1 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6 2017-11-17 Steve Ellcey <sellcey@cavium.com>
9 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
11 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
13 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
15 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
16 rcrt1.o%s/grcrt1.o%s for -static-pie.
18 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
20 * i386.c (ix86_multiplication_cost, ix86_division_cost,
21 ix86_shift_rotate_cost): Break out from ...
22 (ix86_rtx_costs): ... here.
23 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
26 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
28 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
29 when entry block was promoted unlikely.
30 (estimate_bb_frequencies): Increase frequency scale.
31 * profile-count.h (profile_count): Export precision info.
33 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
35 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
36 disturbin profile of entry block.
38 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
40 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
43 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
45 * ipa-cp.c (update_profiling_info): Handle conversion to local
47 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
49 2017-11-17 Jeff Law <law@redhat.com>
51 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
52 from evrp_dom_walker class. Various methods moved into new class.
53 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
54 (evrp_range_analyzer::enter): New method.
55 (evrp_range_analyzer::leave): New method.
56 (evrp_dom_walker): Remove delegators no longer needed by this class.
57 Replace vr_values data member with evrp_range_analyzer
59 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
60 method extracted from evrp_dom_walker::before_dom_children.
61 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
62 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
64 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
65 Add private copy constructor and move assignment operators.
66 Privatize methods and class data where trivially possible.
67 (evrp_dom_walker::cleanup): New function, extracted from
68 execute_early_vrp. Simplify access to class data.
70 * vr-values.h (get_output_for_vrp): Prototype.
71 * vr-values.c (get_output_for_vrp): New function extracted from
72 vrp_visit_assignment_or_call and extract_range_from_stmt.
73 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
75 2017-11-17 Luis Machado <luis.machado@linaro.org>
77 * config/aarch64/aarch64.c
78 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
79 (qdf24xx_tunings) <autoprefetcher_model>: Set to
80 tune_params::AUTOPREFETCHER_WEAK.
82 2017-11-17 Tamar Christina <tamar.christina@arm.com>
85 * config/arm/arm.c (arm_valid_target_attribute_rec):
86 Parse "arch=" and "+<ext>".
87 (arm_valid_target_attribute_tree): Re-init global options.
88 (arm_option_override): Make non-static.
89 (arm_options_perform_arch_sanity_checks): Make errors fatal.
90 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
91 (__ARM_FEATURE_CRC32): Support undef.
92 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
93 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
95 2017-11-17 David Malcolm <dmalcolm@redhat.com>
97 * gdbinit.in (break-on-diagnostic): New command.
99 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
101 * config/i386/i386.c (ix86_expand_epilogue): Change simple
102 return to indirect jump for EH return if control-flow
103 protection is enabled. Change explicit 'false' argument in
104 pro_epilogue_adjust_stack with a value of
106 * config/i386/i386.md (simple_return_indirect_internal):
107 Remove SImode restriction to support 64-bit.
109 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
111 * combine.c (added_notes_insn): New.
112 (try_combine): Handle added_notes_insn like added_links_insn.
113 Rewrite return value code.
114 (distribute_notes): Set added_notes_insn to the earliest insn we added
117 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
119 PR rtl-optimization/82621
120 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
121 dest of one of those SETs is unused.
123 2017-11-17 Richard Biener <rguenther@suse.de>
126 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
127 * tree-pretty-print.c (dump_generic_node): Handle
128 annot_expr_parallel_kind.
129 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
130 * gimplify.c (gimple_boolify): Likewise.
132 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
134 * config.gcc (extra_headers): Add cet.h for x86 targets.
135 * config/i386/cet.h: New file.
136 * doc/install.texi: Add --enable-cet/--disable-cet.
138 2017-11-17 Richard Biener <rguenther@suse.de>
140 PR tree-optimization/83017
141 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
142 (gen_parallel_loop): Properly count iterations.
143 (parallelize_loops): Handle loop->can_be_parallel independent
144 of flag_loop_parallelize_all. Make static profitability test match
146 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
147 * invoke.texi (parloops-min-per-thread): Document.
149 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
151 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
152 upstreaming review comments.
154 2017-11-17 Tamar Christina <tamar.christina@arm.com>
156 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
157 with fast unaligned access.
158 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
160 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
162 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
163 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
165 2017-11-17 Richard Biener <rguenther@suse.de>
167 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
168 folding of references.
170 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
173 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
174 of replacing call to strncmp with corresponding call to strcmp when
177 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
179 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
180 option prefer-avx256 for skylake-avx512 configuration.
181 * config/i386/i386.c (ix86_option_override_internal): Ditto.
182 (get_builtin_code_for_version): Ditto.
184 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
185 Monk Chiang <sh.chiang04@gmail.com>
187 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
188 (FIXED_REGISTERS): Reserve more register numbers.
189 (CALL_USED_REGISTERS): Likewise.
190 (REG_ALLOC_ORDER): Likewise.
191 (REG_CLASS_CONTENTS): Likewise.
192 (REGISTER_NAMES): Likewise.
194 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
195 Kito Cheng <kito.cheng@gmail.com>
197 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
199 * config/nds32/iterators.md: Add vector mode iterators and attributes.
201 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
203 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
204 parameter list for vec_splats.
206 2017-11-16 Joseph Myers <joseph@codesourcery.com>
208 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
210 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
213 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
216 * doc/install.texi: Document incompatibility of Perl >=5.6.26
217 with the required version of automake 1.11.6.
219 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
221 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
223 (DU_C2_3_power9): Correct reservation combinations.
224 (FP_DIV_power9, VEC_DIV_power9): New.
225 (power9-alu): Split out rotate/shift...
226 (power9-rot): ...to here, correct dispatch resource.
227 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
229 (power9-fp): Correct latency.
230 (power9-sdiv): Add div/sqrt resource.
231 (power9-ddiv): Correct latency, add div/sqrt resource.
232 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
233 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
235 (power9-qpdiv, power9-qpmul): Adjust resource usage.
237 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
239 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
240 switch statement mapping KF built-ins to TF built-ins if we don't
241 have the proper ISA 3.0 assembler support.
243 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
245 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
246 (lower_emutls_function_body): Do not compute it.
248 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
250 * ipa-split.c (split_bb_info): Turn time to sreal.
251 (split_point): Likewise.
252 (dump_split_point): Likewise.
253 (fine_split_points): Likewise.
254 (execute_split_functions): Only zero split_bbs; turn time to sreals.
256 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
258 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
261 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
263 * predict.c (combine_predictions_for_bb): Preserve zero predicted
265 (expensive_function_p): Remove useless assert.
266 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
268 2017-11-16 Martin Sebor <msebor@redhat.com>
270 PR tree-optimization/82588
271 PR tree-optimization/82583
272 * tree-vrp.c (check_array_ref): Handle flexible array members,
273 string literals, and inner indices.
274 (search_for_addr_array): Add detail to diagnostics.
276 2017-11-16 Nathan Sidwell <nathan@acm.org>
280 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
281 (SET_DECL_ASSEMBLER_NAME): Forward to
282 overwrite_decl_assembler_name.
283 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
284 (overwrite_decl_assembler_name): Declare.
285 * tree.c (overwrite_decl_assembler_name): New.
286 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
287 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
288 (LANG_HOOKS_INITIALIZER): Add it.
289 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
290 * langhooks.c (lhd_set_decl_assembler_name): Use
291 SET_DECL_ASSEMBLER_NAME.
292 (lhd_overwrite_decl_assembler_name): Default implementation.
294 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
295 Jackson Woodruff <jackson.woodruff@arm.com>
297 PR tree-optimization/71026
298 * match.pd: Canonicalize constant multiplies in division.
300 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
302 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
303 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
304 Remove OPT_fomit_frame_pointer entry.
305 * common/config/arc/arc-common.c: Likewise.
306 * common/config/arm/arm-common.c: Likewise.
307 * common/config/avr/avr-common.c: Likewise.
308 * common/config/c6x/c6x-common.c: Likewise.
309 * common/config/cr16/cr16-common.c: Likewise.
310 * common/config/cris/cris-common.c: Likewise.
311 * common/config/epiphany/epiphany-common.c: Likewise.
312 * common/config/fr30/fr30-common.c: Likewise.
313 * common/config/frv/frv-common.c: Likewise.
314 * common/config/ia64/ia64-common.c: Likewise.
315 * common/config/iq2000/iq2000-common.c: Likewise.
316 * common/config/lm32/lm32-common.c: Likewise.
317 * common/config/m32r/m32r-common.c: Likewise.
318 * common/config/mcore/mcore-common.c: Likewise.
319 * common/config/microblaze/microblaze-common.c: Likewise.
320 * common/config/mips/mips-common.c: Likewise.
321 * common/config/mmix/mmix-common.c: Likewise.
322 * common/config/mn10300/mn10300-common.c: Likewise.
323 * common/config/nios2/nios2-common.c: Likewise.
324 * common/config/pa/pa-common.c: Likewise.
325 * common/config/pdp11/pdp11-common.c: Likewise.
326 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
327 * common/config/riscv/riscv-common.c: Likewise.
328 * common/config/rs6000/rs6000-common.c: Likewise.
329 * common/config/rx/rx-common.c: Likewise.
330 * common/config/s390/s390-common.c: Likewise.
331 * common/config/sh/sh-common.c: Likewise.
332 * common/config/sparc/sparc-common.c: Likewise.
333 * common/config/tilegx/tilegx-common.c: Likewise.
334 * common/config/tilepro/tilepro-common.c: Likewise.
335 * common/config/v850/v850-common.c: Likewise.
336 * common/config/visium/visium-common.c: Likewise.
337 * common/config/xstormy16/xstormy16-common.c: Likewise.
338 * common/config/xtensa/xtensa-common.c: Likewise.
339 * invoke.texi (-fomit-frame-pointer): Update documentation.
341 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
343 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
345 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
347 * cfg.c (scale_bbs_frequencies_int,
348 cale_bbs_frequencies_gcov_type): Remove.
349 * cfg.h (scale_bbs_frequencies_int,
350 cale_bbs_frequencies_gcov_type): Remove.
352 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
354 * tree-ssa-loop-manip.c
355 (scale_dominated_blocks_in_loop): Update to profile counts.
356 (tree_transform_and_unroll_loop): Likewise.
358 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
360 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
361 scale_bbs_frequencies_int.
363 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
365 * final.c (compute_alignments): Use counts rather than frequencies.
367 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
369 * cfgloopanal.c: Include sreal.h
370 (average_num_loop_insns): Use counts and sreal for accounting.
372 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
374 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
377 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
378 Kito Cheng <kito.cheng@gmail.com>
380 * config/nds32/constraints.md: Provide more constraints.
381 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
382 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
383 support constraints usage.
385 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
387 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
389 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
390 Kito Cheng <kito.cheng@gmail.com>
392 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
393 * config/nds32/nds32.opt: Refine the layout.
394 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
395 TARGET_EXT_STRING): Support new options.
396 * config/nds32/nds32.h: Likewise.
397 * config/nds32/nds32.md: Likewise.
398 * config/nds32/nds32-predicates.c: Likewise.
399 * config/nds32/constraints.md: Likewise.
400 * common/config/nds32/nds32-common.c: Likewise.
402 2017-11-16 Julia Koval <julia.koval@intel.com>
405 * config/i386/gfniintrin.h: Add sse check.
406 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
408 2017-11-16 Julia Koval <julia.koval@intel.com>
410 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
411 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
412 (ix86_handle_option): Handle -mavx512vbmi2.
413 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
414 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
415 * config/i386/i386-c.c (__AVX512VBMI2__): New.
416 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
417 (ix86_valid_target_attribute_inner_p): Ditto.
418 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
419 * config/i386/i386.opt (mavx512vbmi2): New option.
420 * doc/invoke.texi: Add new option.
422 [2017-11-16 Julia Koval <julia.koval@intel.com>
424 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
425 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
426 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
427 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
428 _mm512_gf2p8mul_epi8): New intrinsics.
429 * config/i386/i386-builtin-types.def
430 (V64QI_FTYPE_V64QI_V64QI): New type.
431 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
432 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
433 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
434 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
435 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
436 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
438 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
440 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
441 explicitly as a stream of bytes.
443 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
445 * config/rs6000/altivec.h (vec_xst_be): New #define.
446 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
447 and externalize from *altivec_vperm_<mode>_internal.
448 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
450 (XL_BE_V8HI): Likewise.
451 (XL_BE_V4SI): Likewise.
452 (XL_BE_V4SI): Likewise.
453 (XL_BE_V2DI): Likewise.
454 (XL_BE_V4SF): Likewise.
455 (XL_BE_V2DF): Likewise.
456 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
457 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
458 all array entries with these keys: VSX_BUILTIN_VEC_XL,
459 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
460 VSX_BUILTIN_VEC_XST_BE.
461 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
462 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
464 (altivec_init_builtins): Replace conditional calls to def_builtin
465 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
466 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
467 with unconditional calls. Remove calls to def_builtin for
468 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
469 __builtin_vec_xst_be.
470 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
471 to define_expand, and add alternate RTL generation for P8.
472 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
474 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
475 add alternate RTL generation for P8.
476 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
477 vsx_ld_elemrev_v16qi.
478 (vsx_st_elemrev_v8hi): Convert define_insn
479 to define_expand, and add alternate RTL generation for P8.
480 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
482 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
483 add alternate RTL generation for P8.
484 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
485 vsx_st_elemrev_v16qi.
487 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
490 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
491 TARGET_AVX512ER check.
492 (ix86_option_override_internal): Set MASK_VZEROUPPER if
493 neither -mzeroupper nor -mno-zeroupper is used and
494 TARGET_EMIT_VZEROUPPER is set.
495 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
496 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
498 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
500 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
501 folding of vector compares.
502 (fold_build_vec_cmp): New helper function.
503 (fold_compare_helper): New helper function.
504 (builtin_function_type): Add compare builtins to the list of functions
505 having unsigned arguments. Cosmetic updates to comment indentation.
506 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
507 the not+eq combination.
509 2017-11-15 Bin Cheng <bin.cheng@arm.com>
511 PR tree-optimization/82726
512 PR tree-optimization/70754
513 * tree-predcom.c (order_drefs_by_pos): New function.
514 (combine_chains): Move code setting has_max_use_after to...
515 (try_combine_chains): ...here. New parameter. Sort combined chains
516 according to position information.
517 (tree_predictive_commoning_loop): Update call to above function.
