629 PR tree-optimization/80216
28630 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28631 function name. Limit recursion depth.
28632 (record_temporary_equivalences): Corresponding changes.
28634 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28636 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28639 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28642 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28644 * cfgcleanup.c (reg_note_cfa_p): New array.
28645 (insns_have_identical_cfa_notes): New function.
28646 (old_insns_match_p): Don't cross-jump in between /f
28647 and non-/f instructions. If both i1 and i2 are frame related,
28648 verify all CFA notes, their order and content.
28650 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28653 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28654 HImode and SImode with zero extend to DImode to one insn.
28655 (bswap<mode>2_extenddi): Likewise.
28656 (bswapsi2_extenddi): Likewise.
28657 (bswaphi2_extendsi): Likewise.
28658 (bswaphi2): Combine bswap HImode and SImode into one insn.
28659 Separate memory insns from swapping register.
28660 (bswapsi2): Likewise.
28661 (bswap<mode>2): Likewise.
28662 (bswaphi2_internal): Delete, no longer used.
28663 (bswapsi2_internal): Likewise.
28664 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28665 store, and gpr<-gpr swap insns.
28666 (bswap<mode>2_store): Likewise.
28667 (bswaphi2_reg): Register only splitter, combine with the splitter.
28668 (bswaphi2 splitter): Likewise.
28669 (bswapsi2_reg): Likewise.
28670 (bswapsi2 splitter): Likewise.
28671 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28672 the insns into load, store, and register/register insns.
28673 (bswapdi2_ldbrx): Likewise.
28674 (bswapdi2_load): Likewise.
28675 (bswapdi2_store): Likewise.
28676 (bswapdi2_reg): Likewise.
28678 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28680 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28681 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28683 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28686 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28688 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28689 special handling for target option conflicts between dform
28690 options (-mpower9-dform, -mpower9-dform-vector,
28691 -mpower9-dform-scalar) and -mno-direct-move.
28693 2017-03-27 Richard Biener <rguenther@suse.de>
28695 PR tree-optimization/80181
28696 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28698 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28700 * config/arc/predicates.md (move_double_src_operand): Replace the
28701 call to move_double_src_operand with a call to address_operand.
28703 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28705 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28706 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28707 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28709 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28711 * config/arc/predicates.md (long_immediate_loadstore_operand):
28712 Consider scaled addresses cases.
28714 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28716 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28717 restored when in interrupt.
28718 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28719 doesn't have delay slot.
28721 2017-03-27 Richard Biener <rguenther@suse.de>
28724 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28725 inlined thunk clones.
28727 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28730 * asan.c (instrument_derefs): Copy over last operand from
28731 original COMPONENT_REF to the new COMPONENT_REF with
28732 DECL_BIT_FIELD_REPRESENTATIVE.
28733 * ubsan.c (instrument_object_size): Likewise.
28735 2017-03-27 Richard Biener <rguenther@suse.de>
28737 PR tree-optimization/80170
28738 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28739 sure DR/SCEV didnt fold in constants we do not see when looking
28740 at the reference base alignment.
28742 2017-03-27 Richard Biener <rguenther@suse.de>
28744 PR middle-end/80171
28745 * gimple-fold.c (fold_ctor_reference): Properly guard against
28746 NULL return value from canonicalize_constructor_val.
28748 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28751 * config/i386/i386.c (ix86_expand_builtin)
28752 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28753 flags reg setting and flags reg using instructions.
28754 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28755 clobbering instructions to zero extend op2.
28757 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28759 * doc/install.texi (Configuration) <--with-aix-soname>:
28760 Update link to AIX ld.
28762 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28764 PR rtl-optimization/80160
28765 PR rtl-optimization/80159
28766 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28767 reg_alternate_class into account.
28769 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28772 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28773 to consider in curr_insn_transform.
28775 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28777 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28778 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28779 and emit_mode_inner.
28781 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28783 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28784 argument to the overloaded builtin variants. Use the new flag to
28785 deprecate certain builtin variants.
28786 * config/s390/s390-builtin-types.def: Add new builtin types.
28787 * config/s390/s390-builtins.h: Support new flags field for
28788 overloaded builtins.
28789 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28790 (s390_macro_to_expand): Enable vector float data type.
28791 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28792 builtins by incrementing the __VEC__ version number.
28793 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28795 (s390_resolve_overloaded_builtin): Emit error messages depending
28796 on the builtin flags.
28797 * config/s390/s390.c (s390_expand_builtin): Support additional
28798 flags argument. Change error message to match the messages
28799 emitted in s390-c.c.
28800 * config/s390/s390.md: New UNSPEC_* constants.
28801 (op_type): Add new instruction types.
28802 * config/s390/vecintrin.h: Add new builtins and test data class
28804 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28805 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28806 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28807 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28808 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28809 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28811 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28812 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28813 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28814 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28816 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28817 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28818 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28819 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28821 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28822 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28823 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28824 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28825 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28826 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28827 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28829 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28830 ("vec_scatter_element<V_HW_4:mode>_DI")
28831 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28832 ("vec_fpint<mode>", "vflls")
28833 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28834 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28835 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28836 ("*vec_cmphe<mode>_cc"): ... these.
28838 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28839 mode constant instead of magic value.
28841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28843 * config/s390/s390.c (s390_expand_vec_compare): Support other
28844 vector floating point modes than just V2DF.
28845 (s390_expand_vcond): Likewise.
28846 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28847 (s390_cannot_change_mode_class): Prevent mode changes between TF
28848 and V1TF in vector registers.
28849 * config/s390/s390.md (DF, SF): New mode attributes.
28850 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28851 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28852 SFmode support for VRs.
28853 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28855 (VFT, VF_HW): New mode iterators.
28856 (vw, sdx): New mode attributes.
28857 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28858 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28859 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28860 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28861 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28862 also the new vector floating point modes. Renaming to ...
28864 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28865 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28866 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28867 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28868 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28869 ("vec_unordered<mode>"): ... these.
28871 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28872 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28873 ("*vec_extendv2df"): New insn definitions.
28875 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28877 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28878 ("mulditi3_2", "*muldi3_sign"): New patterns.
28879 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28880 rename the pattern definition.
28882 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28884 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28886 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28888 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28890 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28891 instruction if possible.
28892 * config/s390/vector.md (vec_halfnumelts): New mode
28894 ("*vec_vllezlf<mode>"): New pattern.
28896 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28898 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28899 ("popcountv4si2", "popcountv2di2"): Rename to ...
28900 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28901 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28903 ("popcount<mode>2_vxe"): New pattern.
28905 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28907 * common/config/s390/s390-common.c (processor_flags_table): Add
28909 * config.gcc: Add arch12.
28910 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28911 Default to arch12 for unknown CPU model numbers.
28912 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28913 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28914 PROCESSOR_max sanity check.
28915 * config/s390/s390-opts.h (enum processor_type): Add
28917 * config/s390/s390.c (processor_table): Add arch12.
28918 (s390_expand_builtin): Add check for B_VXE flag.
28919 (s390_issue_rate): Add PROCESSOR_ARCH12.
28920 (s390_get_sched_attrmask): Likewise.
28921 (s390_get_unit_mask): Likewise.
28922 (s390_sched_score): Enable z13 scheduling for arch12.
28923 (s390_sched_reorder): Likewise.
28924 (s390_sched_variable_issue): Likewise.
28925 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28927 (s390_tune_attr): Use z13 scheduling also for arch12.
28928 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28929 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28930 (TARGET_VXE_P): New macros.
28931 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28932 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28933 * config/s390/s390.opt: Add arch12 as processor_type.
28935 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28937 * config/s390/s390.md
28938 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28939 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28940 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28942 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28943 Rename expanders to ...
28945 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28946 ("fixuns_truncdddi2_emu"): ... these.
28948 ("fixuns_trunc<mode>si2_emu"): New expander.
28950 ("*fixuns_truncdfdi2_z13"): Rename to ...
28951 ("*fixuns_truncdfdi2_vx"): ... this.
28953 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28955 * config/s390/2964.md: Remove the single element vector compare
28956 instructions which are no longer used.
28957 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28959 (s390_canonicalize_comparison): Remove handling of DFmode
28961 (s390_expand_vec_compare_scalar): Remove function.
28962 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28963 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28965 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28969 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28971 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28972 will anyway by matched by mov<mode>_64dfp.
28974 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28976 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28977 vlef/vstef. Add missing operand to vleif.
28979 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28981 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28982 pair for all vector types with 64 bit elements.
28983 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28984 * config/s390/vector.md (V_HW_64): ... here.
28985 (V_128_NOSINGLE): New mode iterator.
28986 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28987 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28988 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28989 ("*vec_load_pairv2di"): Change to ...
28990 ("*vec_load_pair<mode>"): ... this one.
28992 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28994 * config/s390/constraints.md: Add comments.
28995 (jKK): Reject element sizes > 8 bytes.
28996 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28998 * config/s390/s390.md: Add the s_operand checks formerly in
28999 s390_split_ok_p to various splitters where they are still
29001 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29002 for 128 bit vectors. Plus two splitters.
29004 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29006 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29009 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29012 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29013 error if the boundary argument is not constant.
29015 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29017 PR rtl-optimization/80112
29018 * loop-doloop.c (doloop_condition_get): Don't check condition
29019 if cmp isn't SET with IF_THEN_ELSE src.
29021 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29023 PR tree-optimization/80158
29024 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29025 replacing a candidate statement, also replace it for the
29026 candidate's alternate interpretation.
29027 (replace_rhs_if_not_dup): Likewise.
29028 (replace_one_candidate): Likewise.
29030 2017-03-24 Richard Biener <rguenther@suse.de>
29032 PR tree-optimization/80167
29033 * graphite-isl-ast-to-gimple.c
29034 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29036 (translate_isl_ast_to_gimple::get_rename): Likewise.
29038 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29040 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29041 handling of certain combinations of target options, including the
29042 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29043 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29045 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29048 * config/arm/arm.md (*load_multiple): Add reload_completed to
29049 matching condition.
29051 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29052 Richard Biener <rguenth@suse.de>
29054 PR tree-optimization/79908
29055 PR tree-optimization/80136
29056 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29057 been cast away, gimplify_and_add suffices.
29059 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29061 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29063 2017-03-23 Richard Biener <rguenther@suse.de>
29065 PR tree-optimization/80032
29066 * gimplify.c (gimple_push_cleanup): Forced unconditional
29067 cleanups still have to go to the conditional_cleanups
29070 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29072 PR tree-optimization/80072
29073 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29075 (next_operand_entry_id): Change type to unsigned int.
29076 (sort_by_operand_rank): Make sure to return the right return value
29077 even if unsigned fields are bigger than INT_MAX.
29078 (struct oecount): Change cnt and id type to unsigned int.
29079 (oecount_hasher::equal): Formatting fix.
29080 (oecount_cmp): Make sure to return the right return value
29081 even if unsigned fields are bigger than INT_MAX.
29082 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29085 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29086 TREE_READONLY on result if writing it more than once.
29089 * doc/invoke.texi (-fsanitize=thread): Document that with
29090 -fnon-call-exceptions atomics are not able to throw
29094 * tsan.c: Include tree-eh.h.
29095 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29096 maybe_clean_or_replace_eh_stmt where needed.
29097 (instrument_memory_accesses): Add cfg_changed argument.
29098 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29099 if it returned true.
29100 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29102 PR rtl-optimization/63191
29103 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29104 wrapper function, moved the whole old content into ...
29105 (ix86_delegitimize_address_1): ... this. New inline function.
29106 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29107 true as last argument instead of ix86_delegitimize_address.
29109 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29111 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29112 cortexa57_branch_cost.
29114 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29116 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29118 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29121 * doc/md.texi (Constraints): Document wA constraint.
29122 * config/rs6000/constraints.md (wA): New.
29123 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29124 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29125 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29126 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29128 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29131 * gimplify.c (is_oacc_declared): New function.
29132 (oacc_default_clause): Use it to set default flags for acc declared
29133 variables inside parallel regions.
29134 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29135 declared variables.
29136 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29137 declare attribute to any decl as necessary.
29139 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29142 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29143 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29144 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29145 (arm_arch_lpae): This.
29146 * config/arm/arm.c (arm_arch7ve): Rename into ...
29147 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29148 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29151 2017-03-22 Martin Liska <mliska@suse.cz>
29154 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29155 error message instead of an ICE.
29157 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29159 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29161 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29163 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29165 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29168 2017-03-21 Martin Sebor <msebor@redhat.com>
29170 * doc/extend.texi: Use "cannot" instead of "can't."
29171 * doc/hostconfig.texi: Same.
29172 * doc/install.texi: Same.
29173 * doc/invoke.texi: Same.
29174 * doc/loop.texi: Same.
29175 * doc/md.texi: Same.
29176 * doc/objc.texi: Same.
29177 * doc/rtl.texi: Same.
29178 * doc/tm.texi: Same.
29179 * doc/tm.texi.in: Same.
29180 * doc/trouble.texi: Same.
29182 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29185 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29186 (collect_checksum_attributes): Set it.
29187 (die_checksum_ordered): Use it.
29189 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29191 PR tree-optimization/79908
29192 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29193 change: For a VA_ARG whose LHS has been cast away, use
29194 force_gimple_operand to construct the side effects.
29196 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29198 PR translation/80001
29199 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29200 more amenable to translation.
29201 (oacc_loop_auto_partitions): Likewise.
29203 2017-03-21 Marek Polacek <polacek@redhat.com>
29204 Martin Sebor <msebor@redhat.com>
29206 PR tree-optimization/80109
29207 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29208 on INTEGRAL_TYPE_P.
29210 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29211 Segher Boessenkool <segher@kernel.crashing.org>
29214 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29215 check reg_used_between_p between insn and one of succ or succ2
29216 depending on if succ is artificial insn not inserted into insn
29219 2017-03-21 Martin Liska <mliska@suse.cz>
29221 PR gcov-profile/80081
29222 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29223 * doc/gcc.texi: Include gcov-dump stuff.
29224 * doc/gcov-dump.texi: New file.
29226 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29228 PR rtl-optimization/79150
29229 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29230 conditional jump, if the jump is the last insn of the loop.
29232 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29233 Richard Biener <rguenth@suse.de>
29235 PR tree-optimization/79908
29236 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29237 been cast away, use force_gimple_operand to construct the side
29240 2017-03-21 Martin Liska <mliska@suse.cz>
29242 PR libfortran/79956
29243 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29246 2017-03-21 Brad Spengler <spender@grsecurity.net>
29249 * plugin.c (htab_hash_plugin): New function.
29250 (add_new_plugin): Use it and adjust.
29251 (parse_plugin_arg_opt): Adjust.
29252 (init_one_plugin): Likewise.
29254 2017-03-21 Richard Biener <rguenther@suse.de>
29256 PR tree-optimization/80032
29257 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29258 if set force the cleanup to happen unconditionally.
29259 (gimplify_target_expr): Push inserted clobbers with force_uncond
29260 to avoid them being removed by control-dependent DCE.
29262 2017-03-21 Richard Biener <rguenther@suse.de>
29264 PR tree-optimization/80122
29265 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29266 va_arg_pack_len when the inlined call stmt requires pack
29268 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29270 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29273 * tsan.c (instrument_builtin_call): If the memory model argument
29274 is not a constant, assume it is valid.
29277 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29280 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29282 PR rtl-optimization/79910
29283 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29284 if its dest is used by an insn before I2 (other than the combined
29285 insns themselves, which are properly handled already).
29287 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29290 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29292 * combine.c (record_used_regs): New static function.
29293 (try_combine): Handle situations where there is an additional
29294 instruction between I2 and I3 which needs to have a LOG_LINK
29298 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29300 * combine.c (try_combine): Delete redundant i1 test. Call
29301 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29303 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29306 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29307 alternatives 13/14.
29309 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29311 PR tree-optimization/80054
29312 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29313 the optimization if a PHI or any of its arguments is not dominated
29314 by the candidate's basis. Use gphi* rather than gimple* as
29316 (replace_profitable_candidates): Clean up a gimple* variable that
29317 should be a gphi* variable.
29319 2017-03-20 Martin Sebor <msebor@redhat.com>
29322 * doc/extend.texi (attribute constructor): Document present limitation.
29324 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29327 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29328 __POWER9_VECTOR__ #ifdef control, change template definition to
29329 use Power9-specific built-in function.
29330 (vec_any_eq): Likewise.
29331 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29332 to control outcomes from this test.
29333 (vector_ae_<mode>p): For VEC_F modes, likewise.
29335 2017-03-20 Ian Lance Taylor <iant@google.com>
29337 * config/i386/i386.c (ix86_function_regparm): Save an extra
29338 register for -fsplit-stack with DECL_STATIC_CHAIN.
29340 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29343 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29344 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29346 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29348 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29350 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29353 2017-03-20 Marek Polacek <polacek@redhat.com>
29356 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29358 2017-03-20 Richard Biener <rguenther@suse.de>
29360 PR tree-optimization/80113
29361 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29362 allocate extra SSA name for PHI def.
29363 (add_close_phis_to_outer_loops): Likewise.
29364 (add_close_phis_to_merge_points): Likewise.
29365 (copy_loop_close_phi_args): Likewise.
29366 (copy_cond_phi_nodes): Likewise.
29368 2017-03-20 Martin Liska <mliska@suse.cz>
29370 PR middle-end/79753
29371 * tree-chkp.c (chkp_build_returned_bound): Do not build
29372 returned bounds for a LHS that's not a BOUNDED_P type.
29374 2017-03-20 Martin Liska <mliska@suse.cz>
29378 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29379 COMPLEX_CST and VECTOR_CST.
29381 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29384 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29385 target operand. A new splitter adds the clobber statement in case
29386 the target operand is dead anyway.
29388 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29390 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29391 to age-old versions of binutils and glibc.
29393 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29395 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29397 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29399 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29401 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29403 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29404 requirement for binutils 2.13.
29406 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29408 * combine.c (try_combine): Delete redundant i1 test. Call
29409 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29411 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29413 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29414 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29416 <riscv64-*-elf>: Re-arrange section
29417 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29418 <riscv32-*-linux>: Likewise.
29419 <riscv64-*-elf>: Likewise
29420 <riscv64-*-linux>: Likewise.
29422 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29425 * aarch64.opt(verbose-cost-dump): Fix typo.
29427 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29430 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29431 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29433 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29435 * reload.c (find_reloads): When reloading a nonoffsettable address,
29436 use RELOAD_OTHER for it and its address reloads.
29438 PR rtl-optimization/79910
29439 * combine.c (record_used_regs): New static function.
29440 (try_combine): Handle situations where there is an additional
29441 instruction between I2 and I3 which needs to have a LOG_LINK
29444 2017-03-17 Jeff Law <law@redhat.com>
29446 PR tree-optimization/71437
29447 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29448 conditional in the hash table first.
29449 (vrp_dom_walker::before_dom_children): Extract condition from
29450 ASSERT_EXPR. Record condition, its inverion and any implied
29451 conditions as well.
29453 2017-03-17 Marek Polacek <polacek@redhat.com>
29454 Markus Trippelsdorf <markus@trippelsdorf.de>
29456 PR tree-optimization/80079
29457 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29460 2017-03-17 Richard Biener <rguenther@suse.de>
29462 PR middle-end/80075
29463 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29464 Properly verify the LHS before the RHS possibly claims to be
29466 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29469 2017-03-17 Martin Jambor <mjambor@suse.cz>
29471 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29472 (List of -O2 options): Likewise.
29473 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29476 2017-03-17 Tom de Vries <tom@codesourcery.com>
29478 * gcov-dump.c (print_usage): Print bug_report_url.
29480 2017-03-17 Richard Biener <rguenther@suse.de>
29482 PR middle-end/80050
29483 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29484 (parser::peek): Likewise.
29486 2017-03-17 Richard Biener <rguenther@suse.de>
29488 PR tree-optimization/80048
29489 * sese.c (free_sese_info): Properly release rename_map and
29490 copied_bb_map elements.
29492 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29494 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29495 Add linked-list forward and backlinks. Insert on
29496 construction, remove on destruction.
29497 (class pass_store_merging): Add m_stores_head field.
29498 (pass_store_merging::terminate_and_process_all_chains):
29499 Iterate over m_stores_head list.
29500 (pass_store_merging::terminate_all_aliasing_chains):
29502 (pass_store_merging::execute): Check for debug stmts first.
29503 Push new chains onto the m_stores_head stack.
29505 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29508 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29509 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29510 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29512 2017-03-16 Jeff Law <law@redhat.com>
29514 PR tree-optimization/71437
29515 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29516 member function. Implementation moved into after_dom_children
29517 member function and into the threader's thread_outgoing_edges
29519 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29520 some code into new thread_outgoing_edges.
29521 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29522 definition. Simplify marker handling (do it here). Assume we always
29523 have the available expression and the const/copies tables.
29524 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29526 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29527 * tree-vrp.c (equiv_stack): No longer file scoped.
29528 (vrp_dom_walker): New class.
29529 (vrp_dom_walker::before_dom_children): New member function.
29530 (vrp_dom_walker::after_dom_children): Likewise.
29531 (identify_jump_threads): Setup domwalker. Use it rather than
29532 walking edges in a random order by hand. Simplify setup/finalization.
29533 (finalize_jump_threads): Remove.
29534 (vrp_finalize): Do not call identify_jump_threads here.
29535 (execute_vrp): Do it here instead and call thread_through_all_blocks
29538 PR tree-optimization/71437
29539 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29541 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29543 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29544 (dom_opt_dom_walker::thread_across_edge): Remove
29545 handle_dominating_asserts argument. All callers changed.
29546 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29547 changes. Remove calls to lhs_of_dominating_assert. Other
29548 uses of handle_dominating_asserts turn into unconditional code
29549 (simplify_control_stmt_condition_1): Likewise.
29550 (simplify_control_stmt_condition): Likewise.
29551 (thread_through_normal_block, thread_across_edge): Likewise.
29552 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29553 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29554 object if it is not an SSA_NAME.
29555 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29556 before calling into the VRP specific simplifiers.
29557 (identify_jump_threads): Remove handle_dominating_asserts
29560 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29563 * tree-diagnostic.c (default_tree_printer): No longer static.
29564 * tree-diagnostic.h (default_tree_printer): New prototype.
29566 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29568 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29569 Change ins into fmov.
29571 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29573 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29574 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29575 Use h_con constraint for operand 1.