518 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
520 2017-11-15 Bin Cheng <bin.cheng@arm.com>
522 PR tree-optimization/82726
524 2017-01-23 Bin Cheng <bin.cheng@arm.com>
526 PR tree-optimization/70754
527 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
528 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
529 combined stmt before it if not NULL.
530 (combine_chains): Process refs reversely and compute dominance point
534 2017-02-23 Bin Cheng <bin.cheng@arm.com>
536 PR tree-optimization/79663
537 * tree-predcom.c (combine_chains): Process refs in reverse order
538 only for ZERO length chains, and add explaining comment.
540 2017-11-15 Tamar Christina <tamar.christina@arm.com>
542 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
543 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
544 * doc/invoke.texi (ARM Options): Add armv8.3-a.
546 2017-11-15 Tamar Christina <tamar.christina@arm.com>
548 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
550 2017-11-15 Martin Liska <mliska@suse.cz>
552 * tree-cfg.c (pass_warn_function_return::execute):
553 Compare warn_return_type for greater than zero.
555 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
559 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
560 to return true on Xeon and not on Xeon Phi.
561 (ix86_check_avx256_register): Changed to ...
562 (ix86_check_avx_upper_register): ... this. Add extra check for
563 VALID_AVX512F_REG_OR_XI_MODE.
564 (ix86_avx_u128_mode_needed): Changed
565 ix86_check_avx256_register to ix86_check_avx_upper_register.
566 (ix86_check_avx256_stores): Changed to ...
567 (ix86_check_avx_upper_stores): ... this. Changed
568 ix86_check_avx256_register to ix86_check_avx_upper_register.
569 (ix86_avx_u128_mode_after): Changed
570 avx_reg256_found to avx_upper_reg_found. Changed
571 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
572 (ix86_avx_u128_mode_entry): Changed
573 ix86_check_avx256_register to ix86_check_avx_upper_register.
574 (ix86_avx_u128_mode_exit): Ditto.
575 * config/i386/i386.h: (host_detect_local_cpu): New define.
577 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
579 * config/arm/xgene1.md (xgene1): Split into automatons
580 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
581 (xgene1_f_load): Adjust reservations and/or types.
582 (xgene1_f_store): Likewise.
583 (xgene1_load_pair): Likewise.
584 (xgene1_store_pair): Likewise.
585 (xgene1_fp_load1): Likewise.
586 (xgene1_load1): Likewise.
587 (xgene1_store1): Likewise.
588 (xgene1_move): Likewise.
589 (xgene1_alu): Likewise.
590 (xgene1_simd): Likewise.
591 (xgene1_bfm): Likewise.
592 (xgene1_neon_load1): Likewise.
593 (xgene1_neon_store1): Likewise.
594 (xgene1_neon_logic): Likewise.
595 (xgene1_neon_st1): Likewise.
596 (xgene1_neon_ld1r): Likewise.
597 (xgene1_alu_cond): Added.
598 (xgene1_shift_reg): Likwise.
599 (xgene1_bfx): Likewise.
600 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
602 2017-11-15 Jakub Jelinek <jakub@redhat.com>
605 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
607 (can_widen_mult_without_libcall): New function.
608 (expand_mul_overflow): If only checking unsigned mul overflow,
609 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
610 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
611 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
612 (expand_DIVMOD): Formatting fix.
613 * expmed.h (expand_mult): Add NO_LIBCALL argument.
614 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
615 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
617 PR tree-optimization/82977
618 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
619 constructed temporary to strlen_to_stridx.put.
621 2017-11-15 Martin Liska <mliska@suse.cz>
623 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
624 * configure: Regenerate.
626 2017-11-15 Martin Liska <mliska@suse.cz>
629 * config/sh/sh-mem.cc: Use proper probability for
632 2017-11-14 Jeff Law <law@redhat.com>
634 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
635 the red zone for stack_clash_protection_final_dynamic_probe targets
636 when the total dynamic stack size is zero bytes.
638 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
639 blocks is post order.
641 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
643 * dumpfile.h (TDF_COMPARE_DEBUG): New.
644 * final.c (rest_of_clean_state): Set it for the
645 -fcompare-debug dump.
646 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
647 class when TDF_COMPARE_DEBUG is set.
649 * dwarf2out.c (gen_producer_string): Discard
652 2017-11-15 Joseph Myers <joseph@codesourcery.com>
655 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
656 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
657 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
658 (__TGMATH_REAL_2_3): Remove macros.
659 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
660 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
662 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
664 (remquo): Define using __TGMATH_REAL_3.
666 2017-11-14 Jeff Law <law@redhat.com>
668 * vr-values.c: New file with contents extracted from tree-vrp.c.
669 * Makefile.in (OBJS): Add vr-values.o
670 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
671 (set_value_range, set_and_canonicalize_value_range): Likewise.
672 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
673 (value_range_constant_singleton, symbolic_range_p): Likewise.
674 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
675 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
676 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
677 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
678 (find_case_label_range, find_case_label_index): Likewise.
679 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
680 (range_int_cst_singleton_p, value_inside_range): Likewise.
681 (get_single_symbol): Likewise.
682 (switch_update): Move structure definition here.
683 (to_remove_edges, to_update_switch_stmts): Provide externs.
684 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
685 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
686 (vrp_val_is_min, set_value_range): Likewise.
687 (set_and_canonicalize_value_range, copy_value_range): Likewise.
688 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
689 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
690 (range_is_nonnull, range_int_cst_p): Likewwise.
691 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
692 (get_single_symbol, operand_less_p): Likewise
693 (compare_values_warnv, compare_values): Likewise.
694 (value_inside_range, value_range_constant_singleton): Likewise.
695 (zero_nonzero_bitgs_from_vr): Likewise.
696 (extract_range_from_binary_expr_1): Likewise.
697 (overflow_comparison_p): Likewise.
698 (to_remove_edges, to_update_switch_stmts): Likewise.
699 (find_case_label-index, find_case_label_range): Likewise.
700 (switch_update, set_value_range_to_nonnegative): Remove.
701 (set_value_range_to_truthvalue): Likewise.
702 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
703 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
704 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
705 (find_case_label_ranges, test_for_singularity): Likewise.
706 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
707 (x_vr_values): Move to its remaining use site.
709 2017-11-10 Jeff Law <law@redhat.com>
711 * vr-values.h (VR_INITIALIZER): Move #define here.
712 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
713 * Makefile.in (OBJS): Add tree-evrp.o
714 * tree-vrp.h (assert_info): Move structure definition here.
715 (set_value_range_to_varying): Prototype.
716 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
717 (infer_value_range, register_edge_assert_for): Likewise.
718 (stmt_interesting_for_vrp): Likewise.
719 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
720 (set_value_range_to_varying): No longer static.
721 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
722 (infer_value_range, register_edge_assert_for): Likewise.
724 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
726 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
727 generate the XXBRD instruction.
729 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
730 (rs6000_builtin_type_compatible): Treat _Float128 and long double
731 as being compatible if -mabi=ieeelongdouble.
732 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
733 to setup float128 built-ins with hardware support.
734 (BU_FLOAT128_HW_2): Likewise.
735 (BU_FLOAT128_HW_3): Likewise.
736 (BU_FLOAT128_HW_VSX_1): Likewise.
737 (BU_FLOAT128_HW_VSX_2): Likewise.
738 (scalar_extract_expq): Change float128 built-in functions to
739 accommodate having both KFmode and TFmode functions. Use the
740 KFmode variant as the default.
741 (scalar_extract_sigq): Likewise.
742 (scalar_test_neg_qp): Likewise.
743 (scalar_insert_exp_q): Likewise.
744 (scalar_insert_exp_qp): Likewise.
745 (scalar_test_data_class_qp): Likewise.
746 (sqrtf128_round_to_odd): Delete processing the round to odd
747 built-in functions as special built-in functions, and define them
748 as float128 built-ins. Use the KFmode variant as the default.
749 (truncf128_round_to_odd): Likewise.
750 (addf128_round_to_odd): Likewise.
751 (subf128_round_to_odd): Likewise.
752 (mulf128_round_to_odd): Likewise.
753 (divf128_round_to_odd): Likewise.
754 (fmaf128_round_to_odd): Likewise.
755 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
756 support for KFmode and TFmode xststdcqp calls.
757 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
758 point, switch the built-in handlers for the get/set float128
759 exponent, get float128 mantissa, float128 test built-ins, and the
760 float128 round to odd built-in functions. Eliminate creating the
761 float128 round to odd built-in functions as special built-ins.
762 (rs6000_init_builtins): Eliminate special creation of the float128
763 round to odd built-in functions.
764 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
765 function insns to support both TFmode and KFmode variants.
766 (xsxsigqp_<mode>): Likewise.
767 (xsiexpqpf_<mode>): Likewise.
768 (xsiexpqp_<mode>): Likewise.
769 (xststdcqp_<mode>): Likewise.
770 (xststdcnegqp_<mode>): Likewise.
771 (xststdcqp_<mode>): Likewise.
773 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
775 * tree-ssa-threadupdate.c (compute_path_counts): Remove
776 unused path_in_freq_ptr parameter.
777 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
779 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
781 * ipa-inline.c (edge_badness): Dump sreal frequency.
782 (compute_inlined_call_time): Match natural implementaiton ...
783 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
784 forgotten division by CGRAPH_FREQ_BASE.
786 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
788 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
789 Solaris 11. Update comment.
790 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
792 * config/sol2.h (STARTFILE_SPEC): Likewise.
793 * configure: Regenerate.
795 2017-11-14 Carl Love <cel@us.ibm.com>
797 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
798 le_ and be_ prefixes to swap* variables. Remove
799 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
802 2017-11-14 Jason Merrill <jason@redhat.com>
804 Support GTY((cache)) on hash_map.
805 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
806 (ggc_cache_remove): Override it instead of ggc_mx.
807 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
808 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
809 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
810 (simple_cache_map_traits): Override maybe_mx.
811 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
812 (hash_map): Friend gt_cleare_cache.
813 (gt_cleare_cache): New.
814 * tree.h (tree_cache_traits): New hash_map traits class.
815 (tree_cache_map): New typedef.
817 2017-11-14 Richard Biener <rguenther@suse.de>
819 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
820 paramter and handling.
821 (cleanup_control_flow_bb): Likewise.
822 (cleanup_control_flow_pre): New helper performing a DFS walk
823 to call cleanup_control_flow_bb in PRE order.
824 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
825 via cleanup_control_flow_pre.
827 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
829 * config/aarch64/aarch64-simd.md
830 (aarch64_simd_bsl<mode>_internal): Remove DImode.
831 (*aarch64_simd_bsl<mode>_alt): Likewise.
832 (aarch64_simd_bsldi_internal): New.
833 (aarch64_simd_bsldi_alt): Likewise.
835 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
837 * tracer.c (better_p): Do not compare frequencies.
838 * reg-stack.c (better_edge): Likewise.
839 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
842 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
844 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
845 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
846 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
847 * lto-streamer-in.c (input_function): Use update_max_bb_count.
848 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
849 * predict.c (maybe_hot_frequency_p): Inline to ...
850 (maybe_hot_count_p): ... here; rewrite to counts.
851 (counts_to_freqs): Rename to ...
852 (update_max_bb_count): ... this one.
853 (expensive_function_p): Use counts.
854 (estimate_bb_frequencies): Update.
855 (rebuild_frequencies): Update.
856 * predict.h (counts_to_freqs): Rename to ...
857 (update_max_bb_count): ... this one.
858 * profile.c (compute_branch_probabilities): Add debug info
859 * tree-inline.c (expand_call_inline): Update debug info.
860 (optimize_inline_calls): Use update_max_bb_count..
861 (tree_function_versioning): Use update_max_bb_count..
862 * value-prof.c (gimple_value_profile_transformations):
863 Do not use update_max_bb_count.
865 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
867 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
868 always use frequencies.
870 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
872 * bb-reorder.c: Remove frequencies from comments.
873 (better_edge_p): Use profile counts.
874 (find_traces): Dump profile counts.
875 (rotate_loop): Use profile counts.
876 (find_traces_1_round): Likewise.
877 (connect_better_edge_p): Use counts instead of probabilities for
879 (copy_bb_p): Drop early check for non-0 frequency.
880 (sanitize_hot_paths): Update comments.
882 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
884 * ipa-split.c (struct split_point): Add count.
885 (consider_split): Do not compute incoming frequency; compute incoming
886 count and store it to split_point.
887 (split_function): Set count of the call to split part correctly.
889 2017-11-13 Carl Love <cel@us.ibm.com>
891 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
893 2017-11-13 Tom Tromey <tom@tromey.com>
895 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
897 2017-11-13 Carl Love <cel@us.ibm.com>
899 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
900 Add support for builtins:
901 unsigned int vec_first_{,miss}_match_{,or_eos}index,
902 vector {un,}signed {char,int,short},
903 vector {un,}signed {char,int,short}) arguments.
904 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
905 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
906 Add BU_P9V_AV_2 expansions for the builtins.
907 * config/rs6000/altivec.h (vec_first_match_index,
908 vec_first_mismatch_index, vec_first_match_or_eos_index,
909 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
910 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
911 new extern declaration.
912 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
913 * config/rs6000/vsx.md (first_match_index_<mode>,
914 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
915 first_mismatch_or_eos_index_<mode>): Add define expand.
916 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
917 * doc/extend.texi: Update the built-in documenation file for the new
920 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
922 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
923 into the min/max operations for _Float<N> and _Float<N>X types.
925 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
928 * dwarf2out.c (do_eh_frame): New static variable.
929 (dwarf2out_begin_prologue): Set it.
930 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
932 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
934 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
936 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
938 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
941 * debug.h (dwarf2out_do_eh_frame): Declare.
942 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
943 (dwarf2out_do_frame): Use it.
944 (dwarf2out_do_cfi_asm): Likewise.
945 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
946 (dwarf2out_assembly_start): Likewise.
947 (dwarf2out_begin_prologue): Fix comment.
948 * toplev.c (compile_file): Always call dwarf2out_frame_finish
949 if the target needs either debug or unwind DWARF2 info.
950 * lto-opts.c (lto_write_options): Do not save -fexceptions,
951 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
952 -ftrapping-math, -ftrapv and -fwrapv.
954 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
956 * cgraph.c (cgraph_edge::sreal_frequency): New function.
957 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
958 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
959 (estimate_edge_size_and_time): Likewise.
960 (ipa_merge_fn_summary_after_inlining): Likewise.
961 * ipa-inline.c (cgraph_freq_base_rec): Remove.
962 (compute_uninlined_call_time): Use sreal_frequency.
963 (compute_inlined_call_time): Likewise.
964 (ipa_inline): Do not initialize cgraph_freq_base_rec.
965 * profile-count.c: Include sreal.h.
966 (profile_count::to_sreal_scale): New.
967 * profile-count.h: Forward declare sreal.
968 (profile_count::to_sreal_scale): Declare.
970 2017-11-13 Nathan Sidwell <nathan@acm.org>
972 * diagnostic.c (maybe_line_and_column): New.
973 (diagnostic_get_location_text): Use it.
974 (diagnostic_report_current_module): Likewise.
975 (test_diagnostic_get_location_text): Add tests.
977 2017-11-13 Luis Machado <luis.machado@linaro.org>
979 * doc/md.texi (Specifying processor pipeline description): Fix
980 incorrect latency for the div instruction example.
982 2017-11-13 Jakub Jelinek <jakub@redhat.com>
984 PR tree-optimization/78821
985 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
986 that bit_not_p is the same.
987 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
988 (split_group): Count precisely bit_not_p bits in each statement.
989 (invert_op): New function.
990 (imm_store_chain_info::output_merged_store): Use invert_op to
991 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
992 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
994 2017-11-13 Martin Liska <mliska@suse.cz>
996 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
997 (struct source_info): Likewise.
998 (add_branch_counts): Likewise.
999 (add_line_counts): Likewise.
1000 (function_summary): Likewise.
1001 (output_intermediate_line): Likewise.
1002 (generate_results): Likewise.
1004 2017-11-13 Martin Liska <mliska@suse.cz>
1006 * gcov.c (struct block_info): Remove typedef for block_t.
1007 (struct line_info): Likewise.
1008 (line_info::has_block): Likewise.
1009 (EXIT_BLOCK): Likewise.
1010 (unblock): Likewise.
1011 (circuit): Likewise.
1012 (get_cycles_count): Likewise.
1013 (process_file): Likewise.
1014 (read_graph_file): Likewise.
1015 (solve_flow_graph): Likewise.
1016 (find_exception_blocks): Likewise.
1017 (add_line_counts): Likewise.
1018 (accumulate_line_info): Likewise.
1019 (output_line_details): Likewise.
1021 2017-11-13 Martin Liska <mliska@suse.cz>
1023 * gcov.c (struct arc_info): Remove typedef for arc_t.
1024 (struct line_info): Likewise.
1025 (add_branch_counts): Likewise.
1026 (output_branch_count): Likewise.
1027 (function_info::~function_info): Likewise.
1028 (circuit): Likewise.
1029 (output_intermediate_line): Likewise.
1030 (read_graph_file): Likewise.
1031 (solve_flow_graph): Likewise.
1032 (find_exception_blocks): Likewise.
1033 (add_line_counts): Likewise.
1034 (accumulate_line_info): Likewise.
1035 (output_line_details): Likewise.
1036 (output_function_details): Likewise.
1038 2017-11-13 Martin Liska <mliska@suse.cz>
1040 * gcov.c (struct function_info): Remove typedef for function_t.
1041 (struct source_info): Likewise.
1042 (source_info::get_functions_at_location): Likewise.
1043 (solve_flow_graph): Likewise.
1044 (find_exception_blocks): Likewise.
1045 (add_line_counts): Likewise.
1046 (output_intermediate_file): Likewise.
1047 (process_file): Likewise.
1048 (generate_results): Likewise.
1049 (release_structures): Likewise.
1050 (read_graph_file): Likewise.
1051 (read_count_file): Likewise.
1052 (accumulate_line_counts): Likewise.
1053 (output_lines): Likewise.
1055 2017-11-13 Martin Liska <mliska@suse.cz>
1057 * gcov.c (function_info::function_info): Remove num_counts
1058 and add vector<gcov_type>.
1059 (function_info::~function_info): Use the vector.
1060 (process_file): Likewise.
1061 (read_graph_file): Likewise.
1062 (read_count_file): Likewise.
1063 (solve_flow_graph): Likewise.
1065 2017-11-13 Martin Liska <mliska@suse.cz>
1067 * gcov.c (function_info::is_artificial): New function.
1068 (process_file): Erase all artificial early.
1069 (generate_results): Skip as all artificial are already
1072 2017-11-13 Martin Liska <mliska@suse.cz>
1074 * gcov.c (read_graph_file): Store to global vector of functions.
1075 (read_count_file): Iterate the vector.
1076 (process_file): Likewise.
1077 (generate_results): Likewise.
1078 (release_structures): Likewise.
1080 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1082 PR tree-optimization/82954
1083 * gimple-ssa-store-merging.c
1084 (imm_store_chain_info::coalesce_immediate_stores): If
1085 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1087 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1089 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1090 Upddate call to ENDIAN_LANE_N.
1091 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1092 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1093 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1094 and use aarch64_endian_lane_rtx.
1095 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1097 2017-11-12 Tom de Vries <tom@codesourcery.com>
1099 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1101 2017-11-12 Tom de Vries <tom@codesourcery.com>
1103 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1105 2017-11-12 Tom de Vries <tom@codesourcery.com>
1107 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1109 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1110 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1111 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1113 2017-11-11 Martin Sebor <msebor@redhat.com>
1116 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1119 * prefic.c (translate_name): Replace strncpy with memcpy to
1120 avoid -Wstringop-truncation.
1122 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1124 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1126 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1128 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1130 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1132 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1135 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1137 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1138 enable generating XXBRH if the value is in a vector register.
1139 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1140 value is in a vector register.
1141 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1142 register to register bswap64's instead of doing the GPR sequence
1143 used on previous machines.
1144 (bswapdi2_xxbrd): New insn.
1145 (bswapdi2_reg): Disallow on ISA 3.0.
1146 (register to register bswap64 splitter): Do not split the insn on
1147 ISA 3.0 systems that use XXBRD.
1149 2017-11-10 Martin Sebor <msebor@redhat.com>
1152 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1154 (find_subframework_file): Same.
1156 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1158 * auto-profile.c (afdo_indirect_call): Drop frequency.
1159 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1160 (cgraph_node::create_edge): Drop frequency argument.
1161 (cgraph_node::create_indirect_edge): Drop frequency argument.
1162 (cgraph_edge::make_speculative): Drop frequency arguments.
1163 (cgraph_edge::resolve_speculation): Do not update frequencies
1164 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1165 (cgraph_node::dump): Check consistency in IPA mode.
1166 (cgraph_edge::maybe_hot_p): Use IPA counter.
1167 (cgraph_edge::verify_count_and_frequency): Rename to ...
1168 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1169 (cgraph_node::verify_node): Update.
1170 * cgraph.h (struct cgraph_edge): Drop frequency.
1171 (cgraph_edge::frequency): New function.
1172 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1174 (cgraph_edge::rebuild_edges): Likewise.
1175 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1176 (duplicate_thunk_for_node): Do not pass frequency.
1177 (cgraph_node::create_clone): Scale only counts.
1178 (cgraph_node::create_virtual_clone): Do not pass frequency.
1179 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1180 (cgraph_node::create_version_clone): Do not pass frequency.
1181 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1182 (cgraph_node::expand_thunk): Do not pass frequency.
1183 (cgraph_node::create_wrapper): Do not pass frequency.
1184 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1186 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1187 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1188 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1189 (gather_caller_stats): Use frequency function.
1190 (ipcp_cloning_candidate_p): Use frequency function.
1191 (ipcp_propagate_stage): Use frequency function.
1192 (get_info_about_necessary_edges): Use frequency function.
1193 (update_profiling_info): Update only IPA profile.
1194 (update_specialized_profile): Use frequency functoin.
1195 (perhaps_add_new_callers): Update only IPA profile.
1196 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1197 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1198 (dump_ipa_call_summary): Use frequency function.
1199 (estimate_edge_size_and_time): Use frequency function.
1200 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1201 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1202 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1203 (update_noncloned_counts): ... ths one; scale counts only.
1204 (clone_inlined_nodes): Do not scale frequency.
1205 (inline_call): Do not pass frequency.
1206 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1207 (compute_inlined_call_time): Use IPA profile.
1208 (want_inline_small_function_p): Use IPA profile.
1209 (want_inline_self_recursive_call_p): Use IPA profile.
1210 (edge_badness): Use IPA profile.
1211 (lookup_recursive_calls): Use IPA profile.
1212 (recursive_inlining): Do not pass frequency.
1213 (resolve_noninline_speculation): Do not update frequency.
1214 (inline_small_functions): Collect max of IPA profile.
1215 (dump_overall_stats): Dump IPA porfile.
1216 (dump_inline_stats): Dump IPA porfile.
1217 (ipa_inline): Collect IPA stats.
1218 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1219 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1220 (ipa_propagate_frequency): Use frequency function.
1221 (ipa_profile): Cleanup.
1222 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1223 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1224 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1225 (input_node): Do not stream frequency.
1226 (input_edge): Do not stream frequency.
1227 (merge_profile_summaries): Scale only IPA profiles.
1228 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1229 * predict.c (drop_profile): Do not recompute frequency.
1230 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1231 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1232 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1233 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1234 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1235 * tree-inline.c (copy_bb): Do not scale frequency.
1236 (expand_call_inline): Do not scale frequency.
1237 (tree_function_versioning): Do not scale frequency.
1238 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1240 2017-11-10 Julia Koval <julia.koval@intel.com>
1242 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1243 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1244 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1245 (_mm256_maskz_gf2p8affine_epi64_epi8)
1246 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1247 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1248 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1249 (__builtin_ia32_vgf2p8affineqb_v32qi)
1250 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1251 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1253 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1257 (arm_option_override): Refactor.
1258 (arm_option_reconfigure_globals): New.
1259 (arm_options_perform_arch_sanity_checks): New.
1260 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1262 (arm_options_perform_arch_sanity_checks): Likewise
1264 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1266 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1267 (power9-qpmul): New.
1268 * rs6000/rs6000.md ("type" attr): Add qmul.
1269 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1270 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1271 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1273 2017-11-10 Martin Sebor <msebor@redhat.com>
1276 * builtins.c (compute_objsize): Handle arrays that
1277 compute_builtin_object_size likes to fail for. Make extern.
1278 * builtins.h (compute_objsize): Declare.
1279 (check_strncpy_sizes): New function.
1280 (expand_builtin_strncpy): Call check_strncpy_sizes.
1281 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1282 -Wstringop-truncation.
1283 (gimple_fold_builtin_strncat): Same.
1284 * gimple.c (gimple_build_call_from_tree): Set call location.
1285 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1286 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1287 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1288 (handle_builtin_strlen): Use strlen_to_stridx.
1289 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1291 Use strlen_to_stridx.
1292 (pass_strlen::execute): Release strlen_to_stridx.
1293 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1294 (-Wstringop-truncation): Document new option.
1296 2017-11-10 Martin Liska <mliska@suse.cz>
1298 PR gcov-profile/82702
1299 * gcov.c (main): Handle intermediate files in a different
1301 (get_gcov_intermediate_filename): New function.
1302 (output_gcov_file): Remove support of intermediate files.
1303 (generate_results): Allocate intermediate file.
1304 (release_structures): Clean-up properly fn_end.
1305 (output_intermediate_file): Start iterating with line 1.
1307 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1309 PR tree-optimization/82929
1310 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1311 ops_swapped_p non-static data member.
1312 (store_immediate_info::store_immediate_info): Clear it.
1313 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1314 ops set ops_swapped_p.
1315 (count_multiple_uses): Handle ops_swapped_p.
1317 2017-11-10 Martin Liska <mliska@suse.cz>
1319 * coverage.c (coverage_init): Stream information about
1320 support of has_unexecuted_blocks.
1321 * doc/gcov.texi: Document that.
1322 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1323 * gcov.c (read_graph_file): Likewise.
1324 (output_line_beginning): Fix a small issue with
1327 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1329 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1330 reference of trivial component.
1332 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1335 * gimple-ssa-store-merging.c
1336 (pass_store_merging::terminate_all_aliasing_chains): For
1337 gimple_store_p stmts also call refs_output_dependent_p.
1339 PR rtl-optimization/82913
1340 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1343 2017-11-09 Jeff Law <law@redhat.com>
1345 * vr-values.h: New file with vr_values class.
1346 * tree-vrp.c: Include vr-values.h
1347 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1348 data objects into the vr_values class.
1349 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1350 (get_value_range): Make it a member function within vr_values class.
1351 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1352 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1353 (op_with_constant_singleton_value_range): Likewise.
1354 (extract_range_for_var_from_comparison_expr): Likewise.
1355 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1356 (extract_range_from_binary_expr): Likewise.
1357 (extract_range_from_unary_expr): Likewise.
1358 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1359 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1360 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1361 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1362 (compare_name_with_value, compare_names): Likewise.
1363 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1364 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1365 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1366 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1367 (extract_range_from_phi_node): Likewise.
1368 (simplify_truth_ops_using_ranges): Likewise.
1369 (simplify_div_or_mod_using_ranges): Likewise.
1370 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1371 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1372 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1373 (simplify_float_conversion_using_ranges): Likewise.
1374 (simplify_internal_call_using_ranges): Likewise.
1375 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1376 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1377 poitner via x_vr_values for calls into gimple folder.
1378 (vrp_initialize_lattice): Make this the vr_values ctor.
1379 (vrp_free_lattice): Make this the vr_values dtor.
1380 (set_vr_value): New function.
1381 (class vrp_prop): Add vr_values data member. Add various member
1382 functions as well as member functions that delegate to vr_values.
1383 (check_array_ref): Make a member function within vrp_prop class.
1384 (search_for_addr_array, vrp_initialize): Likewise.
1385 (vrp_finalize): Likewise. Revamp to avoid direct access to
1386 vr_value, values_propagated, etc.
1387 (check_array_bounds): Extract vrp_prop class instance pointer from
1388 walk info structure. Use it to call member functions.
1389 (check_all_array_refs): Make a member function within vrp_prop class.
1390 Smuggle class instance pointer via walk info structure.
1391 (x_vr_values): New local static.
1392 (vrp_valueize): Use x_vr_values to get class instance.
1393 (vr_valueize_1): Likewise.
1394 (class vrp_folder): Add vr_values data member. Add various member
1395 functions as well as member functions that delegate to vr_values.
1396 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1397 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1398 class instance from vr_values. Use it to call member functions.
1399 (vrp_dom_walker): Add vr_values data member.
1400 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1401 instance via x_vr_values.