29576 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29577 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29579 2017-03-15 Jeff Law <law@redhat.com>
29581 PR tree-optimization/71437
29582 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29583 (record_temporary_equivalences): Use it.
29585 PR tree-optimization/71437
29586 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29587 tree-ssa-scopedtables.
29588 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29589 (record_conditions, record_cond, vuse_eq): Likewise.
29590 (record_edge_info): Adjust to API tweak of record_conditions.
29591 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29592 (record_temporary_equivalences, optimize_stmt): Likewise.
29593 (eliminate_redundant_computations): Likewise.
29594 (record_equivalences_from_stmt): Likewise.
29595 * tree-ssa-scopedtables.c: Include options.h and params.h.
29596 (vuse_eq): New function, moved from tree-ssa-dom.c
29597 (build_and_record_new_cond): Likewise.
29598 (record_conditions): Likewise. Accept vector of conditions rather
29599 than edge_equivalence structure for first argument.
29600 for the first argument.
29601 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29602 from tree-ssa-dom.c.
29603 (avail_exprs_stack::record_cond): Likewise.
29604 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29605 from tree-ssa-dom.c.
29606 (avail_exprs_stack): Add new member functions lookup_avail_expr
29608 (record_conditions): Declare.
29610 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29613 * lra-constraints.c (process_alt_operands): Increase reject for
29614 reloading an input/output operand.
29616 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29619 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29620 insns to convert from signed/unsigned char/short to IEEE 128-bit
29622 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29624 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29627 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29628 subreg of inner mode for values already in registers.
29630 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29632 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29633 iteration reg is used after the loop.
29635 2017-03-14 Martin Sebor <msebor@redhat.com>
29637 PR tree-optimization/79800
29638 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29639 precision in negative-positive range.
29640 (format_floating): Call non-const overload with adjusted precision.
29642 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29645 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29648 2017-03-14 Martin Sebor <msebor@redhat.com>
29650 PR middle-end/80020
29651 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29652 * builtins.def (aligned_alloc): Use it.
29655 * Makefile.in (GTFILES): Add calls.c.
29656 * calls.c: Include "gt-calls.h".
29658 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29660 PR rtl-optimization/79728
29661 * regs.h (struct target_regs): New field
29662 x_contains_allocatable_regs_of_mode.
29663 (contains_allocatable_regs_of_mode): New macro.
29664 * reginfo.c (init_reg_sets_1): Initialize it, and change
29665 contains_reg_of_mode so it includes global regs as well.
29666 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29667 rather than contains_regs_of_mode.
29669 2017-03-14 Martin Liska <mliska@suse.cz>
29671 * doc/invoke.texi: Document options that can't be combined with
29672 -fcheck-pointer-bounds.
29674 2017-03-14 Martin Liska <mliska@suse.cz>
29676 PR middle-end/79831
29677 * doc/invoke.texi (-Wchkp): Document the option.
29679 2017-03-14 Martin Liska <mliska@suse.cz>
29681 * Makefile.in: Install gcov-dump.
29683 2017-03-14 Martin Liska <mliska@suse.cz>
29685 * multiple_target.c (expand_target_clones): Bail out for
29686 an invalid attribute.
29688 2017-03-14 Richard Biener <rguenther@suse.de>
29690 * alias.c (struct alias_set_entry): Pack properly.
29691 * cfgloop.h (struct loop): Likewise.
29692 * cse.c (struct set): Likewise.
29693 * ipa-utils.c (struct searchc_env): Likewise.
29694 * loop-invariant.c (struct invariant): Likewise.
29695 * lra-remat.c (struct cand): Likewise.
29696 * recog.c (struct change_t): Likewise.
29697 * rtl.h (struct address_info): Likewise.
29698 * symbol-summary.h (function_summary): Likewise.
29699 * tree-loop-distribution.c (struct partition): Likewise.
29700 * tree-object-size.c (struct object_size_info): Likewise.
29701 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29702 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29703 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29704 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29705 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29706 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29707 (struct _stmt_vec_info): Likewise.
29709 2017-03-14 Martin Liska <mliska@suse.cz>
29712 * multiple_target.c (create_dispatcher_calls): Check that
29713 a target can create a function dispatcher.
29715 2017-03-14 Martin Liska <mliska@suse.cz>
29718 * multiple_target.c (expand_target_clones): Drop local.local
29719 flag for default implementation.
29721 2017-03-14 Richard Biener <rguenther@suse.de>
29723 PR tree-optimization/80030
29724 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29726 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29728 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29729 gcc_fallthrough() instead of __attribute__((fallthrough));
29731 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29733 * doc/gcc.texi: Remove "up" link to (DIR).
29734 * doc/gccint.texi: Ditto.
29736 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29738 * doc/install.texi (Specific) <avr>: Remove reference to
29741 2017-03-13 Jeff Law <law@redhat.com>
29743 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29744 attribute rather than comments.
29746 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29747 match_scratch operand is highest.
29749 2017-03-13 Martin Liska <mliska@suse.cz>
29751 PR middle-end/78339
29752 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29753 is a CHKP clone, use original declaration.
29755 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29757 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29758 (arc_conditional_register_usage): Use a different allocation order
29759 when optimizing for size.
29760 * common/config/arc/arc-common.c (arc_option_optimization_table):
29761 Section anchors default on when optimizing for size.
29763 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29765 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29767 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29769 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29770 * config/arc/arc.md (cpu_facility): Add cd variant.
29771 (*movqi_insn): Add code density variant.
29772 (*movhi_insn): Likewise.
29773 (*movqi_insn): Likewise.
29774 (*addsi3_mixed): Likewise.
29775 (subsi3_insn): Likewise.
29777 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29779 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29781 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29783 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29784 expressions with MINUS and UNARY ops.
29786 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29789 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29791 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29792 between vec_select and vector argument.
29793 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29794 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29795 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29796 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29797 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29798 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29800 2017-03-13 Richard Biener <rguenther@suse.de>
29803 * params.def (vect-max-peeling-for-alignment): Fix typo.
29805 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29807 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29808 issue that only occurred with binutils below 2.18.
29810 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29812 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29813 refer to binutils 2.11/2.12 minimum.
29815 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29817 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29818 ftp.kernel.org and simplify binutils requirement.
29820 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29822 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29824 (Optimize Options): Ditto. Also remove redundancy.
29826 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29828 PR translation/79848
29829 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29831 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29832 to G_ to avoid double translation.
29834 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29836 PR translation/79923
29837 * auto-profile.c (get_combined_location): Convert leading
29838 character of diagnostics to lower case and remove trailing period.
29839 (read_profile): Likewise for various diagnostics.
29840 * config/arm/arm.c (arm_option_override): Remove trailing period
29841 from various diagnostics.
29842 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29843 (msp430_expand_delay_cycles): Likewise.
29845 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29848 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29849 full command-line argument, rather than just "str".
29850 (aarch64_validate_march): Likewise.
29851 (aarch64_validate_mtune): Likewise.
29853 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29855 PR rtl-optimization/78911
29856 * lra-assigns.c (must_not_spill_p): New function.
29857 (spill_for): Use it.
29859 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29861 PR tree-optimization/79981
29862 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29863 ATOMIC_COMPARE_EXCHANGE ifn result.
29864 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29865 IFN_ATOMIC_COMPARE_EXCHANGE.
29867 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29870 * opts.c (parse_sanitizer_options): Add missing question mark to
29871 "did you mean" message.
29873 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29875 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29877 (VMULEUH_UNS): Likewise.
29878 (VMULOUB_UNS): Likewise.
29879 (VMULOUH_UNS): Likewise.
29880 * config/rs6000/rs6000.c (builtin_function_type): Remove
29881 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29883 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29886 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29887 x with result of extra_parsing_for_operand_code_0.
29888 (function_reader::extra_parsing_for_operand_code_0): Convert
29889 return type from void to rtx, returning x. When reading
29890 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29891 larger size containing struct block_symbol.
29893 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29896 -mfloat128-hardware without -m64.
29898 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29901 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29902 entries to the case statement that marks unsigned arguments to
29903 overloaded functions.
29905 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29907 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29908 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29910 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29913 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29914 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29916 2017-03-10 Martin Liska <mliska@suse.cz>
29920 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29921 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29924 2017-03-10 Olivier Hainque <hainque@adacore.com>
29926 * tree-switch-conversion (array_value_type): Start by resetting
29927 candidate type to it's main variant.
29929 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29931 PR rtl-optimization/79909
29932 * combine.c (try_combine): Use simplify_replace_rtx on individual
29933 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29934 of the whole CALL_INSN_FUNCTION_USAGE.
29936 PR tree-optimization/79972
29937 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29938 get_range_info on SSA_NAMEs. Formatting fixes.
29940 2017-03-10 Richard Biener <rguenther@suse.de>
29941 Jakub Jelinek <jakub@redhat.com>
29943 PR tree-optimization/77975
29944 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29945 edge to be constant.
29946 (get_val_for): For constant x return it. Formatting fix.
29947 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29948 would use the same bases as the current one.
29950 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29952 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29953 instead of vec_select for V1TImode.
29954 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29956 (VSX_LE_128): Add V1TI to this mode iterator.
29957 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29958 (*vsx_le_perm_store_<mode>): Likewise.
29959 (pre-reload splitter for VSX stores): Likewise.
29960 (post-reload splitter for VSX stores): Likewise.
29961 (*vsx_xxpermdi2_le_<mode>): Likewise.
29962 (*vsx_lxvd2x2_le_<mode>): Likewise.
29963 (*vsx_stxvd2x2_le_<mode>): Likewise.
29965 2017-03-09 Michael Eager <eager@eagercon.com>
29967 Correct failures with --enable-checking=yes,rtl.
29969 * config/microblaze/microblaze.c (microblaze_expand_shift):
29970 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29971 test for const0_rtx.
29972 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29973 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29975 2017-03-09 Richard Biener <rguenther@suse.de>
29977 PR tree-optimization/79977
29978 * graphite-scop-detection.c (scop_detection::merge_sese):
29979 Handle the case of extra exits to blocks dominating the entry.
29981 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29983 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29986 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29988 PR rtl-optimization/79949
29989 * lra-constraints.c (process_alt_operands): Check memory when
29990 trying to predict a cycle. Print about the overall increase.
29992 2017-03-09 Richard Biener <rguenther@suse.de>
29994 PR middle-end/79971
29995 * gimple-expr.c (useless_type_conversion_p): Preserve
29996 TYPE_SATURATING for fixed-point types.
29998 2017-03-09 Richard Biener <rguenther@suse.de>
30001 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30002 alignment of BLKmode params.
30004 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30007 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30008 (VALL_NO_V2Q): Likewise.
30010 * config/aarch64/aarch64-simd.md
30011 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30013 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30014 VALL_NO_V2Q mode iterator.
30015 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30017 2017-03-09 Martin Liska <mliska@suse.cz>
30019 PR tree-optimization/79631
30020 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30021 tree_int_cst_sign_bit just for INTEGER constants.
30023 2017-03-09 Martin Liska <mliska@suse.cz>
30027 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30030 2017-03-09 Marek Polacek <polacek@redhat.com>
30033 * tree.c (inchash::add_expr): Handle TREE_VEC.
30035 2017-03-09 Martin Liska <mliska@suse.cz>
30038 (chkp_narrow_size_and_offset): New function.
30039 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30040 (void chkp_parse_bit_field_ref): New function.
30041 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30042 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30044 2017-03-09 Martin Liska <mliska@suse.cz>
30047 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30048 (chkp_find_bounds_1): Remove gcc_unreachable.
30050 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30053 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30054 BUILT_IN_SYNC*, determine the access type from the size suffix and
30055 always build a MEM_REF with that type. Handle forgotten
30056 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30059 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30060 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30061 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30062 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30063 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30064 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30065 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30066 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30067 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30068 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30069 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30070 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30071 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30072 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30073 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30074 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30075 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30076 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30077 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30078 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30079 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30080 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30081 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30082 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30083 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30084 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30085 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30086 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30087 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30088 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30089 definitions outside of __OPTIMIZE__ guarded section.
30092 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30093 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30094 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30095 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30098 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30100 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30101 ("vfenez<mode>"): Add missing constraints.
30103 2017-03-08 Martin Sebor <msebor@redhat.com>
30106 * config/nds32/nds32.c (nds32_option_override):
30107 Fix misspelled diagnostic.
30109 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30112 * gimplify.c (gimplify_omp_for): Replace index var in outer
30113 taskloop statement with an artificial variable and add
30114 OMP_CLAUSE_PRIVATE clause for it.
30116 2017-03-08 Richard Biener <rguenther@suse.de>
30118 PR tree-optimization/79955
30119 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30120 for accesses that are completely outside of the variable.
30122 2017-03-08 Andrew Haley <aph@redhat.com>
30124 PR tree-optimization/79943
30125 * tree-ssa-loop-split.c (compute_new_first_bound): When
30126 calculating the new upper bound, (END-BEG) should be added, not
30129 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30131 * config/avr/avr.md (setmemhi): Make sure match_dup
30132 operand number comes before match_scratch.
30134 2017-03-08 Richard Biener <rguenther@suse.de>
30136 PR tree-optimization/79920
30137 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30138 with ncopies == 1 to ...
30139 (vect_transform_slp_perm_load): ... here. Properly compute
30140 all element loads by iterating VF times over the group. Do
30141 not handle ncopies (computed in a broken way) in
30142 vect_create_mask_and_perm.
30144 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30147 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30148 is a uniform vector, use uniform_vector_p return value instead of
30149 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30151 2017-03-07 Marek Polacek <polacek@redhat.com>
30153 PR middle-end/79809
30154 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30155 (alloca_call_type): Likewise.
30157 2017-03-07 Martin Liska <mliska@suse.cz>
30159 * gcov.c (process_args): Put comment to correct location.
30161 2017-03-07 Martin Liska <mliska@suse.cz>
30163 PR middle-end/68270
30164 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30165 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30166 (chkp_narrow_bounds_for_field): Likewise.
30167 (chkp_parse_array_and_component_ref): Pass one more argument to
30170 2017-03-07 Richard Biener <rguenther@suse.de>
30172 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30175 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30177 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30178 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30180 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30183 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30184 to end of description.
30185 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30187 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30189 PR rtl-optimization/79901
30190 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30192 (*avx512f_<code><mode>3<mask_name>): ... this.
30193 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30194 iterator instead of VI8_AVX2_AVX512BW.
30196 PR rtl-optimization/79901
30197 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30198 min/max expander, expand it using expand_vec_cond_expr.
30201 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30204 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30207 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30208 to void * for PCH reasons.
30209 * dwarf2out.c (output_loc_operands, output_die): Cast
30210 v.val_vec.array to unsigned char *.
30212 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30215 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30218 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30220 PR rtl-optimization/79571
30221 * lra-constraints.c (process_alt_operands): Calculate static
30222 reject and subtract it from overall when only addresses will be
30225 2017-03-06 Julia Koval <julia.koval@intel.com>
30228 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30229 incoming stack boundary to 128 for 64-bit targets.
30231 2017-03-06 Richard Biener <rguenther@suse.de>
30233 PR tree-optimization/79894
30234 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30235 to NULL after folding it.
30237 2017-03-06 Richard Biener <rguenther@suse.de>
30239 PR tree-optimization/79824
30240 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30241 check disabling peeling for gaps.
30243 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30245 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30246 attributes): Document gettimeofday.
30248 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30250 * config/s390/s390.c (s390_option_override_internal): Set
30251 PARAM_MIN_VECT_LOOP_BOUND
30253 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30255 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30256 * config/s390/s390.md: Likewise.
30258 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30261 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30262 (<avx2_avx512>_perm<mode>): Rename to ...
30263 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30265 (<avx512>_perm<mode>_mask): Rename to ...
30266 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30268 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30269 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30270 instead of VI8F_256_512.
30271 (avx512f_perm<mode>): New define_expand.
30272 (avx512f_perm<mode>_mask): Likewise.
30273 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30274 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30276 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30278 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30279 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30281 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30283 2017-03-06 Martin Liska <mliska@suse.cz>
30286 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30287 when having a SSA NAME w/o VAR_DECL assigned to it.
30289 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30291 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30292 msa_dpsub_<su>_d): Fix MODE for vec_select.
30294 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30296 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30298 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30300 2017-03-06 Richard Biener <rguenther@suse.de>
30302 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30303 * plugin.c (register_plugin_info): Likewise.
30304 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30306 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30308 * config/i386/sse.md (sse_storehps, sse_storelps,
30309 avx_<castmode><avxsizesuffix>_<castmode>,
30310 avx512f_<castmode><avxsizesuffix>_<castmode>,
30311 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30312 in condition that at least one operand is not a MEM.
30314 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30316 PR middle-end/79805
30317 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30318 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30320 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30321 gimple_call_nothrow_p flag based on whether original builtin can throw.
30322 If it can, emit following stmts on the fallthrough edge.
30323 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30324 don't create new bb if inserting just debug stmts on the edge, try to
30325 insert them on the fallthru bb or just reset debug stmts.
30327 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30330 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30331 restore recog_data (including the operand rtxes inside it) around
30332 the call to get_insn_template.
30334 2017-03-03 Martin Sebor <msebor@redhat.com>
30336 PR tree-optimization/79699
30337 * context.c (context::~context): Free MPFR caches to avoid
30338 a memory leak on program exit.
30340 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30342 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30343 Use wide_int::ulow () instead of .elt (0).
30345 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30347 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30348 (*pushxf): Limit oF constraint to 32bit targets and add oC
30349 constraint for 64bit targets.
30350 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30351 (*pushdf): Change rmF constraint to rmC.
30353 2017-03-03 Martin Liska <mliska@suse.cz>
30355 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30356 Remove unused variable.
30358 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30361 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30362 is a memory operand, increase num_memory.
30363 (ix86_expand_args_builtin): Likewise.
30365 2017-03-03 Jan Hubicka <jh@suse.cz>
30368 * ipa-devirt.c (maybe_record_node): Properly handle
30369 __cxa_pure_virtual visibility.
30371 2017-03-03 Martin Liska <mliska@suse.cz>
30373 PR tree-optimization/79803
30374 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30376 (pass_loop_prefetch::execute): Disabled optimization if an
30377 assumption about L1 cache size is not met.
30379 2017-03-03 Martin Liska <mliska@suse.cz>
30381 PR rtl-optimization/79574
30382 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30383 (hash_scan_set): Likewise.
30384 (dump_hash_table): Likewise.
30385 (hoist_code): Likewise.
30387 2017-03-03 Richard Biener <rguenther@suse.de>
30389 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30390 in place of uhigh ().
30391 (fixed_convert_from_real): Likewise.
30393 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30396 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30398 2017-03-03 Richard Biener <rguenther@suse.de>
30400 PR middle-end/79818
30401 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30402 TYPE_OVERFLOW_UNDEFINED check.
30404 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30406 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30408 (vector_ae_<mode>_p): Likewise.
30409 (vector_nez_<mode>_p): Likewise.
30410 (vector_ne_v2di_p): Likewise.
30411 (vector_ae_v2di_p): Likewise.
30412 (vector_ne_<mode>_p): Likewise.
30413 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30415 (vsx_tsqrt<mode>2_fe): Likewise.
30417 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30420 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30422 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30424 PR rtl-optimization/79780
30425 * cprop.c (one_cprop_pass): When second and further conditional trap
30426 in a single basic block is turned into an unconditional trap, turn it
30427 into a deleted note to avoid RTL verification failures.
30429 2017-03-02 Richard Biener <rguenther@suse.de>
30431 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30433 2017-03-02 Richard Biener <rguenther@suse.de>
30435 PR tree-optimization/79345
30437 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30438 param and abort the walk, returning -1 if it is hit.
30439 (walk_aliased_vdefs): Take a limit param and pass it on.
30440 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30441 defaulting to 0 and return a signed int.
30442 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30443 (check_defs): New helper.
30444 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30445 about uninitialized memory.
30446 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30447 bogus uninitialized warning.
30448 (fixed_convert_from_real): Likewise.
30450 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30452 PR tree-optimization/66768
30453 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30454 iv_use if base object can't be determined.
30456 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30458 PR tree-optimization/79345
30459 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30460 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30461 (get_pattern_stats): Initialize it.
30462 * genemit.c (gen_expand): Verify match_scratch numbers come after
30463 match_operand/match_dup numbers.
30464 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30465 match_scratch numbers.
30466 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30468 * config/s390/s390.md (trunctdsd2): Likewise.
30470 2017-03-02 Richard Biener <rguenther@suse.de>
30472 * wide-int.h (wide_int_storage::operator=): Implement in terms
30475 2017-03-02 Richard Biener <rguenther@suse.de>
30477 PR tree-optimization/79777
30478 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30479 the to insert expression to sth existing.
30481 2017-03-01 Martin Sebor <msebor@redhat.com>
30483 PR middle-end/79692
30484 * gimple-ssa-sprintf.c
30485 (directive::known_width_and_precision): New function.
30486 (format_integer): Use it.
30487 (get_mpfr_format_length): Consider the full range of precision
30488 when computing %g output with the # flag. Set the likely byte
30489 count to 3 rather than 1 when precision is indeterminate.
30490 (format_floating): Correct the lower bound of precision.
30492 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30494 * doc/invoke.texi: Document default code model for 64-bit Linux.
30496 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30499 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30500 udiv rather than div since input pattern is unsigned.
30502 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30504 * config/i386/i386.c (print_reg): Warn for values of
30505 unsupported size in integer register.
30507 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30510 * config/rs6000/predicates.md (current_file_function_operand): Do
30511 not allow self calls to be local if the function is replaceable.
30513 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30516 * config/rs6000/altivec.h (vec_ctz and others): Change the
30517 preprocessor macro that controls conditional compilation from
30518 _ARCH_PWR9 to __POWER9_VECTOR__.
30519 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30520 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30521 control (instead of _ARCH_PWR9 control) so that template
30522 definition uses power9-specific function.
30523 (vec_any_eq): Likewise.
30524 (vec_all_ne): Change macro definition to use a power9-specific
30525 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30526 _ARCH_PWR9 control).
30527 (vec_any_eq) Likewise.
30528 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30529 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30530 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30531 support for xvcmpnedp instruction.
30532 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30533 macro expansion so that Power9 implementation of vec_all_ne does
30534 not use the AltiVec predicate framework.
30535 (VCMPNEH_P): Likewise.
30536 (VCMPNEW_P): Likewise.
30537 (VCMPNED_P): Likewise.
30538 (VCMPNEFP_P): Likewise.
30539 (VCMPNEDP_P): Likewise.
30540 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30541 implementation of vec_any_eq to not use AltiVec predicate
30543 (VCMPAEH_P): Likewise.
30544 (VCMPAEW_P): Likewise.
30545 (VCMPAED_P): Likewise.
30546 (VCMPAEFP_P): Likewise.
30547 (VCMPAEDP_P): Likewise.
30548 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30549 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30550 not use the AltiVec predicate framework.
30551 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30552 of vec_any_eq to not use AltiVec predicate framework.
30553 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30554 support for predefined __POWER9_VECTOR__ macro to indicate that
30555 Power9 instruction selection is enabled.
30556 (altivec_overloaded_builtins): Remove extraneous
30557 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30558 function argument types RS6000_BTI_bool_V16QI and
30559 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30560 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30561 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30562 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30563 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30564 Power9 for implementations of vec_cmpne. Change the signature for
30565 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30566 (representing vec_all_ne) to remove the previously described first
30567 argument of type RS6000_BTI_INTSI, as this was an artifact of
30568 reliance on the AltiVec predicate framework, which is no longer
30569 used in the implementation of these functions. Add
30570 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30571 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30572 since, unlike the AltiVec predicate framework implementation, we
30573 do not share function descriptors between vec_alle and vec_anyeq.
30574 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30575 set of modes that receive special treatment even when
30576 TARGET_P9_VECTOR is true. The special treatment emits code that
30577 does not depend on Power9 instructions.
30578 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30579 define_expand to not rely on AltiVec predicate framework.
30580 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30582 (vector_ne_v2di_p): Change this define_expand to not rely on
30583 AltiVec predicate framework.
30584 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30586 (vector_ne_<mode>_p): Change this define_expand to not rely on
30587 AltiVec predicate framework.
30588 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30590 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30591 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30592 define_insn pattern.
30593 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30594 define_insn pattern because the xvcmpne<VSs>. instruction is not
30596 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30597 instruction is not supported.
30599 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30601 * config/nvptx/nvptx.c: Include intl.h.
30603 2017-03-01 Martin Jambor <mjambor@suse.cz>
30606 * ipa-prop.h (ipa_bits): Removed field known.
30607 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30608 to pointers. Adjusted their comments to warn about their sharing.
30609 (ipcp_transformation_summary): Change bits to a vector of pointers.
30610 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30611 (ipa_get_ipa_bits_for_value): Declare.
30612 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30613 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30614 (ipa_bits_hash_table): Likewise.
30615 (ipa_vr_ggc_hash_traits): Likewise.
30616 (ipa_vr_hash_table): Likewise.
30617 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30618 being pointers and vr_known being removed.
30619 (ipa_set_jf_unknown): Likewise.
30620 (ipa_get_ipa_bits_for_value): New function.
30621 (ipa_set_jfunc_bits): Likewise.
30622 (ipa_get_value_range): New overloaded functions.
30623 (ipa_set_jfunc_vr): Likewise.
30624 (ipa_compute_jump_functions_for_edge): Use the above functions to
30625 construct bits and vr parts of jump functions.
30626 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30627 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30629 (ipcp_grow_transformations_if_necessary): Also allocate
30630 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30632 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30633 them. Fix too long lines.
30634 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30635 vr_known being removed.
30636 (ipa_read_jump_function): Use new setter functions to construct bits
30637 and vr parts of jump functions or set them to NULL.
30638 (write_ipcp_transformation_info): Adjust for bits being pointers.
30639 (read_ipcp_transformation_info): Likewise.
30640 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30642 Include gt-ipa-prop.h.
30643 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30645 (ipcp_store_bits_results): Likewise.
30646 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30647 Do not write to existing jump functions but use a temporary instead.
30649 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30652 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30653 attempt to use its first operand as BIT_FIELD_REF base.
30655 2017-03-01 Richard Biener <rguenther@suse.de>
30657 PR middle-end/79721
30658 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30659 interpolating formula in wrapping arithmetic.
30660 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30662 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30664 PR tree-optimization/79734
30665 * tree-vect-generic.c (expand_vector_condition): Optimize
30666 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30667 Handle VEC_COND_EXPR where comparison has different inner width from
30668 type's inner width.
30670 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30672 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30673 markup, and similar issues. Remove @opindex entries for things
30674 that aren't options. Add missing -mmpy-option entries.
30676 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30678 PR tree-optimization/79737
30679 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30680 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30681 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30682 instead of byte_size. Formatting fix.
30683 (shift_bytes_in_array_right): Formatting fix.
30685 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30688 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30689 condition on optimize for the leaf function test.
30691 2017-02-28 Martin Liska <mliska@suse.cz>
30694 * read-rtl-function.c (function_reader::handle_unknown_directive):
30695 Bail out when one uses -flto.
30697 2017-02-28 Martin Liska <mliska@suse.cz>
30699 * common.opt: Replace space with tabular for options of <number>
30701 * config/i386/i386.opt: Show <number> value for
30702 -mlarge-data-threshold.
30703 * opts.c (print_filtered_help): Do not display number in hexadecimal
30706 2017-02-28 Martin Liska <mliska@suse.cz>
30708 * common.opt: Fix --help=option -Q for options which are of
30711 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30713 * config/i386/i386.c (print_reg): Error out for values
30714 of 8-bit size in invalid integer register.
30716 2017-02-28 Martin Sebor <msebor@redhat.com>
30718 PR tree-optimization/79691
30719 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30721 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30724 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30725 gcc_unreachable with output_operand_lossage.
30727 2017-02-28 Richard Biener <rguenther@suse.de>
30729 PR tree-optimization/79740
30730 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30732 (visit_nary_op): Insert the nary into the hashtable if we
30733 pattern-matched sth.
30734 * tree-ssa-pre.c (eliminate_insert): Robustify.
30736 2017-02-28 Richard Biener <rguenther@suse.de>
30738 PR middle-end/79731
30739 * fold-const.c (decode_field_reference): Reject out-of-bound
30742 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30744 * config/i386/i386.c: Include intl.h.
30745 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30746 instead of just cond ? "..." : "...".
30747 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30748 * coverage.c (read_counts_file): Likewise.
30749 * omp-offload.c: Include intl.h.
30750 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30751 of just cond ? "..." : "...".
30752 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30753 of just cond ? "..." : "...".
30755 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30758 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30760 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30761 'tune for' CPU name.
30762 * config/arm/arm-cpu-data.h: Regenerated.
30764 2017-02-28 Richard Biener <rguenther@suse.de>
30766 PR tree-optimization/79732
30767 * tree-inline.c (expand_call_inline): Do not shadow var.
30769 2017-02-28 Richard Biener <rguenther@suse.de>
30771 PR tree-optimization/79723
30772 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30773 address-space properly.
30775 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30777 * doc/optinfo.texi (Optimization groups): Fix option used for
30779 * doc/invoke.texi (-fopt-info): Document "omp".
30780 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30781 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30782 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30783 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30784 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30786 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30788 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30789 OPTGROUP_OMP with "omp".
30791 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30794 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30795 for arithmetic shift of unsigned V2DI.
30797 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30799 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30800 arc/linux.h headers.
30801 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30802 (LINK_SPEC): Likewise.
30803 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30804 (EXTRA_SPECS): Likewise.
30805 (STARTFILE_SPEC): Likewise.
30806 (ENDFILE_SPEC): Likewise.
30807 (LIB_SPEC): Likewise.
30808 (TARGET_SDATA_DEFAULT): Likewise.
30809 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30810 (MULTILIB_DEFAULTS): Likewise.
30811 (DWARF2_UNWIND_INFO): Likewise.
30812 * config/arc/big.h: New file.
30813 * config/arc/elf.h: Likewise.
30814 * config/arc/linux.h: Likewise.
30815 * config/arc/t-uClibc: Remove.
30817 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30819 PR tree-optimization/77536
30820 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30821 (tree_transform_and_unroll_loop): Use above function to compute the
30822 estimated niter of unrolled loop and use it when scaling profile.
30823 Also use count info rather than frequency if it's non-zero.
30824 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30825 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30826 (vect_transform_loop): Call above function.
30828 2017-02-27 Richard Biener <rguenther@suse.de>
30830 PR tree-optimization/45397
30831 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30832 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30833 (visit_nary_op): Add pattern matching for CSEing sign-changed
30834 or truncated operations with wider ones.
30836 2017-02-27 Richard Biener <rguenther@suse.de>
30838 PR tree-optimization/79690
30839 * tree-vect-stmts.c (vectorizable_store): Use vector type
30840 built from the DR with address-space.
30842 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30844 * doc/invoke.texi (Optimize Options): Refine the description
30845 of asan-use-after-return.
30847 2017-02-25 Alan Modra <amodra@gmail.com>
30849 PR rtl-optimization/79584
30850 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30851 base, not ad->base_term, the reg within base. Remove assertion
30852 that ad->base == ad->base_term. Replace gen_int_mode using
30853 bogus mode with const0_rtx.
30855 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30857 PR middle-end/79396
30858 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30859 FMA_EXPR like tcc_binary or tcc_unary.
30861 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30864 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30866 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30867 (output_loc_operands): Handle DW_OP_call_ref and
30868 DW_OP_GNU_variable_value.
30869 (struct variable_value_struct): New type.
30870 (struct variable_value_hasher): Likewise.
30871 (variable_value_hash): New variable.
30872 (string_types): Remove.
30873 (copy_loc_descr): New function.
30874 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30875 (prepend_loc_descr_to_each): New function.
30876 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30877 instead of add_loc_descr_to_each if the first argument is single
30878 location list and the second has multiple.
30879 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30880 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30881 when looking for variable value which doesn't have other location info.
30882 (loc_list_from_tree): Formatting fix.
30883 (gen_array_type_die): Simplify DW_AT_string_length handling.
30884 (adjust_string_types): Remove.
30885 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30886 string_types. Call resolve_variable_values.
30887 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30888 (resolve_addr_in_expr): Likewise. Add A argument.
30889 (copy_deref_exprloc): Remove deref argument. Adjust for the
30890 original expression being DW_OP_GNU_variable_value with optionally
30891 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30892 optionally after it.
30893 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30894 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30895 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30896 (variable_value_hasher::hash, variable_value_hasher::equal): New
30898 (resolve_variable_value_in_expr, resolve_variable_value,
30899 resolve_variable_values, note_variable_value_in_expr,
30900 note_variable_value): New functions.
30901 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30904 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30907 * opts.h (handle_generated_option): Add GENERATED_P argument.
30908 * opts-common.c (handle_option): Adjust function comment.
30909 (handle_generated_option): Add GENERATED_P argument, pass it to
30911 (control_warning_option): Pass false to handle_generated_option
30913 * opts.c (maybe_default_option): Pass true to handle_generated_option
30915 * optc-gen.awk: Likewise.
30917 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30919 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30920 a REG, look at the REG it is a SUBREG of.
30921 (splitter for cmpeqsi_t): Ditto.
30923 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30925 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30926 the special USEs with the pattern of the insn, not the insn itself.
30928 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30931 * doc/invoke.texi: Document -mload-store-pairs.
30933 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30934 Sandra Loosemore <sandra@codesourcery.com>
30936 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30937 argument isn't a CONST_INT.
30938 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30939 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30940 (nios2_validate_compare): Bypass alternate compare logic if *op2
30941 is not a CONST_INT.
30942 (ldstwm_operation_p): Return false if first_base is not a REG or
30943 if first_offset is not a CONST_INT.
30945 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30947 * config/cris/cris.md: Use correct operand in a define_peephole2.
30949 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30951 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30953 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30955 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30956 this_insn if it is an INSN or JUMP_INSN.
30957 (force_offsettable): Look at base, not at addr.
30958 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30959 on things that aren't necessarily CONST_INTs.
30961 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30963 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30964 -mfpmath=sse is the default also for x86-32 targets with SSE2
30965 instruction set when @option{-ffast-math} is enabled
30967 2017-02-24 Jeff Law <law@redhat.com>
30969 PR rtl-optimizatoin/79286
30970 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30973 2017-02-24 Richard Biener <rguenther@suse.de>
30975 PR tree-optimization/79389
30976 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30979 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30981 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30982 function comment to reflect reality.
30983 (loop_exits_before_overflow): Fix typo in function description.
30985 2017-02-24 Richard Biener <rguenther@suse.de>
30987 PR tree-optimization/79389
30988 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30989 properly that a threading opportunity exists. Detect conditional
30990 copy/constant propagation opportunities.
30992 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30994 * config/visium/visium.md (type): Add trap.
30995 (b): New mode attribute.
30996 (*btst): Rename into...
30997 (*btst<mode>): ...this and adjust.
30998 (*cbranchsi4_btst_insn): Rename into...
30999 (*cbranch<mode>4_btst_insn): ...this and adjust.
31000 (trap): New define_insn.
31002 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31004 PR tree-optimization/79389
31005 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31006 (noce_reversed_cond_code): New function.
31007 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31008 reversed_comparison_code. Formatting fix.
31009 (noce_try_store_flag): Test rev_cond != NULL in addition to
31010 reversed_comparison_code.
31011 (noce_try_store_flag_constants): Likewise.
31012 (noce_try_store_flag_mask): Likewise.
31013 (noce_try_addcc): Use rev_cond if non-NULL instead of
31014 reversed_comparison_code.
31015 (noce_try_cmove_arith): Likewise. Formatting fixes.
31016 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31017 (noce_find_if_block): Initialize rev_cond.
31018 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31019 instead of false as last argument never attempt to reverse it
31022 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31024 PR tree-optimization/79663
31025 * tree-predcom.c (combine_chains): Process refs in reverse order
31026 only for ZERO length chains, and add explaining comment.
31028 2017-02-23 Jeff Law <law@redhat.com>
31030 PR tree-optimization/79578
31031 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31032 in call to operand_equal_p.
31034 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31037 * config/i386/cpuid.h: Fix another undefined behavior.
31039 2017-02-23 Richard Biener <rguenther@suse.de>
31041 PR tree-optimization/79683
31042 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31043 vector types for data-refs.
31045 2017-02-23 Martin Liska <mliska@suse.cz>
31047 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31049 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31051 PR middle-end/79665
31052 * internal-fn.c (get_range_pos_neg): Moved to ...
31053 * tree.c (get_range_pos_neg): ... here. No longer static.
31054 * tree.h (get_range_pos_neg): New prototype.
31055 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31056 are known to be in between 0 and signed maximum inclusive, try to
31057 expand both unsigned and signed divmod and use the cheaper one from
31060 2017-02-22 Jeff Law <law@redhat.com>
31062 PR tree-optimization/79578
31063 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31064 to compare base operands.
31066 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31069 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31070 gpc_reg_operand instead of fpr_reg_operand.
31072 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31074 * config/mips/mips.c (mips_return_in_memory): Force FP
31075 vector types to be returned in memory for o32 ABI.
31077 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31079 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31080 instead of DW_TAG_member for static data member declarations and don't
31081 set no_linkage_name for static inline data members.
31082 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31085 2017-02-22 Martin Liska <mliska@suse.cz>
31087 * doc/invoke.texi: Replace inequality signs with square brackets
31090 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31093 * lra-constraints.c (simplify_operand_subreg): Handle
31094 WORD_REGISTER_OPERATIONS targets.
31096 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31099 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31100 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31101 elimination by swapping fld*.
31103 2017-02-22 Richard Biener <rguenther@suse.de>
31105 PR tree-optimization/79673
31106 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31107 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31108 irrelevant address-space qualifiers and avoiding a
31109 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31111 2017-02-22 Richard Biener <rguenther@suse.de>
31113 PR tree-optimization/79666
31114 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31115 to not symbolically negate if that may introduce undefined
31118 2017-02-22 Martin Liska <mliska@suse.cz>
31121 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31122 * data-streamer-out.c (streamer_write_gcov_count_stream):
31124 * value-prof.c (stream_out_histogram_value): Make assert more
31125 precise based on type of counter.
31127 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31130 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31131 Use nonimmediate_operand instead of memory_operand for operand 1.
31132 (float-extend standard_x87sse_constant_load splitter): Ditto.
31134 2017-02-21 Jeff Law <law@redhat.com>
31136 PR tree-optimization/79621
31137 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31138 blocks with edges to themselves.
31140 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31143 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31144 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31145 Use gimple_call_builtin_p.
31148 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31149 on temporarily removed DEBUG_INSNs.
31151 PR tree-optimization/79649
31152 * tree-loop-distribution.c (classify_partition): Give up on
31153 non-generic address space loads/stores.
31155 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31157 * doc/loop.texi (Loop manipulation): Remove nonexistent
31158 tree_ssa_loop_version from the documentation.
31159 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31161 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31164 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31165 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31166 * config/rs6000/rs6000.c: Include except.h.
31167 (rs6000_expand_split_stack_prologue): Call
31168 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31170 2017-02-21 Martin Jambor <mjambor@suse.cz>
31173 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31174 have been analyzed.
31176 2017-02-21 Martin Jambor <mjambor@suse.cz>
31178 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31179 for backward compatibility only.
31180 * doc/invoke.texi (Option Summary): Remove all references to
31181 -fipa-cp-alignment.
31183 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31187 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31189 * lra-constraints.c (curr_insn_transform): Handle
31190 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31192 2017-02-21 Martin Liska <mliska@suse.cz>
31194 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31196 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31198 PR translation/79638
31199 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31201 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31204 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31205 (arm_function_ok_for_sibcall): Return false for an indirect call by
31206 descriptor if all the argument registers are used.
31207 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31208 alignment of the function.
31210 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31212 PR tree-optimization/61441
31213 * simplify-rtx.c (simplify_const_unary_operation): For
31214 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31215 the sNaN unmodified.
31217 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31219 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31220 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31221 instead of SYSTEM_HEADER_DIR.
31223 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31224 Martin Liška <mliska@suse.cz>
31226 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31227 Fix typos and grammar, use active voice, and clarify.
31229 2017-02-20 Marek Polacek <polacek@redhat.com>
31231 PR middle-end/79537
31232 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31235 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31237 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31240 * config/i386/i386.c (ix86_expand_builtin): Handle
31241 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31242 ix86_builtins_isa[fcode].isa as a requirement of those
31243 flags and any other flag in the bitmask.
31244 (ix86_init_mmx_sse_builtins): Use 0 instead of
31245 ~OPTION_MASK_ISA_64BIT as mask.
31246 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31247 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31248 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31249 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31251 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31254 * lra-constraints.c (split_reg): Check requested split mode
31255 is supported by the register.
31257 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31259 * lra-constraints.c (simplify_operand_subreg): Remove early
31262 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31265 * lra-constraints.c (curr_insn_transform): Tighten condition
31266 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31268 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31271 * lra-constraints.c (curr_insn_transform): Handle
31272 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31274 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31277 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31279 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31281 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31284 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31287 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31289 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31292 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31294 * final.c (last_columnnum, override_columnnum): New variables.
31295 (final_start_function): Set last_columnnum, pass it to begin_prologue
31296 hook and pass 0 to dwarf2out_begin_prologue.
31297 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31298 to source_line debug hook.
31299 (notice_source_line): Compute last_columnnum and for debug_column_info
31300 return true on column changes.
31301 * debug.h (struct gcc_debug_hooks): Add column argument to
31302 source_line and begin_prologue hooks.
31303 (debug_nothing_int_charstar_int_bool): Remove prototype.
31304 (debug_nothing_int_int_charstar,
31305 debug_nothing_int_int_charstar_int_bool): New prototypes.
31306 (dwarf2out_begin_prologue): Add column argument.
31307 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31308 begin_prologue hooks.
31309 (debug_nothing_int_charstar_int_bool): Remove.
31310 (debug_nothing_int_int_charstar,
31311 debug_nothing_int_int_charstar_int_bool): New functions.
31312 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31313 through to dwarf2out_source_line.
31314 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31315 (dwarf2out_source_line): Add column argument, emit it if requested.
31316 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31318 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31319 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31320 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31321 through to dwarf2out_begin_prologue.
31322 (vmsdbgout_source_line): Add column argument, pass it through to
31323 dwarf2out_source_line.
31324 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31325 dbxout_source_line caller.
31326 (dbxout_source_line): Add column argument.
31328 * common.opt (gno-column-info, gcolumn-info): New options.
31329 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31330 (check_die): Also test for multiple DW_AT_decl_column attributes.
31331 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31332 DW_AT_decl_column if requested.
31333 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31335 (gen_variable_die): Likewise.
31336 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31337 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31340 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31341 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31342 (ix86_handle_option): Handle OPT_m3dnowa.
31343 * doc/invoke.texi (-m3dnowa): Document.
31344 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31345 -m3dnowa instead of -m3dnow -march=athlon.
31348 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31349 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31351 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31354 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31355 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31356 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31357 generator for vsx_xxpermdi_<mode>_be.
31358 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31359 force big-endian semantics.
31360 (vsx_xxpermdi_<mode>_be): New define_expand with same
31361 implementation as previous version of vsx_xxpermdi_<mode>.
31363 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31365 PR tree-optimization/79327
31366 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31367 variable, its initialization and use.
31369 2017-02-17 Julia Koval <julia.koval@intel.com>
31371 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31372 (OPTION_MASK_ISA_PKU_UNSET): New.
31373 (ix86_handle_option): Handle -mrdpid.
31374 * config/i386/cpuid.h (bit_RDPID): New.
31375 * config/i386/driver-i386.c (host_detect_local_cpu):
31376 Detect RDPID feature.
31377 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31378 * config/i386/i386-c.c (ix86_target_macros_internal):
31380 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31381 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31382 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31383 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31384 * config/i386/i386.md (define_insn "rdpid"): New.
31385 * config/i386/i386.opt Add -mrdpid.
31386 * config/i386/immintrin.h (_rdpid_u32): New.
31388 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31390 PR rtl-optimization/79541
31391 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31392 instead of transforming it into USE.
31394 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31396 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31397 If HONOR_SNANS (SFmode) force the input to a register.
31398 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31399 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31400 an frsp or similar insn.
31402 2017-02-17 Martin Liska <mliska@suse.cz>
31404 PR rtl-optimization/79577
31405 * params.def (selsched-max-sched-times): Increase minimum to 1.
31407 2017-02-17 Martin Liska <mliska@suse.cz>
31409 PR rtl-optimization/79574
31410 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31412 2017-02-17 Martin Liska <mliska@suse.cz>
31414 PR tree-optimization/79529
31415 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31416 ssa_defined_default_def_p to handle cases which are implicitly
31418 * tree-ssa.c (ssa_defined_default_def_p): New function.