1402 (identify_jump_threads): Accept vr_values as argument. Store
1403 it into the walker structure.
1404 (evrp_dom_walker): Add vr_values class data member. Add various
1406 (evrp_dom_walker::try_find_new_range): Use vr_values data
1407 member to access the memory allocator.
1408 (evrp_dom_walker::before_dom_children): Store vr_values class
1409 instance into the vrp_folder class.
1410 (evrp_dom_walker::push_value_range): Rework to avoid direct
1411 access to num_vr_values and vr_value.
1412 (evrp_dom_walker::pop_value_range): Likewise.
1413 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1414 Use vr_values to get to dump_all_value_ranges member function.
1415 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1416 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1417 Pass vr_values class instance down to identify_jump_threads.
1418 Remove call to vrp_free_lattice.
1419 (debug_all_value_ranges): Remove.
1421 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1422 (vrp_folder): Likewise.
1424 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1425 Get it from the existing bitmap instead.
1426 (vrp_intersect_ranges_1): Likewise.
1428 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1430 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1432 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1433 set bit_not_p to it.
1434 (imm_store_chain_info::coalesce_immediate_stores): Break group
1435 if bit_not_p is different.
1436 (count_multiple_uses, split_group,
1437 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1438 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1439 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1440 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1441 to store_immediate_info ctor.
1443 2017-11-09 Jim Wilson <jimw@sifive.com>
1445 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1446 (scan_prog_file): Likewise.
1448 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1450 * bb-reorder.c (max_entry_frequency): Remove.
1451 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1452 connect_traces, push_to_next_round_p): Remove prototypes.
1453 (find_traces_1_round): Use counts only.
1454 (push_to_next_round_p): Likewise.
1455 (find_traces): Likewise.
1456 (rotate_loop): Likewise.
1457 (find_traces_1_round): Likewise.
1458 (connect_traces): Likewise.
1459 (edge_order): Likewise.
1461 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1463 * config/arm/arm.c (output_return_instruction): Add comments to
1464 indicate requirement for cmse_nonsecure_entry return to account
1465 for the size of clearing instruction output here.
1466 (thumb_exit): Likewise.
1467 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1468 return in hardfloat mode.
1470 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1472 * config/rs6000/rs6000.c (machine_function): Add a bool,
1473 "toc_is_wrapped_separately".
1474 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1475 if it wasn't explicitly set or unset, we are optimizing for speed, and
1476 doing separate shrink-wrapping.
1477 (rs6000_get_separate_components): Enable the TOC component if
1478 saving the TOC register in the prologue.
1479 (rs6000_components_for_bb): Handle the TOC component.
1480 (rs6000_emit_prologue_components): Store the TOC register where needed.
1481 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1482 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1484 2017-11-09 Martin Jambor <mjambor@suse.cz>
1486 * ipa-param-manipulation.c: New file.
1487 * ipa-param-manipulation.h: Likewise.
1488 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1489 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1490 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1491 (ipa_parm_adjustment): Likewise.
1492 (ipa_parm_adjustment_vec): Likewise.
1493 (ipa_get_vector_of_formal_parms): Moved declaration to
1494 ipa-param-manipulation.h.
1495 (ipa_get_vector_of_formal_parm_types): Likewise.
1496 (ipa_modify_formal_parameters): Likewise.
1497 (ipa_modify_call_arguments): Likewise.
1498 (ipa_combine_adjustments): Likewise.
1499 (ipa_dump_param_adjustments): Likewise.
1500 (ipa_modify_expr): Likewise.
1501 (ipa_get_adjustment_candidate): Likewise.
1502 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1503 ipa-param-manipulation.c.
1504 (ipa_get_vector_of_formal_parm_types): Likewise.
1505 (ipa_modify_formal_parameters): Likewise.
1506 (ipa_modify_call_arguments): Likewise.
1507 (ipa_modify_expr): Likewise.
1508 (get_ssa_base_param): Likewise.
1509 (ipa_get_adjustment_candidate): Likewise.
1510 (index_in_adjustments_multiple_times_p): Likewise.
1511 (ipa_combine_adjustments): Likewise.
1512 (ipa_dump_param_adjustments): Likewise.
1513 * tree-sra.c: Also include ipa-param-manipulation.h
1514 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
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_masked_store): 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_align_stack_vars): 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_variable_length): 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_unaligned_possible): Document.
1541 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1542 Alan Hayward <alan.hayward@arm.com>
1543 David Sherwood <david.sherwood@arm.com>
1545 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1547 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1548 Alan Hayward <alan.hayward@arm.com>
1549 David Sherwood <david.sherwood@arm.com>
1551 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1552 previously undocumented selectors.
1553 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1555 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1557 * doc/rtl.texi (const_vector): Say that elements can be
1558 const_wide_ints too.
1559 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1560 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1561 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1562 * optabs.c (expand_vector_broadcast): Likewise.
1564 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1565 Alan Hayward <alan.hayward@arm.com>
1566 David Sherwood <david.sherwood@arm.com>
1568 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1569 scaled index even if the unscaled address was invalid.
1570 Don't increase the complexity of using a scale in that case.
1572 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1573 Alan Hayward <alan.hayward@arm.com>
1574 David Sherwood <david.sherwood@arm.com>
1576 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1577 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1578 UNITS_PER_WORD bytes.
1579 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1580 based on REGMODE_NATURAL_SIZE of the inner mode.
1581 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1582 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1583 * expmed.c (lowpart_bit_field_p): Divide the value up into
1584 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1585 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1586 whether something is likely to occupy more than one register.
1588 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1591 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1592 profile_count::adjust_for_ipa_scaling.
1593 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1594 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1596 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1598 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1600 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1601 (split_group): Add total_orig and total_new arguments, estimate the
1602 number of statements related to the store group without store merging
1603 and with store merging.
1604 (imm_store_chain_info::output_merged_store): Adjust split_group
1605 callers, punt if estimated number of statements with store merging
1606 is not smaller than estimated number of statements without it.
1608 (handled_load): Remove has_single_use checks.
1609 (pass_store_merging::process_store): Likewise.
1611 2017-11-09 Richard Biener <rguenther@suse.de>
1613 PR tree-optimization/82902
1614 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1616 2017-11-09 Martin Liska <mliska@suse.cz>
1619 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1622 2017-11-09 Martin Liska <mliska@suse.cz>
1624 PR tree-optimization/82669
1625 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1627 2017-11-09 Martin Liska <mliska@suse.cz>
1629 PR gcov-profile/48463
1630 * coverage.c (coverage_begin_function): Output also end locus
1631 of a function and information whether the function is
1633 * gcov-dump.c (tag_function): Parse and print the information.
1634 * gcov.c (INCLUDE_MAP): Add include.
1635 (INCLUDE_SET): Likewise.
1636 (struct line_info): Move earlier in the source file because
1637 of vector<line_info> in function_info structure.
1638 (line_info::line_info): Likewise.
1639 (line_info::has_block): Likewise.
1640 (struct source_info): Add new member index.
1641 (source_info::get_functions_at_location): New function.
1642 (function_info::group_line_p): New function.
1643 (output_intermediate_line): New function.
1644 (output_intermediate_file): Use the mentioned function.
1645 (struct function_start): New.
1646 (struct function_start_pair_hash): Likewise.
1647 (process_file): Add code that identifies group functions.
1648 Assign lines either to global or function scope.
1649 (generate_results): Skip artificial functions.
1650 (find_source): Assign index for each source file.
1651 (read_graph_file): Read new flag artificial and end_line.
1652 (add_line_counts): Assign it either to global of function scope.
1653 (accumulate_line_counts): Isolate core of the function to
1654 accumulate_line_info and call it for both function and global
1656 (accumulate_line_info): New function.
1657 (output_line_beginning): Fix GNU coding style.
1658 (print_source_line): New function.
1659 (output_line_details): Likewise.
1660 (output_function_details): Likewise.
1661 (output_lines): Iterate both source (global) scope and function
1663 (struct function_line_start_cmp): New class.
1664 * doc/gcov.texi: Reflect changes in documentation.
1666 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1669 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1670 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1671 and similarly for not instead of neg.
1673 2017-11-08 Andi Kleen <ak@linux.intel.com>
1675 * config/i386/i386.opt: Add -mforce-indirect-call.
1676 * config/i386/predicates.md: Check for flag_force_indirect_call.
1677 * doc/invoke.texi: Document -mforce-indirect-call
1679 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1681 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1683 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1684 config/riscv/riscv.c (predict.h): New include.
1685 (riscv_slow_unaligned_access_p): No longer static.
1686 (riscv_block_move_straight): Add require.
1687 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1689 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1692 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1693 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1694 nonimmediate_operand predicate for operand 1 instead of
1697 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1699 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1701 * config/aarch64/constraints.md (Uml): New constraint.
1702 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1705 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1707 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1708 of two vec_duplicates into a vec_concat.
1710 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1712 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1713 Simplify vec_merge of vec_duplicate and vec_concat.
1714 * config/aarch64/constraints.md (Utq): New constraint.
1715 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1718 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1720 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1721 Simplify vec_merge of vec_duplicate and const_vector.
1722 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1724 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1725 mode iterator. Update predicate on operand 1 to
1726 handle non-const_vec constants. Delete constraints.
1727 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1729 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1731 PR tree-optimization/78821
1732 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1734 (store_operand_info::store_operand_info): Initialize it to false.
1735 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1736 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1737 store in the group, and if chain_info is non-NULL, to ignore altogether
1739 (compatible_load_p): Fail if bit_not_p does not match.
1740 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1741 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1742 (pass_store_merging::process_store): Adjust
1743 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1744 call terminate_all_aliasing_chains newly when adding a store into
1745 a chain with non-NULL chain_info.
1747 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1749 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1751 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1753 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1755 (aarch64_layout_frame): Initialise emit_frame_chain.
1756 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1757 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1759 2017-11-08 Martin Liska <mliska@suse.cz>
1761 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1762 of gimple_call_internal_p.
1764 2017-11-07 Tom de Vries <tom@codesourcery.com>
1766 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1768 2017-11-07 Tom de Vries <tom@codesourcery.com>
1770 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1773 2017-11-08 Martin Liska <mliska@suse.cz>
1776 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1778 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1780 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1782 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1785 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1786 (*cmp<mode>_ccz_1): New insn with $k alternative.
1789 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1790 RTX_COMM_COMPARE as commutative as well.
1791 (ix86_binary_operator_ok): Formatting fix.
1792 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1793 *<code><mode>3<mask_name><round_saeonly_name>,
1794 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1795 *<s>mul<mode>3_highpart<mask_name>,
1796 *vec_widen_umult_even_v16si<mask_name>,
1797 *vec_widen_umult_even_v8si<mask_name>,
1798 *vec_widen_umult_even_v4si<mask_name>,
1799 *vec_widen_smult_even_v16si<mask_name>,
1800 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1801 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1802 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1803 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1804 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1805 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1806 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1807 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1808 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1809 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1810 ix86_binary_operator_ok. Formatting fixes.
1811 (*<plusminus_insn><mode>3<mask_name><round_name>,
1812 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1815 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1817 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1818 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1819 eq and ne if TARGET_ISEL.
1820 (cmp): New code_iterator.
1821 (UNS, UNSU_, UNSIK): New code_attrs.
1822 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1823 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1825 ("eq<mode>3"): ... this.
1826 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1828 ("ne<mode>3"): ... this.
1830 2017-11-07 Julia Koval <julia.koval@intel.com>
1833 * common/config/i386/i386-common.c
1834 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1835 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1836 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1837 * config/i386/i386.opt: Ditto.
1838 * config/i386/i386.c (ix86_target_string): Ditto.
1839 (ix86_option_override_internal): Ditto.
1840 (ix86_init_mpx_builtins): Move MPX to args2.
1841 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1842 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1843 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1844 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1845 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1846 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1847 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1849 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1852 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1853 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1854 (zero-extendsidi peephole2): Remove peephole.
1856 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1859 * stor-layout.c: Include attribs.h.
1860 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1861 explicit lookup of "aligned" attribute.
1863 2017-11-07 Andrew Waterman <andrew@sifive.com>
1865 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1866 (riscv_expand_block_move): Likewise.
1867 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1869 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1870 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1871 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1872 (riscv_adjust_block_mem): Likewise.
1873 (riscv_block_move_loop): Likewise.
1874 (riscv_expand_block_move): Likewise.
1875 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1877 2017-11-07 Michael Clark <michaeljclark@mac.com>
1879 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1880 (MUSL_DYNAMIC_LINKER): Likewise.
1882 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1884 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1885 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1888 2017-11-07 Richard Biener <rguenther@suse.de>
1890 * match.pd: Fix build.
1892 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1893 Jackson Woodruff <jackson.woodruff@arm.com>
1895 PR tree-optimization/71026
1896 * match.pd: Canonicalize negate in division.
1898 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1901 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1903 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1905 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1906 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1908 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1910 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1911 non-scalar integral types.
1912 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1913 (-(A-B), -(~A)): New transformations.
1915 2017-11-07 Tom de Vries <tom@codesourcery.com>
1917 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1918 semicolon after "do {} while (0)".
1919 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1920 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1921 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1922 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1923 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1925 2017-11-07 Tom de Vries <tom@codesourcery.com>
1927 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1928 after "do {} while (0)".
1929 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1930 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1931 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1932 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1933 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1935 2017-11-07 Tom de Vries <tom@codesourcery.com>
1938 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1940 (arm_rtx_costs_internal): Add missing semicolon after
1941 HANDLE_NARROW_SHIFT_ARITH call.
1943 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1945 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1946 disable isel if it was not set explicitly.
1948 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1950 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1952 (add_type_attribute) likewise.
1954 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1956 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1957 of struct ix86_frame.
1958 (ix86_initial_elimination_offset): Likewise.
1959 (ix86_expand_split_stack_prologue): Likewise.
1961 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1963 * tree-vrp.h (enum value_range_type): Update stale comment.
1965 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1966 Alan Hayward <alan.hayward@arm.com>
1967 David Sherwood <david.sherwood@arm.com>
1969 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1970 (aarch64_expand_vec_perm_const): Take the number of units too.
1971 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1972 (aarch64_expand_vec_perm_const): Likewise.
1973 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1974 (vec_perm<mode>): Update accordingly.
1976 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1977 Alan Hayward <alan.hayward@arm.com>
1978 David Sherwood <david.sherwood@arm.com>
1980 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1981 Take the number of units too.
1982 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1983 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1984 but check for a vector mode before rather than after the call.