31419 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31420 which are implicitly defined.
31421 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31423 2017-02-17 Richard Biener <rguenther@suse.de>
31425 PR middle-end/79576
31426 * params.def (max-ssa-name-query-depth): Limit to 10.
31428 2017-02-17 Richard Biener <rguenther@suse.de>
31430 PR tree-optimization/79552
31431 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31434 2017-02-17 Richard Biener <rguenther@suse.de>
31437 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31439 2017-02-17 Marek Polacek <polacek@redhat.com>
31441 PR middle-end/79536
31442 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31443 (fold_negate_expr): New wrapper.
31445 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31447 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31448 Correct terminology and de-emphasize pre-standard behavior.
31450 2017-02-16 Alan Modra <amodra@gmail.com>
31452 PR rtl-optimization/79286
31453 * ira.c (def_dominates_uses): New function.
31454 (update_equiv_regs): Don't create an equivalence for insns that
31455 may trap where the register def does not dominate the use.
31457 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31459 PR rtl-optimization/78127
31460 * lra.c (lra): Call lra_eliminate before finish the loop after
31463 2017-02-16 Richard Biener <rguenther@suse.de>
31465 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31467 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31468 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31469 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31470 (isl_val_int_from_wi): New function.
31471 (extract_affine_gmp): Rename to ...
31472 (extract_affine_wi): ... this, take a widest_int.
31473 (extract_affine_int): Just wrap extract_affine_wi.
31474 (add_param_constraints): Use isl_val_int_from_wi.
31475 (add_loop_constraints): Likewise, and extract_affine_wi.
31477 2017-02-15 Jeff Law <law@redhat.com>
31479 PR middle-end/79521
31480 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31481 ira_init_register_move_cost_if_necessary.
31483 2017-02-15 Martin Sebor <msebor@redhat.com>
31485 PR middle-end/32003
31486 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31487 removed in a prior commit.
31489 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31491 PR tree-optimization/79347
31492 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31493 counters during peeling.
31495 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31497 * Makefile.in (site.exp): Remove "set ISLVER".
31499 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31502 * real.c (real_from_integer): Call real_convert even for decimal.
31504 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31507 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31509 2017-02-14 Andrew Pinski <apinski@cavium.com>
31511 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31512 cores and change the partno/implementer to be correct.
31513 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31514 the 'B" as the implementer.
31515 * config/aarch64/aarch64-tune.md: Regenerate.
31517 2017-02-14 Carl Love <cel@us.ibm.com>
31519 * config/rs6000/rs6000.c: Add case statement entry to make the
31520 xvcvuxdsp built-in argument unsigned.
31521 * config/rs6000/vsx.md: Fix the source and return operand types so they
31522 match the instruction definitions from the ISA document. Fix typo
31523 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31526 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31529 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31530 member early_clobber_alts.
31531 * lra-lives.c (reg_early_clobber_p): New.
31532 (process_bb_lives): Use it.
31533 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31534 (debug_operand_data): Initialize early_clobber_alts.
31535 (setup_operand_alternative): Set up early_clobber_alts.
31536 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31537 alternatives to new_insn_reg.
31538 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31540 (lra_update_insn_regno_info): Pass the new arg.
31542 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31544 PR middle-end/79505
31545 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31546 (new_oacc_loop_raw): Don't clear already cleared fields.
31549 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31550 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31551 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31553 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31556 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31558 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31561 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31562 the extra instruction to the right place to store 128-bit constant
31565 2017-02-14 Martin Sebor <msebor@redhat.com>
31567 PR middle-end/79448
31568 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31569 warning for strings of unknown length.
31571 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31573 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31575 2017-02-14 Jeff Law <law@redhat.com>
31578 * ira-costs.c (scan_one_insn): Initialize register move costs
31579 for pseudos seen in USE/CLOBBER insns.
31581 PR tree-optimization/79095
31582 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31583 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31584 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31585 if the operands are known to be not equal, then the resulting range
31587 (intersect_ranges): If the new range is ~[0,0] and the old range is
31588 wide, then prefer ~[0,0].
31589 * tree-vrp.c (overflow_comparison_p_1): New function.
31590 (overflow_comparison_p): New function.
31591 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31592 if NAME is used in an overflow test.
31593 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31594 overflow check that can be expressed as an equality test, then adjust
31595 ops to be that equality test.
31597 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31599 * config/s390/s390-builtin-types.def: Remove flags argument.
31600 * config/s390/s390.c (s390_init_builtins): Likewise.
31602 2017-02-14 Martin Liska <mliska@suse.cz>
31604 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31605 vector. Fix trailing white spaces.
31607 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31609 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31612 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31614 PR rtl-optimization/68664
31615 * config/arm/arm.c (arm_sched_can_speculate_insn):
31616 New function. Declare prototype.
31617 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31619 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31621 PR rtl-optimization/68664
31622 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31624 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31626 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31628 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31629 max skip bytes for function, loop and jump.
31631 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31633 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31634 ABS_EXPR for gimple dump.
31636 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31639 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31641 PR tree-optimization/79408
31642 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31643 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31644 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31645 also if rhs1 is INTEGER_CST.
31647 2017-02-14 Richard Biener <rguenther@suse.de>
31649 PR middle-end/79432
31650 * tree-into-ssa.c (insert_phi_nodes): When the function can
31651 have abnormal edges rewrite SSA names with broken use-def
31652 dominance out of SSA and register them for PHI insertion.
31654 2017-02-13 Martin Sebor <msebor@redhat.com>
31656 PR middle-end/79496
31657 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31658 clearing info.nowrite flag when snprintf size argument is a range.
31660 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31662 * cprop.c (cprop_jump): Add missing space in string literal.
31663 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31664 (get_constraint_for_component_ref): Likewise.
31665 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31666 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31667 * lra-constraints.c (process_alt_operands): Likewise.
31668 * ipa-inline.c (inline_small_functions): Likewise.
31669 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31670 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31671 * trans-mem.c (diagnose_tm_1_op): Likewise.
31672 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31673 (grid_parallel_clauses_gridifiable): Likewise.
31675 * config/nvptx/mkoffload.c (process): Add space in between
31678 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31679 "MOD4_SSE_REGS" and "ALL_REGS".
31681 * spellcheck.c (test_data): Add , in between "foo" and "food".
31683 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31686 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31687 boundary crossing check and subsequent code generation agree.
31689 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31691 * config/aarch64/aarch64.c (has_memory_op): Delete.
31692 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31695 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31697 PR rtl-optimization/79388
31698 PR rtl-optimization/79450
31699 * combine.c (distribute_notes): When removing TEM_INSN for which
31700 corresponding dest has last value recorded, invalidate that last
31703 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31705 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31706 of explicit '@'. Add missing assembly comment marker on branch costs
31709 2017-02-13 Nathan Sidwell <nathan@acm.org>
31711 * gengtype-lex.l (<in_struct>): Add '/'.
31713 2017-02-13 Martin Liska <mliska@suse.cz>
31716 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31718 2017-02-13 Richard Biener <rguenther@suse.de>
31720 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31722 * configure: Re-generate.
31723 * config.in: Likewise.
31724 * graphite-dependences.c: Simplify as if
31725 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31726 * graphite-isl-ast-to-gimple.c: Likewise.
31727 * graphite-optimize-isl.c: Likewise.
31728 * graphite-poly.c: Likewise.
31729 * graphite-sese-to-poly.c: Likewise.
31730 * graphite.h: Likewise.
31731 * toplev.c: Include isl/version.h and use isl_version () for
31732 printing the ISL version.
31733 * doc/install.texi: Update ISL requirement.
31735 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31737 * doc/standards.texi (Standards): Update reference to
31740 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31742 * doc/extend.texi (Named Address Spaces): sourceware.org now
31744 * doc/install.texi (Binaries): Ditto.
31747 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31749 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31750 standard terminology "stringize"/"stringizing" throughout.
31751 * doc/cppinternals.texi: Likewise.
31753 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31755 * doc/extend.texi: Fix some spelling mistakes and typos.
31756 * doc/invoke.texi: Likewise.
31758 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31761 * params.def (inline-min-speedup) Change from 10 to 8.
31763 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31765 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31768 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31771 * ipa-inline-analysis.c (get_minimal_bb): New function.
31772 (record_modified): Use it.
31773 (remap_edge_change_prob): Handle also ancestor functions.
31775 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31777 * doc/contrib.texi (Contributors): Remove broken link into
31778 the Mauve CVS repository.
31780 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31782 PR middle-end/79454
31783 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31784 result computation whenever lhs doesn't have vector mode, not
31785 just when it has BLKmode.
31787 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31789 * doc/makefile.texi (profiledbootstrap): Refer to the
31790 installation instructions only in textual form.
31792 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31795 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31797 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31799 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31800 (Specific): Update mingw-w64 reference.
31802 (Specific): Remove broken link to Renesas RX processor.
31804 2017-02-10 Richard Biener <rguenther@suse.de>
31806 * toplev.c (process_options): Do not mention obsolete graphite
31807 options when printing sorry message about missing graphite support.
31808 Mention -floop-nest-optimize.
31810 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31812 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31813 (vtst_p16): Likewise.
31814 (vtstq_p8): Likewise.
31815 (vtstq_p16): Likewise.
31817 (vtstq_p64): Likewise.
31818 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31819 (vset_lane_p64): New.
31820 (vsetq_lane_p64): New.
31822 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31824 PR tree-optimization/79411
31825 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31826 stmt operands are SSA_NAMEs used in abnormal phis.
31827 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31830 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31833 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31836 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31838 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31840 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31842 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31846 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31847 not arbitrary TREE_CONSTANT.
31850 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31851 "omp declare target link" attribute unless is_global_var.
31852 * omp-offload.c (find_link_var_op): Likewise.
31854 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31855 Chung-Lin Tang <cltang@codesourcery.com>
31857 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31859 (gimplify_adjust_omp_clauses): Don't delete TILE.
31860 (gimplify_omp_for): Deal with TILE.
31861 * internal-fn.c (expand_GOACC_TILE): New function.
31862 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31864 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31865 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31867 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31868 avoid DIV for outermost collapse var.
31869 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31870 Remove out of date comments, fix whitespace.
31871 * omp-general.c (omp_extract_for_data): Deal with tiling.
31872 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31873 adjust OLF_DIM_BASE value.
31874 (struct omp_for_data): Add tiling field.
31875 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31876 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31877 for auto loops. Remove default auto determining, moved to
31878 oacc_loop_fixed_partitions.
31879 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31880 stmts, add e_mask field.
31881 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31882 (oacc_thread_numbers): Use oacc_dim_call.
31883 (oacc_xform_tile): New.
31884 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31885 (finish_oacc_loop): Adjust for ifns vector.
31886 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31887 add case for GOACC_TILE fns.
31888 (oacc_loop_xform_loop): Delete.
31889 (oacc_loop_process): Iterate over call list directly, and add
31890 handling for GOACC_TILE fns.
31891 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31893 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31894 vector partitioning to outer loops. Assign 2 partitions to loops
31895 when available. Add TILE handling.
31896 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31897 (execite_oacc_device_lower): Process GOACC_TILE fns,
31898 ignore unknown specs.
31899 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31900 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31901 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31903 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31905 * configure.ac (ACX_BUGURL): Update.
31906 * configure: Regenerate.
31908 2017-02-09 Richard Biener <rguenther@suse.de>
31910 PR tree-optimization/69823
31911 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31912 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31914 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31916 * config/arc/arc-c.def: Add __NPS400__ definition.
31917 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31918 (TARGET_NPS400): Define.
31920 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31922 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31924 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31925 pointer, arch_info.
31926 (arc_cpu_types): Fill the arch_info field with a pointer into the
31927 arc_arch_types table.
31928 (arc_selected_cpu): Declare.
31929 * config/arc/arc.c (arc_selected_cpu): Make global.
31930 (arc_selected_arch): Delete.
31931 (arc_base_cpu): Delete.
31932 (arc_override_options): Remove references to deleted variables,
31933 update access to arch information.
31934 (ARC_OPT): Update access to arch information.
31935 (ARC_OPTX): Likewise.
31936 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31937 (TARGET_ARC600): Update access to arch information.
31938 (TARGET_ARC601): Likewise.
31939 (TARGET_ARC700): Likewise.
31940 (TARGET_EM): Likewise.
31941 (TARGET_HS): Likewise.
31942 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31945 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31948 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31949 condition/operands for integer GE/LE/GEU/LEU operations.
31951 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31953 PR translation/79397
31954 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31957 2017-02-08 Martin Jambor <mjambor@suse.cz>
31960 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31961 whether allocation happened.
31962 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31963 nothing was allocated.
31965 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31967 PR tree-optimization/79408
31968 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31969 constant, but SSA_NAME with a known integer range, use the minimum
31970 of that range instead of op1 to determine if modulo can be replaced
31971 with its first operand.
31973 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31975 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31977 2017-02-08 Richard Biener <rguenther@suse.de>
31979 PR tree-optimization/71824
31980 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31981 Check all loops contained in the merged region.
31983 2017-02-07 Andrew Pinski <apinski@cavium.com>
31985 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31987 2017-02-07 Andrew Pinski <apinski@cavium.com>
31989 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31990 (thunderxt88): Likewise.
31991 (thunderxt81): Disable LSE and change v8.1 to v8.
31992 (thunderxt83): Likewise.
31994 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31995 Richard Biener <rguenther@suse.de>
31997 PR middle-end/79399
31998 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31999 type from int to size_t.
32000 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32002 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32004 PR rtl-optimization/79386
32005 * cprop.c (bypass_conditional_jumps): Initialize
32006 bypass_last_basic_block already before splitting bbs after
32007 unconditional traps...
32008 (bypass_conditional_jumps): ... rather than here.
32011 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32012 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32013 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32014 fix -masm=intel patterns.
32016 2017-02-07 Richard Biener <rguenther@suse.de>
32018 PR tree-optimization/79256
32019 PR middle-end/79278
32020 * builtins.c (get_object_alignment_2): Use min_align_of_type
32021 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32022 and ADJUST_FIELD_ALIGN.
32024 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32026 * doc/tm.texi: Regenerate.
32027 * stor-layout.c (layout_decl): Adjust.
32028 (update_alignment_for_field): Likewise.
32029 (place_field): Likewise.
32030 (min_align_of_type): Likewise.
32031 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32032 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32033 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32034 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32035 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32036 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32037 * config/i386/i386.c (x86_field_alignment): Likewise.
32038 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32039 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32040 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32041 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32042 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32043 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32047 2017-01-30 Richard Biener <rguenther@suse.de>
32049 PR tree-optimization/79256
32050 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32051 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32054 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32056 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32057 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32058 builtins to SImode and emit a zero-extend, if necessary.
32060 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32062 * docs/invoke.texi (RISC-V Options): Alphabetize.
32064 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32066 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32069 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32071 * config/riscv/riscv.c: New file.
32072 * common/config/riscv/riscv-common.c: Likewise.
32073 * config.gcc: Likewise.
32074 * config/riscv/constraints.md: Likewise.
32075 * config/riscv/elf.h: Likewise.
32076 * config/riscv/generic.md: Likewise.
32077 * config/riscv/linux.h: Likewise.
32078 * config/riscv/multilib-generator: Likewise.
32079 * config/riscv/peephole.md: Likewise.
32080 * config/riscv/pic.md: Likewise.
32081 * config/riscv/predicates.md: Likewise.
32082 * config/riscv/riscv-builtins.c: Likewise.
32083 * config/riscv/riscv-c.c: Likewise.
32084 * config/riscv/riscv-ftypes.def: Likewise.
32085 * config/riscv/riscv-modes.def: Likewise.
32086 * config/riscv/riscv-opts.h: Likewise.
32087 * config/riscv/riscv-protos.h: Likewise.
32088 * config/riscv/riscv.h: Likewise.
32089 * config/riscv/riscv.md: Likewise.
32090 * config/riscv/riscv.opt: Likewise.
32091 * config/riscv/sync.md: Likewise.
32092 * config/riscv/t-elf-multilib: Likewise.
32093 * config/riscv/t-linux: Likewise.
32094 * config/riscv/t-linux-multilib: Likewise.
32095 * config/riscv/t-riscv: Likewise.
32096 * configure.ac: Likewise.
32097 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32098 Waterman as RISC-V maintainers.
32099 * doc/install.texi: Add RISC-V entries.
32100 * doc/invoke.texi: Add RISC-V options section.
32101 * doc/md.texi: Add RISC-V constraints section.
32102 * configure: Regenerated.
32104 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32107 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32108 false values to be constant vectors with all 0 or all 1 bits set.
32109 (vcondu<mode><mode>): Likewise.
32110 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32112 (fpmask_comparison_operator): Update comment.
32113 (vecint_comparison_operator): New predicate.
32114 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32115 vector conditionals when the true and false values are constant
32116 vectors with all 0 bits or all 1 bits set.
32118 2017-02-06 Martin Sebor <msebor@redhat.com>
32120 PR tree-optimization/79376
32121 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32123 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32125 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32126 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32127 to simplify split condition.
32129 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32131 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32132 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32135 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32137 PR rtl-optimization/68664
32138 * target.def (can_speculate_insn): New hook.
32139 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32140 * doc/tm.texi: Regenerate.
32141 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32142 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32143 (rs6000_sched_can_speculate_insn): New function.
32145 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32147 PR tree-optimization/79284
32148 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32149 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32150 vectorizable_mask_load_store, vectorizable_operation,
32151 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32152 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32153 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32154 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32155 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32156 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32157 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32158 is_gimple_assign (stmt). Replace another such test with
32159 is_gimple_assign (stmt).
32161 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32164 * config/avr/avr.c (rtl-iter.h): Include it.
32165 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32166 (avr_legitimate_combined_insn): ...and implementation.
32168 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32170 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32171 * config/s390/s390.c (s390_const_operand_ok)
32172 (s390_canonicalize_comparison, s390_extract_part)
32173 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32174 (s390_contiguous_bitmask_p, s390_rtx_costs)
32175 (legitimize_pic_address): Likewise.
32176 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32177 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32178 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32179 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32180 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32182 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32184 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32185 REGNO($0) == REGNO($1).
32187 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32189 * config/s390/linux.h(SIZE_TYPE): Add comment.
32191 2017-02-06 Julian Brown <julian@codesourcery.com>
32192 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32193 Virendra Pathak <virendra.pathak@broadcom.com>
32195 * config/aarch64/aarch64-cores.def: Change the scheduler
32197 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32198 * config/aarch64/thunderx2t99.md: New file.
32200 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32202 * doc/standards.texi (Go Language): Update link to language
32205 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32207 * tree-eh.c (lower_resx): Sanitize profile.
32208 (cleanup_empty_eh_move_lp): Likewise.
32210 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32213 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32215 * cfgloopmanip.h (loop_version): Update prototype.
32216 * modulo-sched.c (sms_schedule): Update call of loop_version.
32217 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32218 * tree-parloops.c (gen_parallel_loop): Likewise.
32219 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32220 * tree-ssa-loop-split.c (split_loop): Likewise.
32221 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32222 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32224 2017-02-05 Martin Liska <mliska@suse.cz>
32227 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32229 (print_operand_address): Initialize a struct to zero.
32231 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32233 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32234 garbage collector only in textual form.
32236 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32238 * doc/extend.texi (x86 specific memory model extensions for
32239 transactional memory): Simplify a phrase.
32241 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32244 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32245 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32246 (atomic_storedi_1): Likewise.
32248 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32250 PR tree-optimization/79338
32251 * tree-parloops.c (gather_scalar_reductions): Don't call
32252 vect_analyze_loop_form for loop->inner before destroying loop's
32255 2017-02-03 Martin Sebor <msebor@redhat.com>
32257 PR tree-optimization/79327
32258 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32259 when precision has resulted in leading zeros.
32260 (format_integer): Adjust the likely counter to assume an unknown
32261 argument that may be zero is non-zero.
32263 2017-02-03 Jason Merrill <jason@redhat.com>
32266 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32267 avoid copying non-taken branch.
32269 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32271 PR tree-optimization/79340
32272 * tree-vect-loop.c (vectorizable_reduction): Release
32273 vec_defs elements after safe_splicing them into other vectors.
32276 PR tree-optimization/79327
32277 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32278 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32280 (format_integer): Use wide_int_to_tree instead of build_int_cst
32281 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32282 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32283 of shortest and longest sequence.
32285 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32287 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32288 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32290 2017-02-03 Walter Lee <walt@tilera.com>
32293 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32294 after initial stackframe link reg save.
32295 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32297 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32300 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32301 wu for stxssp alternative.
32303 2017-02-03 Martin Sebor <msebor@redhat.com>
32305 PR tree-optimization/79352
32306 * gimple-fold.c (get_range_strlen): Add argument.
32307 (get_range_strlen): Change return type to bool.
32308 (get_maxval_strlen): Pass in a dummy argument.
32309 * gimple-fold.h (get_range_strlen): Change return type to bool.
32310 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32311 * tree.h (array_at_struct_end_p): Add argument.
32312 * tree.c (array_at_struct_end_p): Handle it.
32314 2017-02-03 Martin Liska <mliska@suse.cz>
32317 * multiple_target.c (create_dispatcher_calls): Redirect edge
32318 from a caller of a dispatcher.
32319 (expand_target_clones): Make the clones local.
32320 (ipa_target_clone): Do both target clones and resolvers.
32321 (ipa_dispatcher_calls): Remove the pass.
32322 (pass_dispatcher_calls::gate): Likewise.
32323 (make_pass_dispatcher_calls): Likewise.
32324 * passes.def (pass_target_clone): Put as very first IPA early
32327 2017-02-03 Martin Liska <mliska@suse.cz>
32329 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32330 in case of a function with ifunc attribute.
32332 2017-02-03 Martin Liska <mliska@suse.cz>
32334 * cgraph.c (cgraph_node::dump): Dump function version info.
32335 * symtab.c (symtab_node::dump_base): Add missing new line.
32337 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32339 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32340 (ifcombine_ifandif): Use it.
32342 2017-02-03 Martin Liska <mliska@suse.cz>
32344 * doc/invoke.texi: Document default value for
32345 use-after-scope-direct-emission-threshold.
32347 2017-02-03 Martin Liska <mliska@suse.cz>
32349 PR tree-optimization/79339
32350 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32351 (format_floating): Likewise.