1985 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1986 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1987 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1988 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1989 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1990 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1991 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1992 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1993 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1994 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1995 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1996 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1997 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1998 (aarch64_sqdmull2_n<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_reverse_mask): Take
2005 the number of units too.
2006 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2007 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2008 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2009 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2010 (vec_store_lanesxi<mode>): Update accordingly.
2012 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2013 Alan Hayward <alan.hayward@arm.com>
2014 David Sherwood <david.sherwood@arm.com>
2016 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2017 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2018 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2019 of units rather than the mode.
2020 * config/aarch64/iterators.md (nunits): New mode attribute.
2021 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2022 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2023 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2024 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2025 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2026 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2027 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2028 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2029 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2030 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2031 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2032 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2033 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2034 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2035 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2036 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2037 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2038 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2039 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2040 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2041 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2042 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2043 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2044 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2045 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2046 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2047 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2048 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2049 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2050 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2051 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2052 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2053 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2054 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2055 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2056 (aarch64_simd_vec_setv2di): Likewise.
2058 2017-11-06 Carl Love <cel@us.ibm.com>
2060 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2062 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2063 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2065 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2067 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2068 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2069 Add power 8 macro expansions.
2070 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2071 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2072 power 8 instructions. (VSX_XXBR): Add iterator.
2074 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2076 * config/arm/arm.md (predicable_short_it): Change default to "no",
2077 improve documentation, remove uses that are identical to the default.
2078 (enabled_for_depr_it): Rename to enabled_for_short_it.
2079 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2080 Remove default uses.
2081 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2082 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2083 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2084 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2086 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2089 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2090 float128 helper macros, which are no longer used after deleting
2091 the old 'q' built-in functions, and moving the round to odd
2092 built-in functions to being special built-in functions.
2093 (BU_FLOAT128_2): Likewise.
2094 (BU_FLOAT128_1_HW): Likewise.
2095 (BU_FLOAT128_2_HW): Likewise.
2096 (BU_FLOAT128_3_HW): Likewise.
2097 (FABSQ): Delete old 'q' built-in functions.
2098 (COPYSIGNQ): Likewise.
2099 (SQRTF128_ODD): Move round to odd built-in functions to be
2100 special built-in functions, so that we can handle
2101 -mabi=ieeelongdouble.
2102 (TRUNCF128_ODD): Likewise.
2103 (ADDF128_ODD): Likewise.
2104 (SUBF128_ODD): Likewise.
2105 (MULF128_ODD): Likewise.
2106 (DIVF128_ODD): Likewise.
2107 (FMAF128_ODD): Likewise.
2108 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2109 built-in names to 'f128'.
2110 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2111 old 'q' built-in functions, as the machine independent code for
2112 'f128' built-in functions handles this.
2113 (rs6000_expand_builtin): Add expansion for float128 round to odd
2114 functions, keying off on -mabi=ieeelongdouble of whether to use
2115 the KFmode or TFmode variant.
2116 (rs6000_init_builtins): Initialize the _Float128 round to odd
2118 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2119 _Float128 'q' built-in functions are now mapped into the new
2120 'f128' built-in functions.
2122 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2124 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2125 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2127 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2130 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2131 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2132 * toplev.c: Include "ipa-fnsummary.h".
2133 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2135 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2137 PR tree-optimization/82838
2138 * gimple-ssa-store-merging.c
2139 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2140 on a separate gimple_seq which is then appended to seq.
2142 2017-11-06 Jeff Law <law@redhat.com>
2145 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2146 (get_probe_interval): New functions.
2147 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2148 (ix86_adjust_stack_and_probe): Likewise.
2149 (output_adjust_stack_and_probe): Likewise.
2150 (ix86_emit_probe_stack_range): Likewise.
2151 (ix86_expand_prologue): Likewise.
2153 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2155 PR tree-optimization/82816
2156 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2157 if the modes of the two types are the same.
2158 (convert_plusminus_to_widen): Likewise.
2160 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2162 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2164 (usadv16qi): New define_expand.
2165 (usadv8hi): New define_expand.
2167 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2170 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2172 (inline_transform): Likewise.
2173 * predict.c (counts_to_freqs): Remove useless conditional.
2174 * profile-count.h (profile_count::apply_scale): Move sanity check.
2175 * tree-inline.c (copy_bb): Always scale.
2176 (copy_cfg_body): Likewise.
2178 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2181 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2183 (*cmp_ite0): Add enabled_for_depr_it attribute.
2184 (*cmp_ite1): Likewise.
2186 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2188 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2191 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2193 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2194 return the value by pointer.
2195 (extract_range_from_multiplicative_op_1): Update accordingly.
2196 Return as soon as an operation fails.
2198 2017-11-05 Tom de Vries <tom@codesourcery.com>
2201 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2202 (DEF_SANITIZER_BUILTIN): ... here.
2203 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2204 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2206 2017-11-05 Tom de Vries <tom@codesourcery.com>
2209 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2211 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2212 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2213 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2215 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2216 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2217 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2219 2017-11-05 Tom de Vries <tom@codesourcery.com>
2222 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2225 2017-11-04 Michael Clark <michaeljclark@mac.com>
2227 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2228 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2230 (*addsi3_extended): Likewise.
2231 (*addsi3_extended2): Likewise.
2232 (<optab>si3): Likewise.
2233 (<optab>di3): Likewise.
2234 (<optab><mode>3): Likewise.
2235 (<*optabe>si3_internal): Likewise.
2236 (zero_extendqi<SUPERQI:mode>2): Likewise.
2237 (*add<mode>hi3): Likewise.
2238 (*xor<mode>hi3): Likewise.
2239 (<optab>di3): Likewise.
2240 (*<optab>si3_extend): Likewise.
2241 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2242 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2243 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2245 2017-11-04 Andrew Waterman <andrew@sifive.com>
2247 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2248 TARGET_STRICT_ALIGN based upon -mtune argument.
2250 2017-11-04 Andrew Waterman <andrew@sifive.com>
2252 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2254 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2256 config/i386/i386.c (choose_basereg): Use optional scratch
2257 register and add assertion.
2258 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2259 needed, and don't allocate stack.
2260 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2261 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2262 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2264 2017-11-03 Jeff Law <law@redhat.com>
2266 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2267 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2268 to probe at the start of a noreturn function.
2270 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2272 PR tree-optimization/78821
2273 * gimple-ssa-store-merging.c: Update the file comment.
2274 (MAX_STORE_ALIAS_CHECKS): Define.
2275 (struct store_operand_info): New type.
2276 (store_operand_info::store_operand_info): New constructor.
2277 (struct store_immediate_info): Add rhs_code and ops data members.
2278 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2279 and op1r arguments to the ctor, initialize corresponding data members.
2280 (struct merged_store_group): Add load_align_base and load_align
2282 (merged_store_group::merged_store_group): Initialize them.
2283 (merged_store_group::do_merge): Update them.
2284 (merged_store_group::apply_stores): Pick the constant for
2285 encode_tree_to_bitpos from one of the two operands, or skip
2286 encode_tree_to_bitpos if neither operand is a constant.
2287 (class pass_store_merging): Add process_store method decl. Remove
2288 bool argument from terminate_all_aliasing_chains method decl.
2289 (pass_store_merging::terminate_all_aliasing_chains): Remove
2290 var_offset_p argument and corresponding handling.
2291 (stmts_may_clobber_ref_p): New function.
2292 (compatible_load_p): New function.
2293 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2294 if there is overlap and rhs_code is not INTEGER_CST. For
2295 non-overlapping stores terminate group if rhs is not mergeable.
2296 (get_alias_type_for_stmts): Change first argument from
2297 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2298 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2299 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2301 (get_location_for_stmts): Change first argument from
2302 auto_vec<gimple *> & to vec<gimple *> &.
2303 (struct split_store): Remove orig_stmts data member, add orig_stores.
2304 (split_store::split_store): Create orig_stores rather than orig_stmts.
2305 (find_constituent_stmts): Renamed to ...
2306 (find_constituent_stores): ... this. Change second argument from
2307 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2308 to info structures rather than the statements.
2309 (split_group): Rename ALLOW_UNALIGNED argument to
2310 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2311 it. Adjust find_constituent_stores caller.
2312 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2313 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2314 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2315 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2316 (mem_valid_for_store_merging): New function.
2317 (handled_load): New function.
2318 (pass_store_merging::process_store): New method.
2319 (pass_store_merging::execute): Use process_store method. Adjust
2320 terminate_all_aliasing_chains caller.
2322 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2324 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2325 Return true for more constants, symbols and label references.
2326 (aarch64_valid_floating_const): Remove unused function.
2328 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2331 * config/aarch64/aarch64.c (aarch64_layout_frame):
2332 Undo forcing of LR at bottom of frame.
2334 2017-11-03 Jeff Law <law@redhat.com>
2337 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2338 for int_registers_saved.
2340 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2341 extracted from tree-ssa-dom.c.
2342 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2343 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2344 (record_equivalences_from_incoming_edge): Add additional argument
2345 to single_pred_edge_ignoring_loop_edges call.
2346 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2347 (uncprop_dom_walker::before_dom_children): Add additional argument
2348 to single_pred_edge_ignoring_loop_edges call.
2349 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2350 single_pred_edge_ignoring_loop_edges rather than open coding.
2351 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2353 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2355 * match.pd (-(-A)): Rewrite.
2357 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2359 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2360 (rs6000_emit_int_cmove): New declaration.
2361 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2362 (rs6000_emit_sISEL): Delete.
2363 (rs6000_emit_int_cmove): Make non-static.
2364 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2365 instead of rs6000_emit_sISEL.
2367 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2369 * asan.c (create_cond_insert_point): Maintain profile.
2370 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2372 * basic-block.h (struct basic_block_def): Remove frequency.
2373 (EDGE_FREQUENCY): Use to_frequency
2374 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2376 (find_traces): Update to use to_frequency.
2377 (find_traces_1_round): Likewise; use only IPA counts.
2378 (bb_to_key): Likewise.
2379 (connect_traces): Use IPA counts only.
2380 (copy_bb_p): Update to use to_frequency.
2381 (fix_up_crossing_landing_pad): Likewise.
2382 (sanitize_hot_paths): Likewise.
2383 * bt-load.c (basic_block_freq): Likewise.
2384 * cfg.c (init_flow): Set count_max to uninitialized.
2385 (check_bb_profile): Remove frequencies; check counts.
2386 (dump_bb_info): Do not dump frequencies.
2387 (update_bb_profile_for_threading): Update counts only.
2388 (scale_bbs_frequencies_int): Likewise.
2389 (MAX_SAFE_MULTIPLIER): Remove.
2390 (scale_bbs_frequencies_gcov_type): Update counts only.
2391 (scale_bbs_frequencies_profile_count): Update counts only.
2392 (scale_bbs_frequencies): Update counts only.
2393 * cfg.h (struct control_flow_graph): Add count-max.
2394 (update_bb_profile_for_threading): Update prototype.
2395 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2396 (find_many_sub_basic_blocks): Likewise.
2397 * cfgcleanup.c (try_forward_edges): Likewise.
2398 (try_crossjump_to_edge): Likewise.
2399 * cfgexpand.c (expand_gimple_cond): Likewise.
2400 (expand_gimple_tailcall): Likewise.
2401 (construct_init_block): Likewise.
2402 (construct_exit_block): Likewise.
2403 * cfghooks.c (verify_flow_info): Check consistency of counts.
2404 (dump_bb_for_graph): Do not dump frequencies.
2405 (split_block_1): Do not update frequencies.
2406 (split_edge): Do not update frequencies.
2407 (make_forwarder_block): Do not update frequencies.
2408 (duplicate_block): Do not update frequencies.
2409 (account_profile_record): Do not update frequencies.
2410 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2411 for global heuristics.
2412 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2413 (expected_loop_iterations_unbounded): Use counts only.
2414 * cfgloopmanip.c (scale_loop_profile): Simplify.
2415 (create_empty_loop_on_edge): Simplify
2417 (duplicate_loop_to_header_edge): Simplify
2418 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2419 (update_br_prob_note): Take care of removing note when profile
2421 (relink_block_chain): Do not dump frequency.
2422 (rtl_account_profile_record): Use to_frequency.
2423 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2424 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2425 (cgraph_update_edges_for_call_stmt_node): Likewise.
2426 (cgraph_edge::verify_count_and_frequency): Update.
2427 (cgraph_node::verify_node): Temporarily disable frequency verification.
2428 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2429 to_cgraph_frequency.
2430 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2431 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2433 (cgraph_node::expand_thunk): Update profile.
2434 * except.c (dw2_build_landing_pads): Do not update frequency.
2435 * final.c (compute_alignments): Use to_frequency.
2436 (dump_basic_block_info): Do not dump frequency.
2437 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2438 (dump_gimple_bb_header): Do not dump frequency.
2439 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2441 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2442 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2443 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2444 (init_before_recovery): Do not update frequency.
2445 (sched_create_recovery_edges): Do not update frequency.
2446 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2447 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2448 (ipa_cp_c_finalize): Set max_count to uninitialized.
2449 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2450 (param_change_prob): Use counts.
2451 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2453 * ipa-split.c (consider_split): Use to_frequency.
2454 (split_function): Use to_frequency.
2455 * ira-build.c (loop_compare_func): Likewise.
2456 (mark_loops_for_removal): Likewise.
2457 (mark_all_loops_for_removal): Likewise.
2458 * loop-doloop.c (doloop_modify): Do not update frequency.
2459 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2461 * lto-streamer-in.c (input_function): Update count_max.
2462 * omp-expand.c (expand_omp_taskreg): Update count_max.
2463 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2464 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2465 (maybe_hot_count_p): Use ipa counts only.
2466 (maybe_hot_bb_p): Simplify.
2467 (maybe_hot_edge_p): Simplify.
2468 (probably_never_executed): Do not take frequency argument.
2469 (probably_never_executed_bb_p): Do not pass frequency.
2470 (probably_never_executed_edge_p): Likewise.
2471 (combine_predictions_for_bb): Check that profile is nonzero.
2472 (propagate_freq): Do not set frequency.
2473 (drop_profile): Simplify.
2474 (counts_to_freqs): Simplify.
2475 (expensive_function_p): Use to_frequency.
2476 (propagate_unlikely_bbs_forward): Simplify.
2477 (determine_unlikely_bbs): Simplify.
2478 (estimate_bb_frequencies): Add hack to silence graphite issues.
2479 (compute_function_frequency): Use ipa counts.
2480 (pass_profile::execute): Update.
2481 (rebuild_frequencies): Use counts only.
2482 (force_edge_cold): Use counts only.
2483 * profile-count.c (profile_count::dump): Dump new count types.
2484 (profile_count::differs_from_p): Check compatiblity.
2485 (profile_count::to_frequency): New function.
2486 (profile_count::to_cgraph_frequency): New function.
2487 * profile-count.h (struct function): Declare.
2488 (enum profile_quality): Add profile_guessed_local and
2489 profile_guessed_global0.
2490 (class profile_proability): Decrease number of bits to 29;
2491 update from_reg_br_prob_note and to_reg_br_prob_note.
2492 (class profile_count: Update comment; decrease number of bits
2493 to 61. Check compatibility.
2494 (profile_count::compatible_p): New private member function.
2495 (profile_count::ipa_p): New member function.
2496 (profile_count::operator<): Handle global zero correctly.
2497 (profile_count::operator>): Handle global zero correctly.
2498 (profile_count::operator<=): Handle global zero correctly.
2499 (profile_count::operator>=): Handle global zero correctly.
2500 (profile_count::nonzero_p): New member function.
2501 (profile_count::force_nonzero): New member function.
2502 (profile_count::max): New member function.
2503 (profile_count::apply_scale): Handle IPA scalling.
2504 (profile_count::guessed_local): New member function.
2505 (profile_count::global0): New member function.
2506 (profile_count::ipa): New member function.
2507 (profile_count::to_frequency): Declare.
2508 (profile_count::to_cgraph_frequency): Declare.
2509 * profile.c (OVERLAP_BASE): Delete.
2510 (compute_frequency_overlap): Delete.
2511 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2512 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2513 * sched-ebb.c (rank): Use counts only.
2514 * shrink-wrap.c (handle_simple_exit): Use counts only.
2515 (try_shrink_wrapping): Use counts only.
2516 (place_prologue_for_one_component): Use counts only.
2517 * tracer.c (find_best_predecessor): Use to_frequency.
2518 (find_trace): Use to_frequency.
2519 (tail_duplicate): Use to_frequency.
2520 * trans-mem.c (expand_transaction): Do not update frequency.
2521 * tree-call-cdce.c: Do not update frequency.
2522 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2523 (gimple_merge_blocks): Likewise.
2524 (gimple_split_edge): Likewise.
2525 (gimple_duplicate_sese_region): Likewise.
2526 (gimple_duplicate_sese_tail): Likewise.
2527 (move_sese_region_to_fn): Likewise.
2528 (gimple_account_profile_record): Likewise.
2529 (insert_cond_bb): Likewise.
2530 * tree-complex.c (expand_complex_div_wide): Likewise.
2531 * tree-eh.c (lower_resx): Update profile.
2532 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2534 (initialize_cfun): Do not initialize frequencies
2535 (freqs_to_counts): Delete.
2536 (copy_cfg_body): Ignore count parameter.
2537 (copy_body): Update.
2538 (expand_call_inline): Update count_max.
2539 (optimize_inline_calls): Update count_max.
2540 (tree_function_versioning): Update count_max.
2541 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2542 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2544 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2545 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2546 (try_peel_loop): Likewise.
2547 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2549 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2550 (tree_transform_and_unroll_loop): Do not use frequencies
2551 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2552 Use reliable prediction only.
2553 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2554 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2555 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2556 probability scaling.
2557 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2558 not update frequency
2559 (any_remaining_duplicated_blocks): Likewise.
2560 (update_profile): Likewise.
2561 (estimated_freqs_path): Delete.
2562 (freqs_to_counts_path): Delete.
2563 (clear_counts_path): Delete.
2564 (ssa_fix_duplicate_block_edges): Likewise.
2565 (duplicate_thread_path): Likewise.
2566 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2567 * tree-tailcall.c (decrease_profile): Do not update frequency.
2568 (eliminate_tail_call): Likewise.
2569 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2570 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2571 (optimize_mask_stores): Likewise.
2572 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2573 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2574 (ubsan_expand_ptr_ifn): Update profile.
2575 * value-prof.c (gimple_ic): Simplify.
2576 * value-prof.h (gimple_ic): Update prototype.
2577 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2578 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2580 (want_inline_self_recursive_call_p): Likewise.
2581 (resolve_noninline_speculation): Only cummulate defined counts.
2582 (inline_small_functions): Use nonzero_p.
2583 (ipa_inline): Do not access freed node.
2585 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2587 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2588 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2590 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2592 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2593 non-legitimate address.
2595 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2597 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2598 (*lt0_<mode>di, *lt0_<mode>si): New.
2600 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2602 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2603 TARGET_PAIRED_FLOAT.
2605 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2606 Jim Wilson <jim.wilson@linaro.org>
2608 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2609 * config/aarch64/aarch64-tune.md: Regenerated.
2610 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2611 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2613 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2615 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2616 (arc_expand_prologue): Restore blink for millicode.
2617 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2619 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2622 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2623 gen_vec_duplicate after forcing the scalar into a register.
2625 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2627 * combine (try_combine): Print the insns input to try_combine to the
2630 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2633 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2634 Remove second argument from aarch64_process_target_attr call.
2635 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2637 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2639 (aarch64_handle_attr_arch): Remove second argument.
2640 (aarch64_handle_attr_cpu): Ditto.
2641 (aarch64_handle_attr_tune): Ditto.
2642 (aarch64_handle_attr_isa_flags): Ditto.
2643 (aarch64_process_one_target_attr): Ditto.
2644 (aarch64_process_target_attr): Ditto.
2645 (aarch64_option_valid_attribute_p): Remove second argument.
2646 on aarch64_process_target_attr call.
2648 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2650 * diagnostic.c: Include "selftest-diagnostic.h".
2651 (selftest::assert_location_text): New function.
2652 (selftest::test_diagnostic_get_location_text): New function.
2653 (selftest::diagnostic_c_tests): Call it.
2655 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2657 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2658 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2659 (class selftest::test_diagnostic_context): Move to...
2660 * selftest-diagnostic.c: New file.
2661 * selftest-diagnostic.h: New file.
2663 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2665 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2666 offset range for FT32B.
2667 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2668 * config/ft32/ft32.md: Add TARGET_NOPM.
2669 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2670 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2672 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2674 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2676 2017-11-02 Jeff Law <law@redhat.com>
2678 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2679 virtual keyword on FINAL OVERRIDE members.
2681 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2682 virtuals together. Add virtual destructor.
2683 (substitute_and_fold_engine): Similarly.
2685 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2687 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2689 2017-11-02 Richard Biener <rguenther@suse.de>
2691 PR tree-optimization/82795
2692 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2694 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2696 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2698 (gcc_GAS_CHECK_FEATURE): Remove.
2699 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2700 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2701 * configure: Regenerate.
2703 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2705 * config/arc/arc.c (hwloop_optimize): Account for empty
2708 2017-11-02 Richard Biener <rguenther@suse.de>
2711 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2712 Truncate ARRAY_REF index and element size.
2714 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2716 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2718 2017-11-01 Jeff Law <law@redhat.com>
2720 * tree-ssa-ccp.c (ccp_folder): New class derived from
2721 substitute_and_fold_engine.
2722 (ccp_folder::get_value): New member function.
2723 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2724 (ccp_fold_stmt): Remove prototype.
2725 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2726 * tree-ssa-copy.c (copy_folder): New class derived from
2727 substitute_and_fold_engine.
2728 (copy_folder::get_value): Renamed from get_value.
2729 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2730 * tree-vrp.c (vrp_folder): New class derived from
2731 substitute_and_fold_engine.
2732 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2733 (vrp_folder::get_value): New member function.
2734 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2735 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2736 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2737 provide a class interface to folder/substitute routines.
2738 (ssa_prop_fold_stmt_fn): Remove typedef.
2739 (ssa_prop_get_value_fn): Likewise.
2740 (subsitute_and_fold): Remove prototype.
2741 (replace_uses_in): Likewise.
2742 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2743 Renamed from replace_uses_in. Call the virtual member function
2744 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2745 (substitute_and_fold_dom_walker): Remove initialization of
2746 data member entries for calbacks. Add substitute_and_fold_engine
2747 member and initialize it.
2748 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2749 member functions for get_value, replace_phi_args_in c
2750 replace_uses_in, and fold_stmt calls.
2751 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2752 substitute_and_fold. Remove assert. Update ctor call.
2754 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2755 (ssa_prop_visit_phi_fn): Likewise.
2756 (class ssa_propagation_engine): New class to provide an interface
2758 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2760 (ssa_prop_visit_phi): Likewise.
2761 (ssa_propagation_engine::simulate_stmt): Moved into class.
2762 Call visit_phi/visit_stmt from the class rather than via
2763 file scoped static variables.
2764 (ssa_propagation_engine::simulate_block): Moved into class.
2765 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2766 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2767 set file scoped statics for the visit_stmt/visit_phi callbacks.
2768 * tree-complex.c (complex_propagate): New class derived from
2769 ssa_propagation_engine.
2770 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2771 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2772 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2774 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2775 ssa_propagation_engine.
2776 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2777 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2778 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2779 * tree-ssa-copy.c (copy_prop): New class derived from
2780 ssa_propagation_engine.
2781 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2782 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2783 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2784 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2785 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2786 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2787 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2789 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2791 PR rtl-optimization/82778
2792 PR rtl-optimization/82597
2793 * compare-elim.c (struct comparison): Add in_a_setter field.
2794 (find_comparison_dom_walker::before_dom_children): Remove killed
2795 bitmap and df_simulate_find_defs call, instead walk the defs.
2796 Compute last_setter and initialize in_a_setter. Merge definitions
2797 with first initialization for a few variables.
2798 (try_validate_parallel): Use insn_invalid_p instead of
2799 recog_memoized. Return insn rather than just the pattern.
2800 (try_merge_compare): Fix up comment. Don't uselessly test if
2801 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2803 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2806 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2807 Alan Hayward <alan.hayward@arm.com>
2808 David Sherwood <david.sherwood@arm.com>
2810 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2811 aarch64_hard_regno_nregs to get the number of registers
2814 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2815 Alan Hayward <alan.hayward@arm.com>
2816 David Sherwood <david.sherwood@arm.com>
2818 * config/aarch64/constraints.md (Upl): Rename to...
2820 * config/aarch64/aarch64.md
2821 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2824 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2825 Alan Hayward <alan.hayward@arm.com>
2826 David Sherwood <david.sherwood@arm.com>
2828 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2829 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2832 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2833 Alan Hayward <alan.hayward@arm.com>
2834 David Sherwood <david.sherwood@arm.com>
2836 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2837 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2838 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2840 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2841 of a single element.
2842 * config/aarch64/iterators.md: Add a comment above the permute
2843 unspecs to say that they are generated directly by
2844 aarch64_expand_vec_perm_const.
2845 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2847 2017-11-01 Nathan Sidwell <nathan@acm.org>
2849 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2851 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2853 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2854 and medany code models, and describe what they do.
2856 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2858 Revert accidental duplicate:
2860 * combine.c (can_change_dest_mode): Reject changes in
2861 REGMODE_NATURAL_SIZE.
2863 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2865 PR rtl-optimization/64682
2866 PR rtl-optimization/69567
2867 PR rtl-optimization/69737
2868 PR rtl-optimization/82683
2869 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2870 register mentioned in the note, drop the note, unless it came from I3,
2871 in which case it should go to I3 again.
2873 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2875 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2876 and return false if not.
2877 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2879 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2881 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2882 range is known to be empty.
2884 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2885 Alan Hayward <alan.hayward@arm.com>
2886 David Sherwood <david.sherwood@arm.com>
2888 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2889 and CONST_VECTOR_NUNITS instead of computing the number of units from
2890 the byte sizes of the vector and element.
2891 (simplify_binary_operation_1): Likewise.
2892 (simplify_const_binary_operation): Likewise.
2893 (simplify_ternary_operation): Likewise.
2895 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2896 Alan Hayward <alan.hayward@arm.com>
2897 David Sherwood <david.sherwood@arm.com>
2899 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2900 (int_mem_offset): ...this new function.
2901 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2902 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2903 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2906 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2907 Alan Hayward <alan.hayward@arm.com>
2908 David Sherwood <david.sherwood@arm.com>
2910 * lower-subreg.c (interesting_mode_p): New function.
2911 (compute_costs, find_decomposable_subregs, decompose_register)
2912 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2913 (resolve_clobber, dump_choices): Use it.
2915 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2916 Alan Hayward <alan.hayward@arm.com>
2917 David Sherwood <david.sherwood@arm.com>
2919 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2921 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2922 Alan Hayward <alan.hayward@arm.com>
2923 David Sherwood <david.sherwood@arm.com>
2925 * alias.c (find_base_value, find_base_term): Only process integer
2926 truncations. Check the precision rather than the size.
2928 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2929 Alan Hayward <alan.hayward@arm.com>
2930 David Sherwood <david.sherwood@arm.com>
2932 * machmode.h (is_narrower_int_mode): New function
2933 * optabs.c (expand_float, expand_fix): Use it.
2934 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2936 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2937 Alan Hayward <alan.hayward@arm.com>
2938 David Sherwood <david.sherwood@arm.com>
2940 * rtl.h (narrower_subreg_mode): New function.
2941 * ira-color.c (update_costs_from_allocno): Use it.
2943 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2944 Alan Hayward <alan.hayward@arm.com>
2945 David Sherwood <david.sherwood@arm.com>
2947 * optabs-query.h (convert_optab_p): New function, split out from...
2948 (convert_optab_handler): ...here.
2949 (widening_optab_handler): Delete.
2950 (find_widening_optab_handler): Remove permit_non_widening parameter.
2951 (find_widening_optab_handler_and_mode): Likewise. Provide an
2952 override that operates on mode class wrappers.
2953 * optabs-query.c (widening_optab_handler): Delete.
2954 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2955 parameter. Assert that the two modes are the same class and that
2956 the "from" mode is narrower than the "to" mode. Use
2957 convert_optab_handler instead of widening_optab_handler.
2958 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2959 instead of widening_optab_handler.
2960 * expr.c (expand_expr_real_2): Update calls to
2961 find_widening_optab_handler.
2962 * optabs.c (expand_widen_pattern_expr): Likewise.
2963 (expand_binop_directly): Take the insn_code as a parameter.
2964 (expand_binop): Only call find_widening_optab_handler for
2965 conversion optabs; use optab_handler otherwise. Update calls
2966 to find_widening_optab_handler and expand_binop_directly.