32353 2017-02-03 Martin Liska <mliska@suse.cz>
32356 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32358 (ipa_node_params_t::remove): Likewise.
32359 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32360 initialization from removed ipa_node_params_t::insert.
32361 (ipa_node_params::~ipa_node_params): Move from removed
32362 ipa_node_params_t::release.
32363 * symbol-summary.h (symbol_summary::m_released): New member.
32364 Do not release a summary twice. Do not allow to call finalizer
32365 for types of a summary that live in GGC memory.
32367 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32369 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32372 2017-02-02 Martin Sebor <msebor@redhat.com>
32374 PR middle-end/79275
32375 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32376 (format_string): Tighten up the range of output for non-constant
32377 strings and correct the expected range for wide non-constant strings.
32379 2017-02-02 Martin Sebor <msebor@redhat.com>
32381 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32383 PR middle-end/32003
32384 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32386 (-fdump-tree-@var): Add to index and document how to come up
32387 with pass-specific option and dump file names.
32388 (-fdump-passes): Clarify where to look for output.
32390 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32392 PR middle-end/77445
32393 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32394 statistics of the analyzed path; allow threading for speed when
32395 any of BBs along the path are optimized for speed.
32397 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32399 PR middle-end/78468
32400 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32401 settings of the virtual registers.
32404 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32406 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32407 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32410 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32412 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32413 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32415 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32417 * config/s390/s390.md: Add missing comments with the expanded
32419 * config/s390/vector.md: Likewise.
32420 * config/s390/vx-builtins.md: Likewise.
32422 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32425 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32426 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32427 conditions on a single line.
32429 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32431 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32432 __S390_VX__ to __VX__.
32434 2017-02-01 Andrew Pinski <apinski@cavium.com>
32436 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32437 stmt_info to record_stmt_cost.
32438 (vect_get_known_peeling_cost): Pass stmt_info if known to
32440 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32441 cpu_vector_cost field into
32442 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32443 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32444 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32445 splitting of scalar_stmt_cost and vec_stmt_cost.
32446 (thunderx_vector_cost): Likewise.
32447 (cortexa57_vector_cost): LIkewise.
32448 (exynosm1_vector_cost): Likewise.
32449 (xgene1_vector_cost): Likewise.
32450 (thunderx2t99_vector_cost): Improve after the splitting of the two
32452 (aarch64_builtin_vectorization_cost): Update for the splitting of
32453 scalar_stmt_cost and vec_stmt_cost.
32455 2017-02-01 Torvald Riegel <triegel@redhat.com>
32456 Richard Henderson <rth@redhat.com>
32458 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32459 conditional on existance of a fast atomic load.
32460 * optabs-query.c (can_atomic_load_p): New function.
32461 * optabs-query.h (can_atomic_load_p): Declare it.
32462 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32463 no fast atomic load is available for the particular size of access.
32464 (expand_atomic_compare_and_swap): Likewise.
32465 (expand_atomic_load): Likewise.
32466 (expand_atomic_store): Likewise.
32467 (expand_atomic_fetch_op): Likewise.
32468 * testsuite/lib/target-supports.exp
32469 (check_effective_target_sync_int_128): Remove x86 because it provides
32470 no fast atomic load.
32471 (check_effective_target_sync_int_128_runtime): Likewise.
32473 2017-02-01 Richard Biener <rguenther@suse.de>
32475 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32476 (graphite_transform_loops): Provide opt-info for optimized nests.
32477 * tree-parloop.c (parallelize_loops): Provide opt-info for
32478 parallelized loops.
32480 2017-02-01 Richard Biener <rguenther@suse.de>
32482 PR middle-end/79315
32483 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32484 was not set before.
32486 2017-02-01 Richard Biener <rguenther@suse.de>
32488 PR tree-optimization/71824
32489 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32490 Verify the loops are valid in the merged SESE region.
32491 (scop_detection::can_represent_loop_1): Check analyzing the
32492 evolution of the number of iterations in the region succeeds.
32494 2017-01-31 Ian Lance Taylor <iant@golang.org>
32496 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32497 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32499 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32501 PR preprocessor/79210
32502 * input.c (get_substring_ranges_for_loc): Replace line_width
32503 assertion with error-handling.
32505 2017-01-31 Richard Biener <rguenther@suse.de>
32507 PR tree-optimization/77318
32508 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32509 (create_pw_aff_from_tree): Take loop parameter.
32510 (add_condition_to_pbb): Pass loop of the condition to
32511 create_pw_aff_from_tree.
32513 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32515 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32516 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32518 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32522 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32524 (convert_int_to_float128): Likewise.
32525 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32526 (convert_int_to_float128): Likewise.
32527 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32528 (UNSPEC_IEEE128_CONVERT): Likewise.
32529 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32530 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32531 Use local variables for IBM extended format.
32532 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32533 (fix_trunc<mode>si2_fprs): Likewise.
32534 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32535 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32536 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32537 to know that we can now have integers of all sizes in vector
32539 (fix<uns>_<mode>di2_hw): Likewise.
32540 (float<uns>_<mode>si2_hw): Likewise.
32541 (fix_<mode>si2_hw): Likewise.
32542 (fixuns_<mode>si2_hw): Likewise.
32543 (float<uns>_<mode>di2_hw): Likewise.
32544 (float_<mode>di2_hw): Likewise.
32545 (float_<mode>si2_hw): Likewise.
32546 (floatuns_<mode>di2_hw): Likewise.
32547 (floatuns_<mode>si2_hw): Likewise.
32548 (xscvqp<su>wz_<mode>): Delete, no longer used.
32549 (xscvqp<su>dz_<mode>): Likewise.
32550 (xscv<su>dqp_<mode>): Likewise.
32551 (ieee128_mfvsrd_64bit): Likewise.
32552 (ieee128_mfvsrd_32bit): Likewise.
32553 (ieee128_mfvsrwz): Likewise.
32554 (ieee128_mtvsrw): Likewise.
32555 (ieee128_mtvsrd_64bit): Likewise.
32556 (ieee128_mtvsrd_32bit): Likewise.
32558 2017-01-31 Martin Liska <mliska@suse.cz>
32561 * ipa-prop.c (ipa_free_all_node_params): Call release method
32562 instead of ~sumbol_summary to not to trigger double times
32565 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32567 PR tree-optimization/71691
32568 * bitmap.h (class auto_bitmap): New.
32569 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32570 is_maybe_undefined instead of ssa_undefined_value_p.
32572 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32574 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32575 __S390_ARCH_LEVEL__ to __ARCH__.
32577 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32579 PR tree-optimization/79267
32580 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32581 if should_remove_lhs_p is true.
32583 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32586 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32587 (add_alignment_attribute): New.
32588 (base_type_die): Add alignment attribute.
32589 (subrange_type_die): Likewise.
32590 (modified_type_die): Likewise.
32591 (gen_array_type_die): Likewise.
32592 (gen_descr_array_type_die: Likewise.
32593 (gen_enumeration_type_die): Likewise.
32594 (gen_subprogram_die): Likewise.
32595 (gen_variable_die): Likewise.
32596 (gen_field_die): Likewise.
32597 (gen_ptr_to_mbr_type_die): Likewise.
32598 (gen_struct_or_union_type_die): Likewise.
32599 (gen_subroutine_type_die): Likewise.
32600 (gen_typedef_die): Likewise.
32601 (base_type_cmp): Compare alignment attribute.
32603 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32606 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32607 (setb_unsigned) New pattern for setb with CCUNS.
32608 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32609 subfc./subfe sequence to avoid overflow problems. Generate a
32610 shorter sequence with cmpld/setb for power9.
32611 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32612 for generating subfc. instruction.
32613 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32614 now uses this instruction.
32616 2017-01-30 Ian Lance Taylor <iant@google.com>
32619 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32620 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32622 2017-01-30 Martin Sebor <msebor@redhat.com>
32624 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32625 Move constant to the right of a relational operator.
32626 (get_mpfr_format_length, format_character, format_string): Ditto.
32627 (should_warn_p, maybe_warn): Same.
32629 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32631 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32634 * asan.c (get_translation_unit_decl): Remove function.
32635 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32637 2017-01-30 Martin Liska <mliska@suse.cz>
32639 PR gcov-profile/79259
32640 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32641 -fprofile-generate.
32643 2017-01-30 Martin Liska <mliska@suse.cz>
32646 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32647 Initialize variables with NULL value.
32649 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32652 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32654 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32656 2017-01-30 Richard Biener <rguenther@suse.de>
32658 PR tree-optimization/79276
32659 * tree-vrp.c (process_assert_insertions): Properly adjust common
32660 when removing a duplicate.
32662 2017-01-30 Richard Biener <rguenther@suse.de>
32664 PR tree-optimization/79256
32665 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32666 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32668 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32670 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32673 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32674 ("*r<noxa>sbg_<mode>_sll_bitmask")
32675 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32676 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32677 Use contiguous_bitmask_nowrap_operand.
32679 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32682 * config/rs6000/altivec.h (vec_xl): Revise #define.
32683 (vec_xst): Likewise.
32685 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32687 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32689 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32691 PR rtl-optimization/79194
32692 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32693 traps before call to bypass_conditional_jumps.
32695 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32697 PR tree-optimization/71374
32698 * lra-constraints.c (check_conflict_input_operands): New.
32699 (match_reload): Use it.
32701 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32704 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32705 account to calculate conflict_set.
32707 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32709 PR rtl-optimization/78559
32710 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32711 other_insn in combine.
32713 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32715 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32716 uint16_type_node for BT_UINT16.
32718 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32720 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32721 "RTL Tests" to menu.
32722 (GIMPLE Tests): New node.
32723 (RTL Tests): New node.
32725 2017-01-27 Richard Biener <rguenther@suse.de>
32727 PR tree-optimization/79245
32728 * tree-loop-distribution.c (distribute_loop): Apply cost
32729 modeling also to detected patterns.
32731 2017-01-27 Richard Biener <rguenther@suse.de>
32733 PR tree-optimization/71433
32734 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32735 (compare_assert_loc): New function.
32736 (process_assert_insertions): Sort and optimize assert locations
32737 to remove duplicates and push down identical assertions on
32738 edges to their destination block.
32740 2017-01-27 Richard Biener <rguenther@suse.de>
32742 PR tree-optimization/79244
32743 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32744 out SSA names even if abnormal.
32746 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32748 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32749 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32750 instead of MPFR_RNDN.
32752 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32755 * arm.c (arm_option_override): Don't call build_target_option_node
32756 until after doing all option overrides.
32757 (arm_valid_target_attribute_tree): Likewise.
32759 2017-01-27 Martin Liska <mliska@suse.cz>
32761 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32762 for {cd}tors and C++ {cd}tors.
32764 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32766 * config/s390/s390.md ("*setmem_long_and")
32767 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32769 2017-01-26 Martin Sebor <msebor@redhat.com>
32771 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32774 2017-01-26 Martin Sebor <msebor@redhat.com>
32776 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32777 HAVE_DFmode before using XFmode or DFmode.
32778 (parse_directive): Avoid using the z length modifier to avoid
32779 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32781 PR middle-end/78703
32782 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32783 to accept adjustment as an array.
32784 (get_int_range): New function.
32785 (struct directive): Make width and prec arrays.
32786 (directive::set_width, directive::set_precision): Call get_int_range.
32787 (format_integer, format_floating): Handle width and precision ranges.
32788 (format_string, parse_directive): Same.
32790 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32793 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32794 comdat_type_p set, just clone them, but keep the children in the
32798 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32799 which have direct callers with -fvar-tracking-assignments enabled
32801 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32802 inside of type units.
32804 2017-01-26 Martin Sebor <msebor@redhat.com>
32806 PR middle-end/78703
32807 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32809 (struct format_result): Replace number_chars, number_chars_min,
32810 and number_chars_max with a single member of struct result_range.
32812 (format_result::operator+=): Adjust.
32813 (struct fmtresult): Remove bounded. Handle likely and unlikely
32815 (fmtresult::adjust_for_width_or_precision): New function.
32816 (fmtresult:type_max_digits): New function.
32817 (bytes_remaining): Handle likely and unlikely counters.
32818 (min_bytes_remaining): Remove.
32819 (format_percent): Simplify.
32820 (format_integer, format_floating): Set likely and unlikely counters.
32821 (get_string_length, format_character, format_string): Same.
32822 (format_plain, should_warn_p): New function.
32823 (maybe_warn): Call should_warn_p. Update diagnostic messages
32824 and handle those for all directives, including plain strings.
32825 (format_directive): Handle likely and unlikely counters.
32826 Remove unnecessary quoting from diagnostics. Add an informational
32828 (add_bytes): Remove.
32829 (pass_sprintf_length::compute_format_length): Simplify.
32830 (try_substitute_return_value): Handle likely and unlikely counters.
32832 2017-01-26 Carl Love <cel@us.ibm.com>
32834 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32835 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32837 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32840 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32841 endianess for subregs into account.
32842 * lra-constraints.c (lra_constraints): Do risky transformations
32843 always on the first iteration.
32844 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32845 last_call_used_reg_set.
32846 (process_bb_lives): Define and use last_call_used_reg_set.
32847 * lra.c (lra): Always continue after lra_constraints on the first
32850 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32852 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32854 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32856 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32858 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32859 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32860 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32861 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32862 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32863 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32864 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32865 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32866 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32868 2017-01-26 Marek Polacek <polacek@redhat.com>
32871 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32872 for the third operand.
32874 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32876 PR middle-end/79236
32877 * omp-low.c (struct omp_context): Add simt_stmt field.
32878 (scan_omp_for): Return omp_context *.
32879 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32880 context to the _simt_ SIMD stmt.
32881 (lower_omp_for): For combined SIMD with sibling _simt_
32882 SIMD, make sure to use the same decls in _looptemp_
32883 clauses as in the sibling.
32885 2017-01-26 David Sherwood <david.sherwood@arm.com>
32887 PR middle-end/79212
32888 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32891 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32894 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32895 emit fld b; fld a; if possible.
32897 * brig-builtins.def: Update copyright years.
32898 * config/arm/arm_acle_builtins.def: Update copyright years.
32900 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32903 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32904 constraint instead of o for the stxsd instruction.
32906 2017-01-25 Carl Love <cel@us.ibm.com>
32908 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32909 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32911 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32913 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32915 2017-01-25 Richard Biener <rguenther@suse.de>
32917 PR tree-optimization/69264
32918 * target.def (vector_alignment_reachable): Improve documentation.
32919 * doc/tm.texi: Regenerate.
32920 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32922 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32923 earlier changes with respect to TYPE_USER_ALIGN.
32924 (vector_alignment_reachable_p): Likewise. Improve dumping.
32926 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32929 * config/arm/arm.md (xordi3): Force constant operand into a register
32932 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32934 * doc/invoke.texi (-fstore-merging): Correct default optimization
32935 levels at which it is enabled.
32936 (-O): Move -fstore-merging from list to...
32939 2017-01-25 Richard Biener <rguenther@suse.de>
32942 * omp-expand.c: Include debug.h.
32943 (expand_omp_taskreg): Make sure to generate early debug before
32944 outlining anything from a function.
32945 (expand_omp_target): Likewise.
32946 (grid_expand_target_grid_body): Likewise.
32948 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32951 * asan.c (get_translation_unit_decl): New function.
32952 (asan_add_global): Extract modules file name from globals
32953 TRANSLATION_UNIT_DECL name.
32955 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32958 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32959 for long calls with APCS frame and VFP.
32961 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32963 * cfg.c (original_copy_tables_initialized_p): New function.
32964 * cfg.h (original_copy_tables_initialized_p): New decl.
32965 * cfgrtl.c (relink_block_chain): Guard the call to
32966 free_original_copy_tables with a call to
32967 original_copy_tables_initialized_p.
32968 * cgraph.h (symtab_node::native_rtl_p): New decl.
32969 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32970 (symtab_node::needed_p): Don't assert for early assembly output
32971 for __RTL functions.
32972 (cgraph_node::finalize_function): Set "force_output" for __RTL
32974 (cgraph_node::analyze): Bail out early for __RTL functions.
32975 (analyze_functions): Update assertion to support __RTL functions.
32976 (cgraph_node::expand): Bail out early for __RTL functions.
32977 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32979 * function.h (struct function): Update comment for field
32981 * gimple-expr.c: Include "tree-pass.h".
32982 (gimple_has_body_p): Return false for __RTL functions.
32983 * Makefile.in (OBJS): Add run-rtl-passes.o.
32984 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32986 (gcc::pass_manager::get_clean_slate): New accessor.
32987 * passes.c: Include "insn-addr.h".
32988 (should_skip_pass_p): Add logging. Update logic for running
32989 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32990 property-provider override so it is only done for gimple passes.
32992 (skip_pass): New function.
32993 (execute_one_pass): Call skip_pass when skipping passes.
32994 * read-md.c (md_reader::read_char): Support filtering
32995 the input to a subset of line numbers.
32996 (md_reader::md_reader): Initialize fields
32997 m_first_line and m_last_line.
32998 (md_reader::read_file_fragment): New function.
32999 * read-md.h (md_reader::read_file_fragment): New decl.
33000 (md_reader::m_first_line): New field.
33001 (md_reader::m_last_line): New field.
33002 * read-rtl-function.c (function_reader::create_function): Only
33003 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33004 curr_properties. Set DECL_INITIAL to a dummy block.
33005 (read_rtl_function_body_from_file_range): New function.
33006 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33008 * run-rtl-passes.c: New file.
33009 * run-rtl-passes.h: New file.
33011 2017-01-24 Jeff Law <law@redhat.com>
33013 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33016 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33018 PR tree-optimization/79159
33019 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33020 (record_nonwrapping_iv): Improve boundary using above function if no
33021 value range information.
33023 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33024 Martin Jambor <mjambor@suse.cz>
33026 * brig-builtins.def: New file.
33027 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33028 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33029 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33030 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33031 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33032 * builtin-types.def (BT_INT8): New.
33033 (BT_INT16): Likewise.
33034 (BT_UINT8): Likewise.
33035 (BT_UINT16): Likewise.
33036 (BT_FN_ULONG): Likewise.
33037 (BT_FN_UINT_INT): Likewise.
33038 (BT_FN_UINT_ULONG): Likewise.
33039 (BT_FN_UINT_LONG): Likewise.
33040 (BT_FN_UINT_PTR): Likewise.
33041 (BT_FN_ULONG_PTR): Likewise.
33042 (BT_FN_INT8_FLOAT): Likewise.
33043 (BT_FN_INT16_FLOAT): Likewise.
33044 (BT_FN_UINT32_FLOAT): Likewise.
33045 (BT_FN_UINT16_FLOAT): Likewise.
33046 (BT_FN_UINT8_FLOAT): Likewise.
33047 (BT_FN_UINT64_FLOAT): Likewise.
33048 (BT_FN_UINT16_UINT32): Likewise.
33049 (BT_FN_UINT32_UINT16): Likewise.
33050 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33051 (BT_FN_INT_PTR_INT): Likewise.
33052 (BT_FN_UINT_PTR_UINT): Likewise.
33053 (BT_FN_LONG_PTR_LONG): Likewise.
33054 (BT_FN_ULONG_PTR_ULONG): Likewise.
33055 (BT_FN_VOID_UINT64_UINT64): Likewise.
33056 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33057 (BT_FN_INT8_INT8_INT8): Likewise.
33058 (BT_FN_INT16_INT16_INT16): Likewise.
33059 (BT_FN_INT_INT_INT): Likewise.
33060 (BT_FN_UINT_FLOAT_UINT): Likewise.
33061 (BT_FN_FLOAT_UINT_UINT): Likewise.
33062 (BT_FN_ULONG_UINT_UINT): Likewise.
33063 (BT_FN_ULONG_UINT_PTR): Likewise.
33064 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33065 (BT_FN_UINT_UINT_UINT): Likewise.
33066 (BT_FN_VOID_UINT_PTR): Likewise.
33067 (BT_FN_UINT_UINT_PTR: Likewise.
33068 (BT_FN_UINT32_UINT64_PTR): Likewise.
33069 (BT_FN_INT_INT_UINT_UINT): Likewise.
33070 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33071 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33072 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33073 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33074 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33075 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33076 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33077 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33078 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33079 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33080 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33081 * doc/frontends.texi: List BRIG FE.
33082 * doc/install.texi (Testing): Add BRIG tesring requirements.
33083 * doc/invoke.texi (Overall Options): Mention BRIG.
33084 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33086 2017-01-24 Richard Biener <rguenther@suse.de>
33088 PR translation/79208
33089 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33091 2017-01-24 Martin Jambor <mjambor@suse.cz>
33094 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33095 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33096 and known_contexts.
33098 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33100 PR middle-end/79123
33101 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33102 casts from signed to unsigned really don't have a range.
33104 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33106 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33107 GMP_RNDx for compatiblity.
33109 2017-01-24 Martin Liska <mliska@suse.cz>
33112 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33113 that would prevent us to call alloca with -1 as argument.
33115 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33117 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33118 Avoid -Wformat-security warning.
33120 2017-01-23 Andrew Pinski <apinski@cavium.com>
33122 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33125 2017-01-23 Martin Sebor <msebor@redhat.com>
33127 PR middle-end/78703
33128 * gimple-ssa-sprintf.c (warn_level): New global.
33129 (format_integer): Use it here and throughout the rest of the file.
33130 Use the same switch to compute sign as base.
33131 (maybe_warn): New function.
33132 (format_directive): Factor out warnings into maybe_warn.
33133 Add debugging output. Use warn_level.
33134 (add_bytes): Use warn_level.
33135 (pass_sprintf_length::compute_format_length): Add debugging output.
33136 (try_substitute_return_value): Same.
33137 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33139 PR middle-end/78703
33140 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33141 (struct fmtresult, format_integer, format_floating): Adjust.
33142 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33143 (get_string_length, format_string,format_directive): Same.
33144 (pass_sprintf_length::compute_format_length): Same.
33145 (try_substitute_return_value): Simplify slightly.
33147 PR middle-end/78703
33148 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33149 (fmtresult::operator+=): Outlined.
33150 (struct fmtresult): Add ctors.
33151 (struct conversion_spec): Rename...
33152 (struct directive): ...to this. Add and remove data members.
33153 (directive::set_width, directive::set_precision): New functions.
33154 (format_percent): Use fmtresult ctor.
33155 (get_width_and_precision): Remove.
33156 (format_integer): Make naming changes. Avoid computing width and
33158 (format_floating): Same. Adjust indentation.
33159 (format_character, format_none): New functions.
33160 (format_string): Moved character handling to format_character.
33161 (format_directive): Remove arguments, change return type.
33162 (parse_directive): New function.
33163 (pass_sprintf_length::compute_format_length): Move directive
33164 parsing to parse_directive.
33166 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33168 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33169 (assign_assembler_name_if_needed): ... this.
33170 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33171 (assign_assembler_name_if_needed): ... this.
33172 (free_lang_data_in_cgraph): Adjust callers.
33173 * cgraphunit.c (cgraph_node::analyze): Likewise.
33174 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33177 2017-01-23 Richard Biener <rguenther@suse.de>
33179 PR tree-optimization/79088
33180 PR tree-optimization/79188
33181 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33182 resetting loop bounds after last path deletion. Reset loop
33183 bounds of the target loop, make code match the comments.
33184 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33185 Make sure loops need no fixups.
33187 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33189 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33190 exponent support with double type for first argument.
33191 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33192 type returned by __builtin_vec_extract_sig,
33193 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33194 functions from "vector int" to "vector unsigned int" or from
33195 "vector long long int" to "vector unsigned long long int".
33196 Changed type returned by __builtin_vec_extract_exp,
33197 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33198 functions from "vector int" to "vector unsigned int" or from
33199 "vector long long int" to "vector unsigned long long int".
33200 Changed return type of __builtin_vec_test_data_class,
33201 __builtin_vec_test_data_class_sp, and
33202 __builtin_vec_test_data_class_dp from "vector int" to
33203 "vector bool int" or from "vector long long int" to "vector bool
33204 long long int" and changed second argument type from "unsigned
33205 int" to "int". Added new overloaded function forms "vector float
33206 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33207 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33208 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33209 double, vector unsigned long long int)" and "vector double
33210 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33211 long int)". Changed return type of
33212 __builtin_scalar_test_data_class and
33213 __builtin_scalar_test_data_class_sp and
33214 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33215 int" and changed second argument from "unsigned int" to "int".
33216 Changed type returned by __builtin_scalar_test_neg,
33217 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33218 from "int" to "bool int". Added new overloaded function form
33219 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33220 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33221 exponent double-precision with floating point first argument.
33222 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33223 documentation of scalar_test_data_class, scalar_test_neg,
33224 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33225 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33226 vec_test_data_class built-in functions to reflect refinements in
33227 their type signatures.
33229 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33231 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33233 (aarch64_elf_asm_destructor): Likewise.
33235 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33237 PR rtl-optimization/78634
33238 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33239 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33240 * ifcvt.c (noce_try_cmove): Add missing cost check.
33242 PR rtl-optimization/71724
33243 * combine.c (if_then_else_cond): Look for situations where it is
33244 beneficial to undo the work of one of the recursive calls.
33246 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33248 PR tree-optimization/70754
33249 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33250 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33251 combined stmt before it if not NULL.
33252 (combine_chains): Process refs reversely and compute dominance point
33255 2017-01-23 Martin Liska <mliska@suse.cz>
33257 PR tree-optimization/79196
33258 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33259 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33261 (strlen_optimize_stmt): Call the renamed function.
33263 2017-01-23 Michael Matz <matz@suse.de>
33265 PR tree-optimization/78384
33266 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33268 2017-01-23 Richard Biener <rguenther@suse.de>
33270 PR tree-optimization/79186
33271 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33272 both incoming edges before moving an assert.
33274 2017-01-23 Martin Jambor <mjambor@suse.cz>
33276 * ipa-prop.c (load_from_param_1): Removed.
33277 (load_from_unmodified_param): Bits from load_from_param_1 put back
33279 (load_from_param): Removed.
33280 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33281 with stmt. Reverted back to use of load_from_unmodified_param.
33283 2017-01-23 Martin Jambor <mjambor@suse.cz>
33286 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33287 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33288 field a pointer to garbage collected vector, mark lattices and
33289 ipcp_orig_node with GTY((skip)).
33290 (ipa_get_param_count): Adjust to descriptors being a pointer.
33291 (ipa_get_param): Likewise.
33292 (ipa_get_type): Likewise.
33293 (ipa_get_param_move_cost): Likewise.
33294 (ipa_set_param_used): Likewise.
33295 (ipa_get_controlled_uses): Likewise.
33296 (ipa_set_controlled_uses): Likewise.
33297 (ipa_is_param_used): Likewise.
33298 (ipa_node_params_t): Move into garbage collector. New methods insert
33300 (ipa_node_params_sum): Annotate wth GTY(()).
33301 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33303 (ipa_load_from_parm_agg): Adjust declaration.
33304 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33305 * ipa-profile.c (ipa_profile): Likewise.
33306 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33307 (ipa_populate_param_decls): Make descriptors parameter garbage
33309 (ipa_dump_param): Adjust to descriptors being a pointer.
33310 (ipa_alloc_node_params): Likewise.
33311 (ipa_initialize_node_params): Likewise.
33312 (load_from_param_1): Make descriptors parameter garbage collected.
33313 (load_from_unmodified_param): Likewise.
33314 (load_from_param): Likewise.
33315 (ipa_load_from_parm_agg): Likewise.
33316 (ipa_node_params::~ipa_node_params): Removed.
33317 (ipa_free_all_node_params): Remove call to delete operator.
33318 (ipa_node_params_t::insert): New.
33319 (ipa_node_params_t::remove): Likewise.
33320 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33321 copy known_csts and known_contexts vectors.
33322 (ipa_read_node_info): Adjust to descriptors being a pointer.
33323 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33325 (ipcp_transform_function): Make descriptors variable garbage
33328 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33330 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33331 * config/i386/avx512dqintrin.h: Ditto.
33332 * config/i386/avx512fintrin.h: Ditto.
33333 * config/i386/i386.c: Handle new builtins.
33334 * config/i386/i386-builtin.def: Add new builtins.
33335 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33336 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33338 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33339 Martin Liska <mliska@suse.cz>
33341 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33342 * asan.c (asan_expand_poison_ifn): Support stores and use
33343 appropriate ASAN report function.
33344 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33345 * internal-fn.def (ASAN_POISON_USE): Declare.
33346 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33347 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33348 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33349 ASAN_POISON calls w/o LHS.
33350 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33351 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33352 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33353 * gimplify.c (asan_poison_variables): Add attribute
33354 use_after_scope_memory to variables that really needs to live
33356 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33357 having the attribute.
33359 2017-01-23 Martin Liska <mliska@suse.cz>
33361 * asan.c (create_asan_shadow_var): New function.
33362 (asan_expand_poison_ifn): Likewise.
33363 * asan.h (asan_expand_poison_ifn): New declaration.
33364 * internal-fn.c (expand_ASAN_POISON): Likewise.
33365 * internal-fn.def (ASAN_POISON): New builtin.
33366 * sanopt.c (pass_sanopt::execute): Expand
33367 asan_expand_poison_ifn.
33368 * tree-inline.c (copy_decl_for_dup_finish): Make function
33370 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33371 * tree-ssa.c (is_asan_mark_p): New function.
33372 (execute_update_addresses_taken): Rewrite local variables
33373 (identified just by use-after-scope as addressable) into SSA.
33375 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33377 * doc/install.texi (Specific): opensource.apple.com uses https
33378 now. Remove trailing slash.
33380 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33382 * README.Portability: Remove note on an Irix compatibility issue.
33384 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33386 * gcov.c (INCLUDE_ALGORITHM): Define.
33387 (INCLUDE_VECTOR): Define.
33388 No longer include <vector> and <algorithm> directly.
33390 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33392 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33394 * doc/invoke.texi (Code Gen Options): Ditto.
33396 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33399 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33401 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33403 rtl-optimization/79125
33404 * cprop.c (local_cprop_pass): Handle cases where we make an
33405 unconditional trap.
33407 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33411 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33412 read from, for big endian.
33414 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33416 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33417 register pauth builtins for LP64 only.
33419 2017-01-20 Marek Polacek <polacek@redhat.com>
33422 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33425 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33427 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33429 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33430 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33431 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33433 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33436 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33437 in big-endian mode when they are not a single duplicated value.
33439 2017-01-20 Richard Biener <rguenther@suse.de>
33441 * BASE-VER: Bump to 7.0.1.
33443 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33445 * omp-low.c (omplow_simd_context): New struct. Use it...
33446 (lower_rec_simd_input_clauses): ...here and...
33447 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33448 references to idx, lane, max_vf, is_simt.
33450 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33452 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33455 2017-01-20 Martin Jambor <mjambor@suse.cz>
33457 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33458 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33460 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33461 (GTFILES): Rename hsa.c to hsa-common.c.
33462 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33463 * hsa-dump.c: Likewise.
33464 * hsa-gen.c: Likewise.
33465 * hsa-regalloc.c: Likewise.
33466 * ipa-hsa.c: Likewise.
33467 * omp-expand.c: Likewise.
33468 * omp-low.c: Likewise.
33469 * toplev.c: Likewise.
33471 2017-01-20 Marek Polacek <polacek@redhat.com>
33474 * doc/invoke.texi: Document -Wduplicated-branches.
33475 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33476 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33477 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33478 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33479 return 0 only when not OEP_LEXICOGRAPHIC.
33480 (fold_build_cleanup_point_expr): Use the expression
33481 location when building CLEANUP_POINT_EXPR.
33482 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33483 * tree.c (add_expr): Handle error_mark_node.
33485 2017-01-20 Martin Liska <mliska@suse.cz>
33488 * tree-profile.c (init_ic_make_global_vars): Do not call
33490 (gimple_init_gcov_profiler): Likewise.
33492 2017-01-20 Martin Liska <mliska@suse.cz>
33495 * cgraph.h (maybe_create_reference): Remove argument and
33497 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33499 * ipa-cp.c (create_specialized_node): Likewise.
33500 * symtab.c (symtab_node::maybe_create_reference): Handle
33501 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33503 2017-01-20 Martin Liska <mliska@suse.cz>
33505 * read-rtl-function.c (function_reader::create_function): Use
33506 build_decl instread of build_decl_stat.
33508 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33510 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33511 * config/i386/avx512dqintrin.h: Ditto.
33512 * config/i386/avx512fintrin.h: Ditto.
33513 * config/i386/i386-builtin-types.def: Add new types.
33514 * config/i386/i386.c: Handle new types.
33515 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33516 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33517 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33518 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33519 (__builtin_ia32_kshiftridi): New.
33520 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33522 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33526 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33527 define to rs6000_init_stack_protect_guard.
33528 (rs6000_init_stack_protect_guard): New function.
33530 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33531 Yunqiang Su <yunqiang.su@imgtec.com>
33533 * config.gcc (supported_defaults): Add madd4.
33534 (with_madd4): Add validation.
33535 (all_defaults): Add madd4.
33536 * config/mips/mips.opt (mmadd4): New option.
33537 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33539 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33541 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33542 (ISA_HAS_FUSED_MADD4): Likewise.
33543 * doc/invoke.texi (-mmadd4): Document the new option.
33544 * doc/install.texi (--with-madd4): Document the new option.
33546 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33548 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33549 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33550 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33551 (aarch64_init_pauth_hint_builtins): New.
33552 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33553 (aarch64_expand_builtin): Expand new builtins.
33555 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33557 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33558 * combine-stack-adj.c (no_unhandled_cfa): Handle
33559 REG_CFA_TOGGLE_RA_MANGLE.
33560 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33561 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33562 info for return address signing.
33563 (aarch64_expand_epilogue): Likewise.
33565 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33567 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33568 * config/aarch64/aarch64-protos.h
33569 (aarch64_return_address_signing_enabled): New declaration.
33570 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33572 (aarch64_expand_prologue): Sign return address before it's pushed onto
33574 (aarch64_expand_epilogue): Authenticate return address fetched from
33576 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33577 (aarch64_attributes): New function attributes for "sign-return-address".
33578 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33579 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33580 ("*do_return"): Generate combined instructions according to key index.
33581 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33582 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33584 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33585 * config/aarch64/aarch64.opt (msign-return-address=): New.
33586 * doc/extend.texi (AArch64 Function Attributes): Documents
33587 "sign-return-address=".
33588 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33590 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33592 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33593 overall option summary.
33595 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33597 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33598 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33599 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33600 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33602 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33604 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33605 -mpower9-minmax by default for -mcpu=power9.
33606 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33607 128-bit floating point.
33609 2017-01-20 Alan Modra <amodra@gmail.com>
33611 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33612 optimizing for size.
33614 2017-01-20 Alan Modra <amodra@gmail.com>
33617 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33618 for strcmp and strncmp from corresponding builtin decl.
33620 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33622 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33623 instead of i386/rtems-64.h.
33624 * config/i386/rtems-64.h: Remove.
33626 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33630 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33632 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33634 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33636 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33637 Change int to HOST_WIDE_INT.
33638 * config/aarch64/aarch64-protos.h
33639 (aarch64_simd_gen_const_vector_dup): Likewise.
33640 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33642 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33644 * langhooks-def.h (lhd_type_for_size): New decl.
33645 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33646 * langhooks.c (lhd_type_for_size): New function, taken from
33649 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33651 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33652 define_bypass for CR latency.
33653 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33654 (power9-alu2): Add define_bypass for CR latency.
33656 (power9-mul): Update insn latency.
33657 (power9-mul-compare): Update insn latency, bypass latency and remove
33660 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33662 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33664 * config/aarch64/aarch64.md
33665 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33666 aarch64_nopcrelative_literal_loads.
33667 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33669 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33671 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33672 TARGET_LOONGSON_3A.
33673 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33675 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33678 * config.gcc (supported_defaults): Add lxc1-sxc1.
33679 (with_lxc1_sxc1): Add validation.
33680 (all_defaults): Add lxc1-sxc1.
33681 * config/mips/mips.opt (mlxc1-sxc1): New option.
33682 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33684 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33685 __mips_no_lxc1_sxc1.
33686 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33687 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33688 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33690 2017-01-19 Richard Biener <rguenther@suse.de>
33692 PR tree-optimization/72488
33693 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33694 sure to restore SSA info.
33695 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33697 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33699 PR rtl-optimization/79121
33700 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33701 of the inner type when shifting an extended value.
33703 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33706 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33707 interposable aliases.
33709 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33712 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33713 Use the evmergelohi instruction.
33714 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33715 (mov_sitf_e500_subreg8_2_be): Likewise.
33716 (mov_sitf_e500_subreg12_2_le): Likewise.
33717 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33718 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33719 (mov_sitf_e500_subreg8_2_le): Likewise.
33720 (mov_sitf_e500_subreg12_2_be): Likewise.
33722 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33724 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33725 attribute from vecsimple to vecperm.
33726 (altivec_vbpermq2): Likewise.
33728 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33731 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33733 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33734 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33735 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33736 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33737 case where N arg is SIZE_MAX.
33738 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33739 (cmpstrsi): Add pattern.
33741 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33743 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33744 __builtin_vec_revb builtins.
33745 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33746 built-in functions to support generation of the ISA 3.0 XXBR<x>
33747 vector byte reverse instructions.
33748 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33749 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33750 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33751 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33752 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33753 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33754 (P9V_BUILTIN_VEC_REVB): Likewise.
33755 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33756 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33757 (p9_xxbrq_v16qi): Likewise.
33758 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33759 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33760 (p9_xxbrh_v8hi): Likewise.
33761 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33762 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33763 vec_revb built-in functions.
33765 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33767 PR rtl-optimization/78952
33768 * config/i386/i386.md (any_extract): New code iterator.
33769 (*insvqi_2): Use any_extract for source operand.
33770 (*insvqi_3): Use any_shiftrt for source operand.
33772 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33774 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33776 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33778 2017-01-18 Matthias Klose <doko@ubuntu.com>
33780 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33782 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33784 * config/rs6000/altivec.h (vec_bperm): Change #define.
33785 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33786 (altivec_vbpermq2): New define_insn.
33787 (altivec_vbpermd): Likewise.
33788 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33789 function interface.
33790 (VBPERMD): Likewise.
33791 (VBPERM): New polymorphic function interface.
33792 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33793 Add entries for P9V_BUILTIN_VEC_VBPERM.
33794 * doc/extend.texi: Add interfaces for vec_bperm.
33796 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33798 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33799 first letter of error messages.
33800 (s390_resolve_overloaded_builtin): Likewise.
33801 * config/s390/s390.c (s390_expand_builtin): Likewise.
33802 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33803 (s390_valid_target_attribute_inner_p): Likewise.
33804 * config/s390/s390.md ("tabort"): Likewise.
33806 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33808 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33809 (ISA_AVOID_DIV_HILO): New macro.
33810 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33811 (ISA_HAS_DDIV): Likewise.
33813 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33815 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33817 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33819 * doc/invoke.texi (fabi-version): Spelling fix.
33821 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33824 * doc/invoke.texi (fabi-version): Mention mangling fix for
33827 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33830 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33832 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33835 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33836 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33838 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33840 (stack_protect_test): Ditto.
33841 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33842 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33844 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33845 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33846 -mstack-protector-guard-offset=.
33847 (RS/6000 and PowerPC Options): Ditto.
33849 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33851 * config/i386/i386.h (MASK_CLASS_P): New define.
33852 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33853 there are no registers from different register sets also when
33854 mask registers are used. Update function comment.
33855 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33856 to (*k/*r) and (*k/*km) alternatives.
33858 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33860 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33861 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33862 (EH_RETURN_HANDLER_RTX): New define.
33863 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33864 Force frame pointer in EH return functions.
33865 (aarch64_expand_epilogue): Add barrier for eh_return.
33866 (aarch64_final_eh_return_addr): Remove.
33867 (aarch64_eh_return_handler_rtx): New function.
33868 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33870 (aarch64_eh_return_handler_rtx): New prototype.
33872 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33874 * config/rs6000/altivec.h (vec_rlmi): New #define.
33875 (vec_vrlnm): Likewise.
33876 (vec_rlnm): Likewise.
33877 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33878 (UNSPEC_VRLNM): Likewise.
33879 (VIlong): New mode iterator.
33880 (altivec_vrl<VI_char>mi): New define_insn.
33881 (altivec_vrl<VI_char>nm): Likewise.
33882 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33884 (VRLDNM): Likewise.
33885 (RLNM): New polymorphic function entry.
33886 (VRLWMI): New monomorphic function entry.
33887 (VRLDMI): Likewise.
33888 (RLMI): New polymorphic function entry.
33889 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33890 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33891 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33894 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33897 * dwarf2out.c (field_byte_offset): Restore the
33898 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33899 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33900 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33903 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33906 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33908 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33911 * dwarf2out.c (add_data_member_location_attribute): For constant
33912 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33913 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33914 DW_AT_byte_size attributes.
33916 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33918 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33919 after forcing to constant memory when the code model is medium.
33921 2017-01-17 Julia Koval <julia.koval@intel.com>
33924 * config/i386/avx512fintrin.h
33925 (_mm512_i32gather_ps): Change __addr type to void const*.
33926 (_mm512_mask_i32gather_ps): Ditto.
33927 (_mm512_i32gather_pd): Ditto.
33928 (_mm512_mask_i32gather_pd): Ditto.
33929 (_mm512_i64gather_ps): Ditto.
33930 (_mm512_mask_i64gather_ps): Ditto.
33931 (_mm512_i64gather_pd): Ditto.
33932 (_mm512_mask_i64gather_pd): Ditto.
33933 (_mm512_i32gather_epi32): Ditto.
33934 (_mm512_mask_i32gather_epi32): Ditto.
33935 (_mm512_i32gather_epi64): Ditto.
33936 (_mm512_mask_i32gather_epi64): Ditto.
33937 (_mm512_i64gather_epi32): Ditto.
33938 (_mm512_mask_i64gather_epi32): Ditto.
33939 (_mm512_i64gather_epi64): Ditto.
33940 (_mm512_mask_i64gather_epi64): Ditto.
33941 (_mm512_i32scatter_ps): Change __addr type to void*.
33942 (_mm512_mask_i32scatter_ps): Ditto.
33943 (_mm512_i32scatter_pd): Ditto.
33944 (_mm512_mask_i32scatter_pd): Ditto.
33945 (_mm512_i64scatter_ps): Ditto.
33946 (_mm512_mask_i64scatter_ps): Ditto.
33947 (_mm512_i64scatter_pd): Ditto.
33948 (_mm512_mask_i64scatter_pd): Ditto.
33949 (_mm512_i32scatter_epi32): Ditto.
33950 (_mm512_mask_i32scatter_epi32): Ditto.
33951 (_mm512_i32scatter_epi64): Ditto.
33952 (_mm512_mask_i32scatter_epi64): Ditto.
33953 (_mm512_i64scatter_epi32): Ditto.
33954 (_mm512_mask_i64scatter_epi32): Ditto.
33955 (_mm512_i64scatter_epi64): Ditto.
33956 (_mm512_mask_i64scatter_epi64): Ditto.
33957 * config/i386/avx512pfintrin.h
33958 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33959 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33960 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33961 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33962 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33963 (_mm512_prefetch_i32scatter_ps): Ditto.
33964 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33965 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33966 (_mm512_prefetch_i64scatter_pd): Ditto.
33967 (_mm512_prefetch_i64scatter_ps): Ditto.
33968 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33969 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33970 * config/i386/avx512vlintrin.h
33971 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33972 (_mm_mmask_i32gather_ps): Ditto.
33973 (_mm256_mmask_i32gather_pd): Ditto.
33974 (_mm_mmask_i32gather_pd): Ditto.
33975 (_mm256_mmask_i64gather_ps): Ditto.
33976 (_mm_mmask_i64gather_ps): Ditto.
33977 (_mm256_mmask_i64gather_pd): Ditto.
33978 (_mm_mmask_i64gather_pd): Ditto.
33979 (_mm256_mmask_i32gather_epi32): Ditto.
33980 (_mm_mmask_i32gather_epi32): Ditto.
33981 (_mm256_mmask_i32gather_epi64): Ditto.
33982 (_mm_mmask_i32gather_epi64): Ditto.
33983 (_mm256_mmask_i64gather_epi32): Ditto.
33984 (_mm_mmask_i64gather_epi32): Ditto.
33985 (_mm256_mmask_i64gather_epi64): Ditto.
33986 (_mm_mmask_i64gather_epi64): Ditto.
33987 (_mm256_i32scatter_ps): Change __addr type to void*.
33988 (_mm256_mask_i32scatter_ps): Ditto.
33989 (_mm_i32scatter_ps): Ditto.