2967 Use convert_optab_handler instead of widening_optab_handler.
2968 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2969 find_widening_optab_handler and use scalar_mode rather than
2971 (convert_plusminus_to_widen): Likewise.
2973 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2974 Alan Hayward <alan.hayward@arm.com>
2975 David Sherwood <david.sherwood@arm.com>
2977 * machmode.h (fixed_size_mode): New class.
2978 * rtl.h (get_pool_mode): Return fixed_size_mode.
2979 * gengtype.c (main): Add fixed_size_mode.
2980 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2981 (get_raw_arg_mode): Likewise.
2982 * doc/tm.texi: Regenerate.
2983 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2984 * targhooks.c (default_get_reg_raw_mode): Likewise.
2985 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2986 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2987 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2988 (msp430_get_raw_result_mode): Likewise.
2989 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2990 * dbxout.c (dbxout_parms): Require fixed-size modes.
2991 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2992 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2993 * omp-low.c (lower_oacc_reductions): Likewise.
2994 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2995 (simplify_subreg): Update accordingly.
2996 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2997 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2998 that aren't fixed-size.
2999 (get_pool_mode): Return a fixed_size_mode.
3000 (output_constant_pool_2): Take a fixed_size_mode.
3002 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3003 Alan Hayward <alan.hayward@arm.com>
3004 David Sherwood <david.sherwood@arm.com>
3006 * doc/rtl.texi (vec_series): Document.
3007 (const): Say that the operand can be a vec_series.
3008 * rtl.def (VEC_SERIES): New rtx code.
3009 * rtl.h (const_vec_series_p_1): Declare.
3010 (const_vec_series_p): New function.
3011 * emit-rtl.h (gen_const_vec_series): Declare.
3012 (gen_vec_series): Likewise.
3013 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3014 (gen_vec_series): Likewise.
3015 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3016 * simplify-rtx.c (simplify_unary_operation): Handle negations
3018 (simplify_binary_operation_series): New function.
3019 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3020 (test_vector_ops_series): New function.
3021 (test_vector_ops): Call it.
3022 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3023 gen_const_vec_series.
3024 (altivec_lvsr): Likewise.
3025 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3027 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3028 Alan Hayward <alan.hayward@arm.com>
3029 David Sherwood <david.sherwood@arm.com>
3031 * doc/rtl.texi (const): Update description of address constants.
3032 Say that vector constants are allowed too.
3033 * common.md (E, F): Use CONSTANT_P instead of checking for
3035 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3036 checking for CONST_VECTOR.
3037 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3039 * expr.c (expand_expr_real_2): Check for vector modes instead
3040 of checking for CONST_VECTOR.
3041 * rtl.h (const_vec_p): New function.
3042 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3043 (unwrap_const_vec_duplicate): Handle them here too.
3045 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3046 David Malcolm <dmalcolm@redhat.com>
3047 Alan Hayward <alan.hayward@arm.com>
3048 David Sherwood <david.sherwood@arm.com>
3050 * rtl.h (vec_duplicate_p): New function.
3051 * selftest-rtl.c (assert_rtx_eq_at): New function.
3052 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3053 (assert_rtx_eq_at): Declare.
3054 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3055 * selftest-run-tests.c (selftest::run_tests): Call it.
3056 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3057 (simplify_unary_operation_1): Recursively handle vector duplicates.
3058 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3060 (simplify_subreg): Handle subregs of vector duplicates.
3061 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3062 (selftest::simplify_rtx_c_tests): New functions.
3064 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3065 Alan Hayward <alan.hayward@arm.com>
3066 David Sherwood <david.sherwood@arm.com>
3068 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3069 (gen_vec_duplicate): Likewise.
3070 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3072 (gen_const_vector): ...here.
3073 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3074 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3075 whose elements are all equal.
3076 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3077 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3078 (simplify_relational_operation): Likewise.
3079 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3081 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3082 (aarch64_expand_vector_init): Likewise.
3083 * config/arm/arm.c (neon_vdup_constant): Likewise.
3084 (neon_expand_vector_init): Likewise.
3085 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3086 (arm_block_set_unaligned_vect): Likewise.
3087 (arm_block_set_aligned_vect): Likewise.
3088 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3089 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3090 (expand_vec_perm_even_odd_pack): Likewise.
3091 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3092 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3093 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3094 gen_const_vec_duplicate.
3095 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3096 * config/mips/mips.c (mips_gen_const_int_vector): Use
3097 gen_const_vec_duplicate.
3098 (mips_expand_vector_init): Use CONST0_RTX.
3099 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3100 (define_split): Use gen_const_vec_duplicate.
3101 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3102 (define_split): Use gen_const_vec_duplicate.
3103 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3104 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3105 * config/spu/spu.c (spu_const): Likewise.
3107 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3108 Alan Hayward <alan.hayward@arm.com>
3109 David Sherwood <david.sherwood@arm.com>
3111 * combine.c (can_change_dest_mode): Reject changes in
3112 REGMODE_NATURAL_SIZE.
3114 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3116 * configure.ac (--enable-libssp): New.
3117 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3118 trying to determine target-specific default. Adjust indentation.
3119 * configure: Regenerated.
3120 * doc/install.texi (Configuration): Expand --disable-libssp
3123 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3125 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3128 2017-10-31 Martin Jambor <mjambor@suse.cz>
3131 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3133 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3135 * auto-profile.c (autofdo_source_profile::read): Use
3136 UNKNOWN_LOCATION rather than 0.
3137 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3138 (warning_at): ...this overload.
3139 (warning_at_rich_loc_n): Rename to...
3140 (warning_n): ...this overload.
3141 (error_at_rich_loc): Rename to...
3142 (error_at): ...this overload.
3143 (pedwarn_at_rich_loc): Rename to...
3144 (pedwarn): ...this overload.
3145 (permerror_at_rich_loc): Rename to...
3146 (permerror): ...this overload.
3147 (inform_at_rich_loc): Rename to...
3148 (inform): ...this overload.
3149 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3150 (diagnostic_n_impl_richloc): Rename to...
3151 (diagnostic_n_impl): ...this rich_location *-based decl.
3152 (inform_at_rich_loc): Rename to...
3153 (inform): ...this, and add an assertion.
3154 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3155 (warning_at_rich_loc): Rename to...
3156 (warning_at): ...this, and add an assertion.
3157 (warning_at_rich_loc_n): Rename to...
3158 (warning_n): ...this, and add an assertion.
3159 (warning_n): Update location_t-based implementation for removal of
3160 location_t-based diagnostic_n_impl.
3161 (pedwarn_at_rich_loc): Rename to...
3162 (pedwarn): ...this, and add an assertion.
3163 (permerror_at_rich_loc): Rename to...
3164 (permerror): ...this, and add an assertion.
3165 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3166 (error_at_rich_loc): Rename to...
3167 (error_at): ...this, and add an assertion.
3168 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3169 (driver::do_spec_on_infiles): Likewise.
3170 * substring-locations.c (format_warning_va): Update for renaming
3171 of inform_at_rich_loc.
3173 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3175 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3176 _Float<N>X built-in functions so that the variant without the
3177 "__builtin_" prefix is only enabled for the GNU C and Objective C
3178 languages when they are in non-strict ANSI/ISO mode.
3179 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3180 * target.def (floatn_builtin_p): Add a target hook to control
3181 whether _Float<N> and _Float<N>X built-in functions without the
3182 "__builtin_" prefix are enabled, and return true for C and
3183 Objective C in the default hook. Include langhooks.h in
3185 * targhooks.h (default_floatn_builtin_p): Likewise.
3186 * targhooks.c (default_floatn_builtin_p): Likewise.
3187 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3188 floatn_builtin_p target hook.
3189 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3191 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3192 Eric Botcazou <ebotcazou@adacore.com>
3194 PR rtl-optimization/81803
3195 * lra-constraints.c (curr_insn_transform): Also reload the whole
3196 register for a strict subreg no wider than a word if this is for
3197 a WORD_REGISTER_OPERATIONS target.
3199 2017-10-31 Jason Merrill <jason@redhat.com>
3201 * gdbinit.in: Skip over inlines from timevar.h.
3203 2017-10-31 Martin Liska <mliska@suse.cz>
3205 * doc/gcov.texi: Document new option.
3206 * gcov.c (print_usage): Likewise print it.
3207 (process_args): Support the argument.
3208 (format_count): New function.
3209 (format_gcov): Use the function.
3211 2017-10-31 Martin Liska <mliska@suse.cz>
3213 * gcov.c (struct name_map): do not use typedef.
3214 Define operator== and operator<.
3215 (name_search): Remove.
3216 (name_sort): Remove.
3217 (main): Do not allocate names.
3218 (process_file): Add vertical space.
3219 (generate_results): Use std::find.
3220 (release_structures): Do not release memory.
3221 (find_source): Use std::find.
3223 2017-10-31 Martin Liska <mliska@suse.cz>
3225 * gcov.c (struct line_info): Remove it's typedef.
3226 (line_info::line_info): Add proper ctor.
3227 (line_info::has_block): Do not use a typedef.
3228 (struct source_info): Do not use typedef.
3229 (circuit): Likewise.
3230 (get_cycles_count): Likewise.
3231 (output_intermediate_file): Iterate via vector iterator.
3232 (add_line_counts): Use std::vector methods.
3233 (accumulate_line_counts): Likewise.
3234 (output_lines): Likewise.
3236 2017-10-31 Martin Liska <mliska@suse.cz>
3238 * gcov.c (struct source_info): Remove typedef.
3239 (source_info::source_info): Add proper ctor.
3240 (accumulate_line_counts): Use struct, not it's typedef.
3241 (output_gcov_file): Likewise.
3242 (output_lines): Likewise.
3243 (main): Do not allocate an array.
3244 (output_intermediate_file): Use size of vector container.
3245 (process_file): Resize the vector.
3246 (generate_results): Do not preallocate, use newly added vector
3248 (release_structures): Do not release sources.
3249 (find_source): Use vector methods.
3250 (add_line_counts): Do not use typedef.
3252 2017-10-31 Martin Liska <mliska@suse.cz>
3254 * doc/gcov.texi: Document that.
3255 * gcov.c (add_line_counts): Mark lines with a non-executed
3257 (output_line_beginning): Handle such lines.
3258 (output_lines): Pass new argument.
3259 (output_intermediate_file): Print it in intermediate format.
3261 2017-10-31 Martin Liska <mliska@suse.cz>
3263 * color-macros.h: New file.
3264 * diagnostic-color.c: Factor out color related to macros to
3266 * doc/gcov.texi: Document -k option.
3267 * gcov.c (INCLUDE_STRING): Include string.h.
3268 (print_usage): Add -k option.
3269 (process_args): Parse it.
3270 (pad_count_string): New function.
3271 (output_line_beginning): Likewise.
3272 (DEFAULT_LINE_START): New macro.
3273 (output_lines): Support color output.
3275 2017-10-31 Martin Liska <mliska@suse.cz>
3277 PR gcov-profile/82633
3278 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3279 their interaction with GCOV infrastructure.
3280 * configure.ac: Add -fkeep-{inline,static}-functions to
3282 * configure: Regenerate.
3284 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3287 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3289 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3292 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3293 into a register if it does not fit into an immediate offset field.
3295 2017-10-31 Olivier Hainque <hainque@adacore.com>
3297 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3299 2017-10-31 Julia Koval <julia.koval@intel.com>
3301 * config.gcc: Add gfniintrin.h.
3302 * config/i386/gfniintrin.h: New.
3303 * config/i386/i386-builtin-types.def
3304 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3305 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3306 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3307 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3308 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3309 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3310 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3311 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3312 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3313 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3314 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3315 * config/i386/immintrin.h: Include gfniintrin.h.
3316 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3318 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3320 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3322 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3324 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3325 (arm_ashldi3_1bit): Remove pattern.
3326 (ashrdi3): Remove shift by 1 expansion.
3327 (arm_ashrdi3_1bit): Remove pattern.
3328 (lshrdi3): Remove shift by 1 expansion.
3329 (arm_lshrdi3_1bit): Remove pattern.
3330 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3331 cost of ashldi3 by 1.
3332 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3333 (<shift>di3_neon): Likewise.
3335 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3337 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3338 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3339 and (*aarch64_simd_mov<VQ:mode>).
3340 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3341 pattern alternative.
3342 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3343 attributes to match pattern alternative.
3345 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3347 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3348 * config/rs6000/emmintrin.h: New file.
3349 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3351 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3353 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3354 * (movdi_vfp_cortexa8): Remove pattern.
3356 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3358 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3360 (Specific, alpha*-dec-osf5.1): Remove.
3361 (Specific, mips-sgi-irix5): Remove.
3362 (Specific, mips-sgi-irix6): Remove.
3364 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3367 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3368 arguments to clear_bit_region_be.
3370 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3372 * gimplify.c: Include memmodel.h.
3374 2017-10-30 Martin Jambor <mjambor@suse.cz>
3376 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3377 condition whether loop should be executed at all.
3379 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3381 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3382 gimple folding of vec_madd() intrinsics.
3383 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3384 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3385 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3387 2017-10-30 Richard Biener <rguenther@suse.de>
3389 PR tree-optimization/82762
3391 2017-10-23 Richard Biener <rguenther@suse.de>
3393 PR tree-optimization/82129
3395 2017-08-01 Richard Biener <rguenther@suse.de>
3397 PR tree-optimization/81181
3398 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3399 (compute_antic): ... end of iteration here.
3401 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3403 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3404 -std=iso9899:2017 and -std=gnu17.
3405 * doc/standards.texi (C Language): Document C17 support.
3406 * doc/cpp.texi (Overview): Mention -std=c17.
3407 (Standard Predefined Macros): Document C11 and C17 values of
3408 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3409 * doc/extend.texi (Inline): Do not list individual options for
3410 standards newer than C99.
3411 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3413 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3416 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3418 * asan.c (asan_finish_file): Align asan globals array by shadow
3421 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3424 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3425 (struct store_immediate_info): Add bitregion_start and bitregion_end
3427 (store_immediate_info::store_immediate_info): Add brs and bre
3428 arguments and initialize bitregion_{start,end} from those.
3429 (struct merged_store_group): Add bitregion_start, bitregion_end,
3430 align_base and mask fields. Drop unnecessary struct keyword from
3431 struct store_immediate_info. Add do_merge method.
3432 (clear_bit_region_be): Use memset instead of loop storing zeros.