33990 (_mm_mask_i32scatter_ps): Ditto.
33991 (_mm256_i32scatter_pd): Ditto.
33992 (_mm256_mask_i32scatter_pd): Ditto.
33993 (_mm_i32scatter_pd): Ditto.
33994 (_mm_mask_i32scatter_pd): Ditto.
33995 (_mm256_i64scatter_ps): Ditto.
33996 (_mm256_mask_i64scatter_ps): Ditto.
33997 (_mm_i64scatter_ps): Ditto.
33998 (_mm_mask_i64scatter_ps): Ditto.
33999 (_mm256_i64scatter_pd): Ditto.
34000 (_mm256_mask_i64scatter_pd): Ditto.
34001 (_mm_i64scatter_pd): Ditto.
34002 (_mm_mask_i64scatter_pd): Ditto.
34003 (_mm256_i32scatter_epi32): Ditto.
34004 (_mm256_mask_i32scatter_epi32): Ditto.
34005 (_mm_i32scatter_epi32): Ditto.
34006 (_mm_mask_i32scatter_epi32): Ditto.
34007 (_mm256_i32scatter_epi64): Ditto.
34008 (_mm256_mask_i32scatter_epi64): Ditto.
34009 (_mm_i32scatter_epi64): Ditto.
34010 (_mm_mask_i32scatter_epi64): Ditto.
34011 (_mm256_i64scatter_epi32): Ditto.
34012 (_mm256_mask_i64scatter_epi32): Ditto.
34013 (_mm_i64scatter_epi32): Ditto.
34014 (_mm_mask_i64scatter_epi32): Ditto.
34015 (_mm256_i64scatter_epi64): Ditto.
34016 (_mm256_mask_i64scatter_epi64): Ditto.
34017 (_mm_i64scatter_epi64): Ditto.
34018 (_mm_mask_i64scatter_epi64): Ditto.
34019 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34020 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34021 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34022 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34023 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34024 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34025 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34026 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34027 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34028 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34029 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34030 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34031 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34032 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34033 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34034 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34035 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34036 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34037 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34038 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34039 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34040 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34041 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34042 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34043 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34044 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34045 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34046 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34047 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34048 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34049 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34050 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34051 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34052 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34053 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34054 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34055 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34056 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34057 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34058 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34059 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34060 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34061 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34062 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34063 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34064 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34065 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34066 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34067 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34068 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34069 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34070 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34071 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34072 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34073 definitions accordingly.
34075 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34076 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34079 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34082 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34085 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34086 hard regs for inner regno.
34088 2017-01-17 Martin Liska <mliska@suse.cz>
34091 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34092 assumption and add comment.
34094 2017-01-17 Nathan Sidwell <nathan@acm.org>
34096 * ipa-visibility.c (localize_node): New function, broken out of ...
34097 (function_and_variable_visibility): ... here. Call it.
34099 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34101 PR middle-end/77445
34102 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34103 correctly set frequency of oudgoing edge.
34104 (duplicate_thread_path): Fix profile updating.
34106 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34109 * configure.ac: Add GCC_BASE_VER.
34110 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34111 version from BASE-VER file.
34112 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34113 (gcc.o): Depend on $(BASEVER).
34114 * common.opt (dumpfullversion): New option.
34115 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34116 * doc/invoke.texi: Document -dumpfullversion.
34117 * doc/install.texi: Document --with-gcc-major-version-only.
34118 * configure: Regenerated.
34120 2017-01-17 Richard Biener <rguenther@suse.de>
34122 PR tree-optimization/71433
34123 * tree-vrp.c (register_new_assert_for): Merge same asserts
34124 on all incoming edges.
34125 (process_assert_insertions_for): Handle insertions at the
34128 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34130 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34131 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34133 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34136 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34139 2017-01-17 Alan Modra <amodra@gmail.com>
34142 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34143 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34144 symbolic stack limit when pic.
34146 2017-01-16 Martin Sebor <msebor@redhat.com>
34148 PR tree-optimization/78608
34149 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34151 2017-01-16 Jeff Law <law@redhat.com>
34154 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34155 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34156 for several include directories that may be relative to sysroot.
34157 * config/i386/x-mingw32 (gplus_includedir): Define.
34158 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34159 (native_system_includedir): Likewise.
34160 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34161 override if TARGET_SYSTEM_ROOT is defined.
34162 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34164 PR tree-optimization/79090
34165 PR tree-optimization/33562
34166 PR tree-optimization/61912
34167 PR tree-optimization/77485
34168 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34169 and computed trims into the dump file.
34171 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34173 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34175 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34178 * gimplify.c (gimplify_init_constructor): If want_value and
34179 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34183 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34184 sequence. Formatting fixes.
34185 (doloop_optimize): Formatting fixes.
34188 * gcc.c (debug_level_greater_than_spec_func): New function.
34189 (static_spec_functions): Add debug-level-gt spec function.
34190 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34192 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34193 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34194 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34195 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34196 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34197 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34199 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34201 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34202 QImode fixups to general and mask registers only.
34204 2017-01-16 Carl Love <cel@us.ibm.com>
34206 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34207 for built-in functions
34208 vector signed char vec_nabs (vector signed char)
34209 vector signed short vec_nabs (vector signed short)
34210 vector signed int vec_nabs (vector signed int)
34211 vector signed long long vec_nabs (vector signed long long)
34212 vector float vec_nabs (vector float)
34213 vector double vec_nabs (vector double)
34214 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34216 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34217 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34218 * doc/extend.texi: Update the documentation file for the new built-in
34221 2017-01-16 Martin Sebor <msebor@redhat.com>
34223 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34226 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34228 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34229 UNSPEC_VSX__XXSPLTD to require special splat handling.
34231 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34234 * system.h: Poison strndup.
34236 2017-01-16 Alan Modra <amodra@gmail.com>
34239 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34242 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34244 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34246 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34248 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34249 call recog here. Assert that INSN_CODE (insn) is non-negative.
34251 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34254 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34256 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34257 in the currently scheduled RTL fragment.
34259 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34261 PR rtl-optimization/78751
34262 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34265 2017-01-14 Jeff Law <law@redhat.com>
34267 PR tree-optimization/79090
34268 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34269 variable length stores.
34270 (compute_trims): Delete dead assignment to *trim_tail.
34271 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34274 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34276 PR rtl-optimization/78626
34277 PR rtl-optimization/78727
34278 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34279 of a block, and split such blocks after everything else is finished.
34281 2017-01-14 Alan Modra <amodra@gmail.com>
34284 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34285 target legitimate_combined_insn.
34286 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34287 (rs6000_legitimate_combined_insn): New function.
34288 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34290 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34291 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34292 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34294 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34296 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34298 2017-01-13 Jeff Law <law@redhat.com>
34300 PR tree-optimization/33562
34301 PR tree-optimization/61912
34302 PR tree-optimization/77485
34303 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34305 (delete_dead_assignment): Likewise.
34306 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34307 statement to delete_dead_call and delete_dead_assignment.
34309 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34312 * substring-locations.c (format_warning_va): Strengthen case 1 so
34313 that both endpoints of the substring must be within the format
34314 range for just the substring to be printed.
34316 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34318 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34319 * config/i386/i386.c (ix86_target_string): Add missing options
34320 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34321 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34322 flags_other and ix86_target_other to flags2_other. Display unknown
34324 (ix86_valid_target_attribute_inner_p): Add missing options and
34325 reorder options by implied ISAs, as in ix86_target_string.
34327 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34329 * hash-table.h (hash_table::too_empty_p): New function.
34330 (hash_table::expand): Use it.
34331 (hash_table::traverse): Likewise.
34332 (hash_table::empty_slot): Use sizeof (value_type) instead of
34333 sizeof (PTR) to convert bytes to elements. Shrink the table
34334 if the current size is excessive for the current number of
34337 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34339 * ira-costs.c (record_reg_classes): Break from the inner loop
34340 early once alt_fail is known to be true. Update outer loop
34341 handling accordingly.
34343 2017-01-13 Jeff Law <law@redhat.com>
34345 * tree-ssa-dse.c (decrement_count): New function.
34346 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34347 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34348 when we know the partially dead statement is a mem* function.
34350 PR tree-optimization/61912
34351 PR tree-optimization/77485
34352 * tree-ssa-dse.c: Include expr.h.
34353 (maybe_trim_constructor_store): New function.
34354 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34356 PR tree-optimization/33562
34357 PR tree-optimization/61912
34358 PR tree-optimization/77485
34359 * doc/invoke.texi: Document new dse-max-object-size param.
34360 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34361 * tree-ssa-dse.c: Include params.h.
34362 (dse_store_status): New enum.
34363 (initialize_ao_ref_for_dse): New, partially extracted from
34365 (valid_ao_ref_for_dse, normalize_ref): New.
34366 (setup_live_bytes_from_ref, compute_trims): Likewise.
34367 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34368 (maybe_trim_partially_dead_store): Likewise.
34369 (maybe_trim_complex_store): Likewise.
34370 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34371 Track what bytes live from the original store. Return tri-state
34372 for dead, partially dead or live.
34373 (dse_dom_walker): Add constructor, destructor and new private members.
34374 (delete_dead_call, delete_dead_assignment): New extracted from
34376 (dse_optimize_stmt): Make a member of dse_dom_walker.
34377 Use initialize_ao_ref_for_dse.
34379 PR tree-optimization/33562
34380 PR tree-optimization/61912
34381 PR tree-optimization/77485
34382 * sbitmap.h (bitmap_count_bits): Prototype.
34383 (bitmap_clear_range, bitmap_set_range): Likewise.
34384 * sbitmap.c (bitmap_clear_range): New function.
34385 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34387 2017-01-13 Martin Liska <mliska@suse.cz>
34390 * function.c (set_cfun): Add new argument force.
34391 * function.h (set_cfun): Likewise.
34392 * ipa-inline-transform.c (inline_call): Use the function when
34393 strict alising from is dropped for function we inline to.
34395 2017-01-13 Richard Biener <rguenther@suse.de>
34397 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34398 for dumping GIMPLE INTEGER_CSTs.
34400 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34402 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34403 to 201112L since C++17.
34405 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34408 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34409 if -fsanitize=kernel-address is present.
34411 2017-01-13 Richard Biener <rguenther@suse.de>
34413 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34414 as _Literal ( type ) number in case usual suffixes do not
34415 preserve all information.
34417 2017-01-13 Richard Biener <rguenther@suse.de>
34419 PR tree-optimization/77283
34420 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34421 and ssa-iterators.h.
34422 (is_feasible_trace): Implement a cost model based on joiner
34425 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34428 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34429 char or short to __float128/_Float128 directly.
34431 2017-01-12 Martin Sebor <msebor@redhat.com>
34433 to -Wformat-overflow.
34434 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34435 (min_bytes_remaining): Same.
34436 (get_string_length): Same.
34437 (format_string): Same.
34438 (format_directive): Same.
34440 (pass_sprintf_length::handle_gimple_call): Same.
34442 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34444 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34445 info.nowrite calls with no lhs that can't throw. Return bool
34446 whether gsi_remove has been called or not.
34447 (pass_sprintf_length::handle_gimple_call): Return bool whether
34448 try_substitute_return_value called gsi_remove. Formatting fix.
34449 (pass_sprintf_length::execute): Don't use gsi_remove if
34450 handle_gimple_call returned true.
34453 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34454 be removed due to side-effects, don't remove following barrier nor
34455 turn the successor edge into fallthru edge.
34457 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34460 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34461 element-reversing loads and stores as not swappable.
34463 2017-01-12 Nathan Sidwell <nathan@acm.org>
34464 Nicolai Stange <nicstange@gmail.com>
34466 * combine.c (try_combine): Don't ignore result of overlap checking
34467 loop. Combine overlap & asm check into single loop.
34469 2017-01-12 Richard Biener <rguenther@suse.de>
34471 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34472 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34474 2017-01-12 Richard Biener <rguenther@suse.de>
34476 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34477 and TS_TARGET_OPTION directly derive from TS_BASE.
34478 * tree-core.h (tree_optimization_option): Derive from tree_base.
34479 (tree_target_option): Likewise.
34481 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34483 * config/i386/i386.c (memory_address_length): Increase len
34484 only when rip_relative_addr_p returns false.
34486 2017-01-11 Julia Koval <julia.koval@intel.com>
34488 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34489 (OPTION_MASK_ISA_SGX_SET): New.
34490 (ix86_handle_option): Handle OPT_msgx.
34491 * config.gcc: Added sgxintrin.h.
34492 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34493 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34494 * config/i386/i386.c (ix86_target_string): Add -msgx.
34496 (ix86_option_override_internal): Handle new options.
34497 (ix86_valid_target_attribute_inner_p): Add sgx.
34498 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34499 * config/i386/i386.opt: Add msgx.
34500 * config/i386/sgxintrin.h: New file.
34501 * config/i386/x86intrin.h: Add sgxintrin.h.
34503 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34506 * fold-const.c (maybe_nonzero_address): Return 1 for function
34508 (tree_single_nonzero_warnv_p): Don't handle function local objects
34512 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34515 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34518 * opts.c: Include "spellcheck.h"
34519 (struct string_fragment): New struct.
34520 (struct edit_distance_traits<const string_fragment &>): New
34522 (get_closest_sanitizer_option): New function.
34523 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34525 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34527 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34529 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34530 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34531 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34532 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34533 for initial die_offset if dwarf_split_debug_info.
34534 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34535 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34537 (output_skeleton_debug_sections): Formatting fix. Use
34538 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34539 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34541 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34543 * config/arm/cortex-a53.md: Add bypasses for
34544 cortex_a53_r2f_cvt.
34545 (cortex_a53_r2f): Only use for transfers.
34546 (cortex_a53_f2r): Likewise.
34547 (cortex_a53_r2f_cvt): Add reservation for conversions.
34548 (cortex_a53_f2r_cvt): Likewise.
34550 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34552 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34553 to all inlined functions, change static to extern.
34555 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34558 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34560 (arm_assemble_integer): Likewise.
34562 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34564 * config.gcc: Use new awk script to check CPU, FPU and architecture
34565 parameters for --with-... options.
34566 * config/arm/parsecpu.awk: New file
34567 * config/arm/arm-cpus.in: New file.
34568 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34570 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34572 * config/arm/t-arm: Update dependency rules.
34573 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34574 of processing .def files.
34575 * config/arm/genopt.sh: Deleted.
34576 * config/arm/gentune.sh: Deleted.
34577 * config/arm/arm-cores.def: Deleted.
34578 * config/arm/arm-arches.def: Deleted.
34579 * config/arm/arm-fpus.def: Deleted.
34580 * config/arm/arm-tune.md: Regenerated.
34581 * config/arm/arm-tables.opt: Regenerated.
34582 * config/arm/arm-cpu.h: New generated file.
34583 * config/arm/arm-cpu-data.h: New generated file.
34584 * config/arm/arm-cpu-cdata.h: New generated file.
34586 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34589 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34591 (input_varpool_node): Unpack dynamically_initialized bit.
34593 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34595 PR rtl-optimization/79032
34596 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34597 the alignment of the adjusted memory reference against that of MODE,
34598 instead of the alignment of the original memory reference.
34600 2017-01-11 Martin Jambor <mjambor@suse.cz>
34602 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34604 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34605 decorated functions.
34607 2017-01-11 Richard Biener <rguenther@suse.de>
34609 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34610 set range/nonnull info for PHI results. Do not set it on
34611 stmts marked for removal.
34613 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34615 * expr.c (store_field): In the bitfield case, fetch the return value
34616 from the registers before applying a single big-endian adjustment.
34617 Always do a final load for a BLKmode value not larger than a word.
34619 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34622 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34623 that we correctly handle column numbers greater than
34624 LINE_MAP_MAX_COLUMN_NUMBER.
34626 2017-01-10 Martin Sebor <msebor@redhat.com>
34628 PR middle-end/78245
34629 * gimple-ssa-sprintf.c (get_destination_size): Call
34630 {init,fini}object_sizes.
34631 * tree-object-size.c (addr_object_size): Adjust.
34632 (pass_through_call): Adjust.
34633 (pass_object_sizes::execute): Adjust.
34634 * tree-object-size.h (fini_object_sizes): Declare.
34636 2017-01-10 Martin Sebor <msebor@redhat.com>
34638 PR tree-optimization/78775
34639 * builtins.c (get_size_range): Move...
34640 * calls.c: ...to here.
34641 (alloc_max_size): Accept zero argument.
34642 (operand_signed_p): Remove.
34643 (maybe_warn_alloc_args_overflow): Call get_size_range.
34644 * calls.h (get_size_range): Declare.
34646 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34648 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34649 from TI's devices.csv file as of September 2016.
34650 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34652 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34654 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34655 * doc/invoke.texi: Likewise.
34656 * doc/md.texi: Likewise.
34657 * doc/objc.texi: Likewise.
34659 2017-01-10 Joshua Conner <joshconner@google.com>
34661 * config/arm/fuchsia-elf.h: New file.
34662 * config/fuchsia.h: New file.
34663 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34664 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34666 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34668 2016-01-10 Richard Biener <rguenther@suse.de>
34670 PR tree-optimization/79034
34671 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34672 Propagate out degenerate PHIs in the joiner.
34674 2017-01-10 Martin Liska <mliska@suse.cz>
34676 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34677 (sort_congruence_classes_by_decl_uid): Likewise.
34678 (sort_congruence_class_groups_by_decl_uid): Likewise.
34679 (sem_item_optimizer::merge_classes): Sort class, groups in these
34680 classes and members in the groups by DECL_UID of declarations.
34681 This would make merge operations stable.
34683 2017-01-10 Martin Liska <mliska@suse.cz>
34685 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34686 usage of m_classes_vec.
34687 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34688 (sem_item_optimizer::get_group_by_hash): Likewise.
34689 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34690 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34691 (sem_item_optimizer::verify_classes): Likewise.
34692 (sem_item_optimizer::process_cong_reduction): Likewise.
34693 (sem_item_optimizer::dump_cong_classes): Likewise.
34694 (sem_item_optimizer::merge_classes): Likewise.
34695 * ipa-icf.h (congruence_class_hash): Rename from
34696 congruence_class_group_hash. Remove declaration of m_classes_vec.
34698 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34700 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34701 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34702 * config.gcc: Add avx512vpopcntdqintrin.h.
34703 * config/i386/avx512vpopcntdqintrin.h: New.
34704 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34705 * config/i386/i386-builtin-types.def: Add new types.
34706 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34707 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34708 __builtin_ia32_vpopcountq_v8di_mask): New.
34709 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34710 __AVX512VPOPCNTDQ__.
34711 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34712 (PTA_AVX512VPOPCNTDQ): Define.
34713 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34714 TARGET_AVX512VPOPCNTDQ_P): Define.
34715 * config/i386/i386.opt: Add mavx512vpopcntdq.
34716 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34717 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34719 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34721 PR middle-end/77484
34722 * predict.def (PRED_CALL): Set to 67.
34724 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34726 * expr.c (store_field): In the bitfield case, if the value comes from
34727 a function call and is of an aggregate type returned in registers, do
34728 not modify the field mode; extract the value in all cases if the mode
34729 is BLKmode and the size is not larger than a word.
34731 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34734 * config/i386/cpuid.h: Fix undefined behavior.
34736 2017-01-04 Jeff Law <law@redhat.com>
34738 PR tree-optimization/79007
34739 PR tree-optimization/67955
34740 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34741 conservative for pt.null when flag_non_call_exceptions is on.
34743 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34745 PR translation/79019
34746 PR translation/79020
34747 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34748 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34749 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34751 * config/avr/avr.opt (maccumulate-args): Likewise.
34752 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34753 * common.opt (freport-bug): Likewise.
34754 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34755 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34756 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34757 translatable string.
34758 * config/i386/i386.c (function_value_32): Likewise.
34759 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34760 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34762 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34763 * common/config/msp430/msp430-common.c (msp430_handle_option):
34765 * symtab.c (symtab_node::verify_base): Likewise.
34766 * opts.c (set_debug_level): Likewise.
34767 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34769 missing whitespace to translatable strings.
34770 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34771 * config/sh/superh.h: Likewise.
34772 * config/i386/xopintrin.h: Likewise.
34773 * config/i386/znver1.md: Likewise.
34774 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34775 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34776 * double-int.h (struct double_int): Likewise.
34777 * double-int.c (div_and_round_double): Likewise.
34778 * wide-int.cc: Likewise.
34779 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34780 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34781 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34782 (crossjumps_occurred): ... this.
34783 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34786 PR tree-optimization/78899
34787 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34788 returning bool return struct loop *, NULL for failure and the new
34790 (versionable_outer_loop_p): Don't version outer loop if it has
34791 dont_vectorized bit set.
34792 (tree_if_conversion): When versioning outer loop, ensure
34793 tree_if_conversion is performed also on the inner loop of the
34794 non-vectorizable outer loop copy.
34795 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34796 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34797 prevent vectorization of it.
34798 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34799 the outer loop vectorization of the non-scalar version is attempted
34800 before vectorization of the inner loop in scalar version. If
34801 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34802 vectorization of its inner loop.
34803 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34804 has 2 inner loops, rename also on edges from bb whose single pred
34805 is outer_loop->header. Fix typo in function comment.
34807 2017-01-09 Martin Sebor <msebor@redhat.com>
34810 * asan.c (asan_emit_stack_protection): Increase local buffer size
34811 to avoid snprintf truncation warning.
34813 2017-01-09 Andrew Pinski <apinski@cavium.com>
34815 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34816 to reference thunderx2t99 for the tuning structure
34817 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34819 (thunderx2t99_extra_costs): This.
34820 * config/aarch64/aarch64-tune.md: Regenerate.
34821 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34822 (vulcan_addrcost_table): This.
34823 (vulcan_regmove_cost): Rename to ...
34824 (thunderx2t99_regmove_cost): This.
34825 (vulcan_vector_cost): Rename to ...
34826 (thunderx2t99_vector_cost): this.
34827 (vulcan_branch_cost): Rename to ...
34828 (thunderx2t99_branch_cost): This.
34829 (vulcan_tunings): Rename to ...
34830 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34831 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34833 2017-01-09 Martin Jambor <mjambor@suse.cz>
34837 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34838 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34839 (propagate_vr_accross_jump_function): Use the above function for all
34840 value range computations for pass-through jump functions and type
34841 converasion from explicit value range values.
34842 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34843 parameters from TYPE_ARG_TYPES.
34844 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34845 (ipa_write_node_info): Stream type of the actual argument.
34846 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34848 2017-01-09 Martin Liska <mliska@suse.cz>
34851 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34852 (lookup_compiler): Do not show error message with have_E.
34854 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34856 PR tree-optimization/78938
34857 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34858 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34859 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34860 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34863 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34865 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34868 2017-01-09 Richard Biener <rguenther@suse.de>
34870 PR tree-optimization/78997
34871 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34872 name condition properly.
34874 2017-01-09 Richard Biener <rguenther@suse.de>
34877 * dwarf2out.c (is_cxx): New overload with context.
34878 (is_naming_typedef_decl): Use it.
34880 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34882 * invoke.texi (Option Summary): Correct spacing in option lists
34883 and add line breaks to fix over-long lines.
34885 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34887 PR middle-end/17660
34889 * extend.texi (Common Variable Attributes): Add xref to GCC
34890 Internals manual to explain mode attribute keywords.
34892 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34895 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34896 and Preprocessor Options.
34897 (Options for Linking): Document -pthread here....
34898 (RS/6000 and PowerPC Options): ...not here.
34899 (Solaris 2 Options): ...or here.
34900 * doc/cppopts.texi: Document -pthread.
34902 2017-01-08 Martin Sebor <msebor@redhat.com>
34904 PR middle-end/77708
34905 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34906 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34907 New member functions.
34908 (format_directive): Used them.
34910 (pass_sprintf_length::handle_gimple_call): Same.
34911 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34912 to avoid truncation for any argument.
34913 (extract_affine_mul): Same.
34914 * tree.c (get_file_function_name): Same.
34916 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34918 PR middle-end/77484
34919 * predict.def (PRED_INDIR_CALL): Set to 86.
34921 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34923 PR preprocessor/54124
34924 * doc/cppopts.texi: Reformat -d subtable to list the full name
34925 of the options. Add cross-reference to the docs for the general
34926 compiler -d options.
34927 * doc/invoke.texi (Developer Options): Add cross-reference to the
34928 preprocessor-specific -d option documentation.
34930 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34932 PR preprocessor/13498
34933 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34934 redudant material, and reflect new command-line options.
34935 (System Headers): Likewise.
34937 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34939 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34940 -isystem, and -idirafter. Copy-edit.
34941 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34942 default for -ftrack-macro-expansion. Delete obsolete and
34943 badly-formatted implementation details about -fdebug-cpp output.
34944 * doc/cppwarnopts.texi: Copy-edit.
34946 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34949 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34950 that the transition from a max line width >= 1<<10 to narrower
34951 lines works correctly.
34953 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34955 * doc/options.texi (PerFunction): New.
34956 * opt-functions.awk (switch_flags): Map both Optimization and
34957 PerFunction to CL_OPTIMIZATION.
34958 * opth-gen.awk: Test for PerFunction flag along with
34960 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34961 it only when the latter is present. Skip those that don't in
34962 the hash function generator.
34963 * common.opt (fvar-tracking): Mark as PerFunction instead of
34965 (fvar-tracking-assignments): Likewise.
34966 (fvar-tracking-assignments-toggle): Likewise.
34967 (fvar-tracking-uninit): Likewise.
34969 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34971 PR translation/79018
34972 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34975 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34978 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34979 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34980 TARGET_LONG_JUMP_TABLE_OFFSETS.
34981 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34982 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34983 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34984 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34985 * config/m68k/m68k.md (tablejump expander): Likewise.
34986 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34987 TARGET_LONG_JUMP_TABLE_OFFSETS.
34988 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34989 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34991 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34992 David Holsgrove <david.holsgrove@xilinx.com>
34994 * common/config/microblaze/microblaze-common.c
34995 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34996 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34998 * config/microblaze/microblaze.c (microblaze_must_save_register)
34999 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35001 (microblaze_eh_return): New function.
35002 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35003 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35004 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35005 * config/microblaze/microblaze.md (eh_return): New pattern.
35007 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35009 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35010 GCC_DIAGNOSTIC_STRINGIFY): Define.
35012 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35014 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35016 * config/arm/arm.md (<mcrr>): New.
35018 * config/arm/arm.c (arm_arch5te): New.
35019 (arm_option_override): Set arm_arch5te.
35020 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35022 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35023 (arm_mcrr_qualifiers): ... this. New.
35024 (MRRC_QUALIFIERS): Define to...
35025 (arm_mrrc_qualifiers): ... this. New.
35026 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35028 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35029 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35030 (MRRCI, mrrc, MRRC): New.
35031 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35032 VUNSPEC_MRRC2): New.
35034 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35036 * config/arm/arm.md (<mcr>): New.
35038 * config/arm/arm.c (arm_coproc_builtin_available): Add
35039 support for mcr, mrc, mcr2 and mrc2.
35040 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35041 (arm_mcr_qualifiers): ... this. New.
35042 (MRC_QUALIFIERS): Define to ...
35043 (arm_mrc_qualifiers): ... this. New.
35044 (MCR_QUALIFIERS): Define to ...
35045 (arm_mcr_qualifiers): ... this. New.
35046 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35048 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35049 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35050 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35051 VUNSPEC_MRC2): New.
35053 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35055 * config/arm/arm.md (*ldc): New.
35059 * config/arm/arm.c (arm_coproc_builtin_available): Add
35060 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35061 (arm_coproc_ldc_stc_legitimate_address): New.
35062 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35063 'qualifier_const_pointer'.
35064 (LDC_QUALIFIERS): Define to...
35065 (arm_ldc_qualifiers): ... this. New.
35066 (STC_QUALIFIERS): Define to...
35067 (arm_stc_qualifiers): ... this. New.
35068 * config/arm/arm-protos.h
35069 (arm_coproc_ldc_stc_legitimate_address): New.
35070 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35071 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35072 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35073 stc2, stcl, stc2l): New.
35074 * config/arm/constraints.md (Uz): New.
35075 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35076 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35077 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35078 VUNSPEC_STC2L): New.
35080 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35082 * config/arm/arm.md (<cdp>): New.
35083 * config/arm/arm.c (neon_const_bounds): Rename this ...
35084 (arm_const_bounds): ... this.
35085 (arm_coproc_builtin_available): New.
35086 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35087 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35088 (CDP_QUALIFIERS): Define to...
35089 (arm_cdp_qualifiers): ... this. New.
35091 (arm_expand_builtin_args): Add case for 6 arguments.
35092 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35093 (arm_const_bounds): ... this.
35094 (arm_coproc_builtin_available): New.
35095 * config/arm/arm_acle.h (__arm_cdp): New.
35097 * config/arm/arm_acle_builtins.def (cdp): New.
35099 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35100 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35101 'arm_const_bounds'.
35102 * config/arm/types.md (coproc): New.
35103 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35104 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35105 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35106 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35108 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35110 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35111 (UBINOP_QUALIFIERS): New.
35113 (acle_builtin_data): New. Change comment.
35114 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35115 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35116 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35117 arm_acle_builtins.def.
35118 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35119 (arm_init_acle_builtins): New.
35120 (CRC32_BUILTIN): Remove.
35121 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35122 crc32cb, crc32ch and crc32cw.
35123 (arm_init_crc32_builtins): Remove.
35124 (arm_init_builtins): Use arm_init_acle_builtins rather
35125 than arm_init_crc32_builtins.
35126 (arm_expand_acle_builtin): New.
35127 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35128 * config/arm/arm_acle_builtins.def: New.
35130 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35132 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35133 (arm_builtin_datum): ... this.
35134 (arm_init_neon_builtin): Rename to ...
35135 (arm_init_builtin): ... this. Add a new parameters PREFIX
35136 and USE_SIG_IN_NAME.
35137 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35138 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35139 'arm_builtin_datum'.
35140 (arm_init_vfp_builtins): Likewise.
35141 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35142 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35143 (arm_expand_neon_args): Rename to ...
35144 (arm_expand_builtin_args): ... this. Rename builtin_arg
35145 enum values and differentiate between ARG_BUILTIN_MEMORY
35146 and ARG_BUILTIN_NEON_MEMORY.
35147 (arm_expand_neon_builtin_1): Rename to ...
35148 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35149 values, arm_expand_builtin_args and add bool parameter NEON.
35150 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35151 (arm_expand_vfp_builtin): Likewise.
35152 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35154 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35156 PR middle-end/77484
35157 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35158 * predict.c (tree_estimate_probability_bb): Reverse direction of
35159 polymorphic call predictor.
35161 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35163 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35164 (determine_pass_name_match): ...this new function and...
35165 (should_skip_pass_p): ...this new function.
35167 2017-01-06 Nathan Sidwell <nathan@acm.org>
35169 * ipa-visibility.c (function_and_variable_visibility): Reformat
35170 comments and long lines. Remove extrneous if.
35171 * symtab.c (symtab_node::make_decl_local): Fix code format.
35172 (symtab_node::set_section_for_node): Fix comment typo.
35174 2017-01-06 Martin Liska <mliska@suse.cz>
35177 * lra-constraints.c: Rename invariant to lra_invariant.
35178 * predict.c (set_even_probabilities): Initialize e to NULL.
35180 2017-01-05 Martin Sebor <msebor@redhat.com>
35182 PR tree-optimization/78910
35183 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35184 (format_integer): Correct off-by-one error in the handling
35185 of precision with negative numbers in signed conversions..
35187 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35189 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35191 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35193 PR tree-optimization/71016
35194 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35195 factor_out_conditional_conversion. Formatting fix.
35196 (factor_out_conditional_conversion): Add cond_stmt argument.
35197 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35198 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35201 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35203 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35204 read-rtl-function.o, and selftest-rtl.o.
35205 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35206 (selftest::aarch64_test_loading_full_dump): New function.
35207 (selftest::aarch64_run_selftests): New function.
35208 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35209 selftest::aarch64_run_selftests.
35210 * config/i386/i386.c
35211 (selftest::ix86_test_loading_dump_fragment_1): New function.
35212 (selftest::ix86_test_loading_call_insn): New function.
35213 (selftest::ix86_test_loading_full_dump): New function.
35214 (selftest::ix86_test_loading_unspec): New function.
35215 (selftest::ix86_run_selftests): Call the new functions.
35216 * emit-rtl.c (maybe_set_max_label_num): New function.
35217 * emit-rtl.h (maybe_set_max_label_num): New decl.
35218 * function.c (instantiate_decls): Guard call to
35219 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35220 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35222 * gensupport.c (gen_reader::gen_reader): Pass "false"
35223 for new "compact" param of rtx_reader.
35224 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35225 rather than an empty string for NULL strings.
35226 * read-md.c: Potentially include config.h rather than bconfig.h.
35227 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35228 (have_error): New global, copied from errors.c.
35229 (md_reader::read_name): Rename to...
35230 (md_reader::read_name_1): ...this, adding "out_loc" param,
35231 and converting "missing name or number" to returning false, rather
35233 (md_reader::read_name): Reimplement in terms of read_name_1.
35234 (md_reader::read_name_or_nil): New function.
35235 (md_reader::read_string): Handle "(nil)" by returning NULL.
35236 (md_reader::md_reader): Add new param "compact".
35237 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35238 (md_reader::read_file): New method.
35239 * read-md.h (md_reader::md_reader): Add new param "compact".
35240 (md_reader::read_file): New method.
35241 (md_reader::is_compact): New accessor.
35242 (md_reader::read_name): Convert return type from void to file_location.
35243 (md_reader::read_name_or_nil): New decl.
35244 (md_reader::read_name_1): New decl.
35245 (md_reader::m_compact): New field.
35246 (noop_reader::noop_reader): Pass "false" for new "compact" param
35248 (rtx_reader::rtx_reader): Add new "compact" param.
35249 (rtx_reader::read_rtx_operand): Make virtual and convert return
35250 type from void to rtx.
35251 (rtx_reader::read_until): New decl.
35252 (rtx_reader::handle_any_trailing_information): New virtual function.
35253 (rtx_reader::postprocess): New virtual function.
35254 (rtx_reader::finalize_string): New virtual function.
35255 (rtx_reader::m_in_call_function_usage): New field.
35256 (rtx_reader::m_reuse_rtx_by_id): New field.
35257 * read-rtl-function.c: New file.
35258 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35259 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35260 (selftest::verify_three_block_rtl_cfg): New decl.
35261 * read-rtl-function.h: New file.
35262 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35263 For host, include function.h, memmodel.h, and emit-rtl.h.
35264 (one_time_initialization): New function.
35265 (struct compact_insn_name): New struct.
35266 (compact_insn_names): New array.
35267 (find_code): Handle insn codes in compact dumps.
35268 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35269 (bind_subst_iter_and_attr): Likewise.
35270 (add_condition_to_string): Likewise.
35271 (add_condition_to_rtx): Likewise.
35272 (apply_attribute_uses): Likewise.
35273 (add_current_iterators): Likewise.
35274 (apply_iterators): Likewise.
35275 (initialize_iterators): Guard usage of apply_subst_iterator with
35276 #ifdef GENERATOR_FILE.
35277 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35278 (md_reader::read_mapping): Likewise.
35279 (add_define_attr_for_define_subst): Likewise.
35280 (add_define_subst_attr): Likewise.
35281 (read_subst_mapping): Likewise.
35282 (check_code_iterator): Likewise.
35283 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35285 (one_time_initialization): New function.
35286 (rtx_reader::read_until): New method.
35287 (read_flags): New function.
35288 (parse_reg_note_name): New function.
35289 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35290 Handle reuse_rtx ids.
35291 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35292 Add parsing support for RTL dumps, mirroring the special-cases in
35293 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35294 values, and calling handle_any_trailing_information.
35295 (rtx_reader::read_rtx_operand): Convert return type from void
35296 to rtx, returning return_rtx. Handle case 'e'. Call
35297 finalize_string on XSTR and XTMPL fields.
35298 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35299 "(nil)" values were omitted. Call the postprocess vfunc on the
35301 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35302 class ctor. Initialize m_in_call_function_usage. Call
35303 one_time_initialization.
35304 * rtl-tests.c (selftest::test_uncond_jump): Call
35305 set_new_first_and_last_insn.
35306 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35307 * selftest-rtl.c: New file.
35308 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35309 (selftest::get_insn_by_uid): New decl.
35310 * selftest-run-tests.c (selftest::run_tests): Call
35311 read_rtl_function_c_tests.
35312 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35313 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35316 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35318 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35319 operands in a special way. Assert that pos+len <= mode precision.
35321 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35323 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35324 3 argument Alias with unlimited for the negative form.
35325 (fno-vect-cost-model): Removed.
35327 2017-01-05 Martin Liska <mliska@suse.cz>
35329 * hsa-gen.c (gen_hsa_divmod): New function.
35330 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35332 2017-01-05 Martin Liska <mliska@suse.cz>
35335 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35338 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35340 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35341 small constant length operands.
35343 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35345 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35346 between loop iterations.
35348 2017-01-05 Martin Liska <mliska@suse.cz>
35351 * gimplify.c (gimplify_decl_expr): Compare to
35352 asan_poisoned_variables instread of checking flags.
35353 (gimplify_target_expr): Likewise.
35354 (gimplify_expr): Likewise.
35355 (gimplify_function_tree): Conditionally initialize
35356 asan_poisoned_variables.
35358 2017-01-04 Jeff Law <law@redhat.com>
35360 PR tree-optimizatin/78812
35361 * rtl.h (contains_mem_rtx_p): Prototype.
35362 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35363 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35364 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35365 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35366 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35368 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35370 * input.c (assert_char_at_range): Default-initialize actual_range.
35372 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35374 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35375 to match the caller.
35377 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35379 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35380 insns after final jump in test to emit dummy move.
35382 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35384 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35385 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35387 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35389 * multiple_target.c (create_dispatcher_calls): Init e_next.
35390 * tree-ssa-loop-split.c (split_loop): Init border.
35391 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35394 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35399 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35400 (altivec_register_operand): Do not return true if the operand
35401 contains a SUBREG mixing SImode and SFmode.
35402 (vsx_register_operand): Likewise.
35403 (vsx_reg_sfsubreg_ok): New predicate.
35404 (vfloat_operand): Do not return true if the operand contains a
35405 SUBREG mixing SImode and SFmode.
35406 (vint_operand): Likewise.
35407 (vlogical_operand): Likewise.
35408 (gpc_reg_operand): Likewise.
35409 (int_reg_operand): Likewise.
35410 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35411 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35412 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35414 (rs6000_emit_move_si_sf_subreg): New helper function.
35415 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35416 fixup SUBREGs involving SImode and SFmode.
35417 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35418 numbers for the new peephole2 optimization.
35419 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35420 the GLIBC math library that do AND/IOR/XOR operations on single
35421 precision floating point.
35422 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35423 target macros to say whether we need to avoid SUBREGs mixing
35425 (TARGET_ALLOW_SF_SUBREG): Likewise.
35426 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35427 (UNSPEC_SI_FROM_SF): Likewise.
35428 (iorxor): Change spacing.
35429 (and_ior_xor): New iterator for AND, IOR, and XOR.
35430 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35431 (movdi_from_sf_zero_ext): Likewise.
35432 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35433 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35434 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35435 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35436 (fms<mode>4): Likewise.
35437 (fnma<mode>4): Likewise.
35438 (fnms<mode>4): Likewise.
35439 (nfma<mode>4): Likewise.
35440 (nfms<mode>4): Likewise.
35442 2017-01-04 Marek Polacek <polacek@redhat.com>
35445 * doc/invoke.texi: Document -Wpointer-compare.
35447 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35449 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35452 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35453 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35456 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35459 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35460 documentation of the powerpc_popcntb_ok attribute.
35461 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35462 code to issue warning messages if a requested CPU configuration is
35463 not supported by the binary (assembler and loader) toolchain.
35464 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35465 made to define a built-in function that has been disabled.
35466 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35467 made to define a built-in function that has been disabled.
35468 (altivec_init_builtins): Add comment explaining why definition
35469 of the DST built-in functions is not preceded by an assertion
35470 check. Add assertions to prevent ICE if attempts are made to
35471 define an altivec predicate or an abs* built-in function that has
35473 (htm_init_builtins): Add comment explaining why definition of the
35474 htm built-in functions is not preceded by an assertion check.
35476 2017-01-04 Jeff Law <law@redhat.com>
35478 PR tree-optimizatin/67955
35479 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35480 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35481 the points-to solution does not include pt_null. Use DECL_PT_UID
35484 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35486 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35487 Use gen_int_mode instead of gen_lopwart for const_int operands.
35489 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35491 PR tree-optimization/71563
35492 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35493 out of range value - has low bits known to be zero.
35495 2017-01-04 Alan Modra <amodra@gmail.com>
35497 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35498 * configure: Regenerate.
35499 * config.in: Regenerate.
35501 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35504 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35505 a substring of the message, but strcmp with the whole message. Ifdef
35506 ENABLE_NLS, translate the message first using dgettext.
35508 2017-01-03 Jeff Law <law@redhat.com>
35510 PR tree-optimizatin/78856
35511 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35512 (mark_threaded_blocks): Remove code to truncate thread paths that
35513 cross multiple loop headers. Instead invalidate the cached loop
35514 iteration information and handle case of a thread path walking
35515 into an irreducible region.
35517 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35520 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35521 assertions. Add support for doing the signbit if the IEEE 128-bit
35522 floating point value is in a GPR.
35523 * config/rs6000/rs6000.md (Fsignbit): Delete.
35524 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35525 Update the length attribute if the value is in a GPR.
35526 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35527 the sign or zero extension instruction, since the value is always 0/1.
35528 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35531 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35532 extracting SImode to a GPR register so that we can generate a
35533 store, limit the vector to be in a traditional Altivec register
35534 for the vextuwrx instruction.
35536 2017-01-03 Ian Lance Taylor <iant@google.com>
35538 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35540 2017-01-03 Martin Sebor <msebor@redhat.com>
35542 PR tree-optimization/78696
35543 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35544 precision. Use MPFR for %f for greater fidelity. Correct handling
35546 (pass_sprintf_length::compute_format_length): Set width and precision
35547 specified by asrerisk to void_node for vararg functions.
35548 (try_substitute_return_value): Adjust dump output.
35550 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35552 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35554 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35556 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35557 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35558 -mlra/-mno-lra was passed to the compiler.
35560 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35562 PR rtl-optimization/65618
35563 * emit-rtl.c (try_split): Move initialization of "before" and
35564 "after" to just before the call to emit_insn_after_setloc.
35566 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35568 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35570 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35572 * dwarf2out.c (gen_enumeration_type_die): When
35573 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35575 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35577 PR tree-optimization/78965
35578 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35579 Change first argument from const call_info & to call_info &. For %n
35580 set info.nowrite to false.
35582 PR middle-end/78901
35583 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35584 possibly throwing calls.
35586 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35587 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35588 and fns handling, rather than in a separate case SSA_NAME.
35590 2017-01-02 Jeff Law <law@redhat.com>
35592 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35593 fixes for first_period and second_period variables.
35595 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35598 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35599 (*insvqi_1): New insn pattern.
35600 (*insvqi_1_mem_rex64): Ditto.
35601 (*insvqi_2): Ditto.
35602 (*insvqi_3): Rename from *insvqi.
35604 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35606 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35608 * doc/cfg.texi (Edges): Remove reference to Java.
35609 (Maintaining the CFG): Ditto.
35611 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35613 PR middle-end/77674
35614 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35615 transparent aliases.
35617 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35619 PR middle-end/77484
35620 * predict.def (PRED_CALL): Update hitrate.
35621 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35622 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35623 into direct/indirect/polymorphic variants.
35625 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35627 Update copyright years.
35629 * gcc.c (process_command): Update copyright notice dates.
35630 * gcov-dump.c (print_version): Ditto.
35631 * gcov.c (print_version): Ditto.
35632 * gcov-tool.c (print_version): Ditto.
35633 * gengtype.c (create_file): Ditto.
35634 * doc/cpp.texi: Bump @copying's copyright year.
35635 * doc/cppinternals.texi: Ditto.
35636 * doc/gcc.texi: Ditto.
35637 * doc/gccint.texi: Ditto.
35638 * doc/gcov.texi: Ditto.
35639 * doc/install.texi: Ditto.
35640 * doc/invoke.texi: Ditto.
35642 Copyright (C) 2017 Free Software Foundation, Inc.
35644 Copying and distribution of this file, with or without modification,
35645 are permitted in any medium without royalty provided the copyright
35646 notice and this notice are preserved.