3433 (merged_store_group::do_merge): New method.
3434 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3435 stores as long as the surrounding bitregions have no gaps.
3436 (merged_store_group::merge_overlapping): Use do_merge.
3437 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3438 is byte aligned, rather than requiring that start and width are
3439 byte aligned. Drop unnecessary struct keyword from
3440 struct store_immediate_info. Allocate and populate also mask array.
3441 Make start of the arrays relative to bitregion_start rather than
3442 start and size them according to bitregion_{end,start} difference.
3443 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3444 struct store_immediate_info.
3445 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3446 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3447 struct keyword from struct store_immediate_info.
3448 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3449 between stores as long as the surrounding bitregions have no gaps.
3451 (struct split_store): Add orig non-static data member.
3452 (split_store::split_store): Initialize orig to false.
3453 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3454 if there is exactly a single original stmt. Change stmts argument
3455 to pointer from reference, if NULL, don't push anything to it. Add
3456 first argument, use it to optimize skipping over orig stmts that
3457 are known to be before bitpos already. Simplify.
3458 (split_group): Return unsigned int count how many stores are or
3459 would be needed rather than a bool. Add allow_unaligned argument.
3460 Change split_stores argument from reference to pointer, if NULL,
3461 only do a dry run computing how many stores would be produced.
3462 Rewritten algorithm to use both alignment and misalign if
3463 !allow_unaligned and handle bitfield stores with gaps.
3464 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3465 from bitregion_start instead of start. Compute allow_unaligned
3466 here, if true, do 2 split_group dry runs to compute which one
3467 produces fewer stores and prefer aligned if equal. Punt if
3468 new count is bigger or equal than original before emitting any
3469 statements, rather than during that. Remove no longer needed
3470 new_ssa_names tracking. Replace num_stmts with
3471 split_stores.length (). Use 32-bit stack allocated entries
3472 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3473 if possible. Handle bitfields with gaps.
3474 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3475 Compute bitregion_{start,end} for the stores and construct
3476 store_immediate_info with that. Formatting fixes.
3478 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3481 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3482 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3484 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3486 * gimplify.c: Include tm_p.h.
3488 * common.opt (gcoff): Re-add as ignored option.
3489 (gcoff1, gcoff2, gcoff3): Likewise.
3491 * Makefile.in (OBJS): Delete sdbout.o.
3492 (GTFILES): Delete $(srcdir)/sdbout.c.
3493 * debug.h: Delete sdb_debug_hooks.
3494 * final.c: Delete sdbout.h include.
3495 (final_scan_insn): Delete SDB_DEBUG check.
3496 (rest_of_clean_state): Likewise.
3497 * output.h: Delete sdb_begin_function_line.
3500 * toplev.c: Delete sdbout.h include.
3501 (process_options): Delete SDB_DEBUG check.
3502 * tree-core.h (tree_type_common): Delete pointer field of
3504 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3505 TYPE_SYMTAB_POINTER.
3506 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3507 (TYPE_SYMTAB_IS_POINTER): Delete.
3508 (TYPE_SYMTAB_IS_DIE): Renumber.
3509 * xcoffout.c: Refer to former sdbout.c file.
3510 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3512 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3513 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3514 (Debugging Options): Delete -gcoff.
3515 (-gstabs): Delete SDB reference.
3517 (-gcoff@var{level}): Delete.
3518 * doc/passes.texi (Debugging information output): Delete SDB and
3519 sdbout.c references.
3520 * doc/tm.texi: Regenerate.
3521 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3522 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3524 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3525 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3526 (SDB_DEBUGGING_INFO): Delete.
3527 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3528 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3529 * target.def (output_source_filename): Delete COFF reference.
3531 * common.opt (gcoff): Delete.
3532 (gxcoff+): Update Negative chain.
3533 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3535 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3536 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3537 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3539 (expand_function_start): Change sdb reference to past tense.
3540 (expand_function_end): Change sdb reference to past tense.
3541 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3542 * opts.c (debug_type_names): Delete coff entry.
3543 (common_handle_option): Delete OPT_gcoff case.
3544 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3546 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3547 * config/cris/cris.h: Delete SDB reference in comment.
3548 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3549 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3550 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3551 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3553 (ix86_expand_prologue): Likewise.
3554 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3555 * config/ia64/ia64.h: Likewise.
3556 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3557 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3559 * config/mmix/mmix.h: Likewise.
3560 * config/nds32/nds32.c: Likewise.
3561 * config/stormy/storym16.h: Likewise.
3562 * config/visium/visium.h: Likewise.
3563 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3565 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3567 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3568 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3569 FRAME_POINTER_REGNUM point at high end of local var area.
3571 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3573 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3574 Move comment around. Do not reset best_edge for a copiable
3575 destination if the copy would cause a partition change.
3576 (better_edge_p): Remove redundant check.
3578 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3580 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3582 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3584 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3585 for math functions that have _Float<N> and _Float<N>X variants.
3586 (mathfn_built_in_2): Add support for math functions that have
3587 _Float<N> and _Float<N>X variants.
3588 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3589 (expand_builtin_mathfn_ternary): Add support for fma with
3590 _Float<N> and _Float<N>X variants.
3591 (expand_builtin): Likewise.
3592 (fold_builtin_3): Likewise.
3593 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3594 create math function _Float<N> and _Float<N>X variants as external
3596 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3597 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3598 the __builtin_ prefix and if not strict ansi, without the prefix.
3599 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3600 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3601 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3602 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3603 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3604 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3605 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3606 function signatures for fma _Float<N> and _Float<N>X variants.
3607 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3608 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3609 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3610 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3611 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3612 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3613 * gencfn-macros.c (print_case_cfn): Add support for math functions
3614 that have _Float<N> and _Float<N>X variants.
3615 (print_define_operator_list): Likewise.
3616 (fltfn_suffixes): Likewise.
3618 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3619 for math functions that have _Float<N> and _Float<N>X variants.
3620 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3621 and _Float<N>X variants.
3622 (COPYSIGN): Likewise.
3625 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3626 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3628 (integer_valued_read_call_p): Likewise.
3629 * fold-const-call.c (fold_const_call_ss): Likewise.
3630 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3631 _Float<N> and _Float<N>X variants.
3632 (fold_const_call_ssss): Add support for fma _Float<N> and
3633 _Float<N>X variants.
3634 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3635 support for copysign and fma _Float<N> and _Float<N>X variants.
3636 (backprop::process_builtin_call_use): Likewise.
3637 * tree-call-cdce.c (can_test_argument_range); Add support for
3638 sqrt _Float<N> and _Float<N>X variants.
3639 (edom_only_function): Likewise.
3640 (get_no_error_domain): Likewise.
3641 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3642 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3643 copysign _Float<N> and _Float<N>X variants.
3644 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3645 handled by machine independent code.
3646 (FMAF128): Likewise.
3647 * doc/cpp.texi (Common Predefined Macros): Document defining
3648 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3649 fma _Float<N> and _Float<N>X variants.
3651 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3654 * config/i386/i386-modes.def (CCFPU): Remove definition.
3655 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3656 (ix86_cc_modes_compatible): Ditto.
3657 (ix86_expand_carry_flag_compare): Ditto.
3658 (ix86_expand_int_movcc): Ditto.
3659 (ix86_expand_int_addcc): Ditto.
3660 (ix86_reverse_condition): Ditto.
3661 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3662 Return true/false for unordered/ordered fp comparisons.
3663 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3664 (ix86_prepare_fp_compare_args): Update for rename.
3665 (ix86_expand_fp_compare): Update for rename. Generate unordered
3666 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3667 (ix86_expand_sse_compare_and_jump): Ditto.
3668 * config/i386/predicates.md (fcmov_comparison_operator):
3669 Remove CCFPU mode handling.
3670 (ix86_comparison_operator): Ditto.
3671 (ix86_carry_flag_operator): Ditto.
3672 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3673 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3674 (*cmpu<mode>_cc_i387): Ditto.
3675 (FPCMP): Remove mode iterator.
3676 (unord): Remove mode attribute.
3677 (unord_subst): New define_subst transformation
3678 (unord): New define_subst attribute.
3680 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3681 (*cmpi<unord>xf_i387): Ditto.
3682 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3683 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3684 using unord_subst transformation.
3685 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3686 (round_saeonly): Also handle CCFP mode.
3687 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3688 Remove UNSPEC_SAHF unspec handling.
3690 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3692 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3694 2017-10-27 Jeff Law <law@redhat.com>
3696 * gimple-ssa-sprintf.c: Include domwalk.h.
3697 (class sprintf_dom_walker): New class, derived from dom_walker.
3698 (sprintf_dom_walker::before_dom_children): New function.
3699 (struct call_info): Moved into sprintf_dom_walker class
3700 (compute_formath_length, handle_gimple_call): Likewise.
3701 (sprintf_length::execute): Call the dominator walker rather
3702 than walking the statements.
3704 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3705 gimple statement locations.
3706 (check_array_bounds): Corresponding changes. Get the statement's
3707 location directly from wi->stmt.
3709 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3712 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3714 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3716 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3717 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3719 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3722 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3723 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3724 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3725 maybe_get_pool_constant.
3726 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3729 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3731 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3732 2.26 caveat. Update gas and gld versions.
3733 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3736 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3738 * cgraph.h (set_malloc_flag): Declare.
3739 * cgraph.c (set_malloc_flag_1): New function.
3740 (set_malloc_flag): Likewise.
3741 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3742 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3744 (read_ipa_call_summary): Add support for reading is_return_callee.
3745 (write_ipa_call_summary): Stream is_return_callee.
3746 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3747 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3748 ipa-prop.h, ipa-fnsummary.h.
3749 (pure_const_names): Change to static.
3750 (malloc_state_e): Define.
3751 (malloc_state_names): Define.
3752 (funct_state_d): Add field malloc_state.
3753 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3754 (check_retval_uses): New function.
3755 (malloc_candidate_p): Likewise.
3756 (analyze_function): Add support for malloc attribute.
3757 (pure_const_write_summary): Stream malloc_state.
3758 (pure_const_read_summary): Add support for reading malloc_state.
3759 (dump_malloc_lattice): New function.
3760 (propagate_malloc): New function.
3761 (warn_function_malloc): New function.
3762 (ipa_pure_const::execute): Call propagate_malloc and
3763 ipa_free_fn_summary.
3764 (pass_local_pure_const::execute): Add support for malloc attribute.
3765 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3766 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3768 2017-10-27 Martin Liska <mliska@suse.cz>
3770 PR gcov-profile/82457
3771 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3772 for fork-like functions to be properly instrumented.
3774 2017-10-27 Richard Biener <rguenther@suse.de>
3777 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3778 info when we redirected EH.
3780 2017-10-26 Michael Collison <michael.collison@arm.com>
3782 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3784 (<optab>_trunchf<GPI:mode>2: New pattern.
3785 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3786 * config/aarch64/iterators.md (wv): New mode attribute.
3787 (vf, VF): New mode attributes.
3788 (vgp, VGP): New mode attributes.
3789 (s): Update attribute with SImode and DImode prefixes.
3791 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3793 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3794 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3795 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3796 (nios2_option_overide): Initialize it. Don't allow R0-relative
3797 addressing with PIC.
3798 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3799 (nios2_symbolic_constant_p): Likewise.
3800 (nios2_legitimate_address_p): Likewise.
3801 (nios2_r0rel_section_name_p): New.
3802 (nios2_symbol_ref_in_r0rel_data_p): New.
3803 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3804 (r0rel_constant_p): New.
3805 (nios2_print_operand_address): Handle r0rel_constant_p.
3806 (nios2_cdx_narrow_form_p): Likewise.
3807 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3808 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3809 (Nios II Options): Document -mr0rel-sec.
3811 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3813 * config/nios2/nios2.c: Include xregex.h.
3814 (nios2_gprel_sec_regex): New.
3815 (nios2_option_overide): Initialize it. Don't allow GP-relative
3816 addressing with PIC.
3817 (nios2_small_section_name_p): Check for regex match.
3818 * config/nios2/nios2.opt (mgprel-sec=): New option.
3819 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3820 (Nios II Options): Document -mgprel-sec.
3822 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3824 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3826 2017-10-26 Tom de Vries <tom@codesourcery.com>
3828 PR tree-optimization/82707
3829 * gimple.c (gimple_copy): Fix unsharing of
3830 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3832 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3834 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3835 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3836 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3837 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3838 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3839 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3840 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3841 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3842 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3843 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3844 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3845 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3846 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3847 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3848 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3849 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3850 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3852 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3854 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3856 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3857 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3858 warning to rs6000.c. Remove the Undocumented flag, since it has
3860 (-mabi=ibmlongdouble): Likewise.
3861 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3862 already set, set the default format for long double.
3863 (rs6000_debug_reg_global): Print whether long double is IBM or
3865 (rs6000_option_override_internal): Rework setting long double
3866 format. Only warn if the user is changing the long double default
3867 and they did not use -Wno-psabi.
3868 * doc/invoke.texi (PowerPC options): Update the documentation for
3869 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3871 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3872 Alan Hayward <alan.hayward@arm.com>
3873 David Sherwood <david.sherwood@arm.com>
3875 * rtl.h (wider_subreg_mode): New function.
3876 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3877 rather than an unsigned int *.
3878 * ira-color.c (regno_max_ref_width): Replace with...
3879 (regno_max_ref_mode): ...this new variable.
3880 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3881 Use wider_subreg_mode.
3882 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3883 rather than an unsigned int *.
3884 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3885 (process_alt_operands): Likewise.
3886 (invariant_p): Likewise.
3887 * lra-spills.c (assign_mem_slot): Likewise.
3888 (add_pseudo_to_slot): Likewise.
3889 * lra.c (collect_non_operand_hard_regs): Likewise.
3890 (add_regs_to_insn_regno_info): Likewise.
3891 * reload1.c (regno_max_ref_width): Replace with...
3892 (regno_max_ref_mode): ...this new variable.
3893 (reload): Update accordingly. Update call to
3894 ira_sort_regnos_for_alter_reg.
3895 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3896 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3897 (scan_paradoxical_subregs): Likewise.
3899 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3901 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3902 (aarch64_frame): Add emit_frame_chain boolean.
3903 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3904 Move eh_return case to aarch64_layout_frame.
3905 (aarch64_layout_frame): Initialize emit_frame_chain.
3906 (aarch64_expand_prologue): Use emit_frame_chain.
3908 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3910 * config/aarch64/aarch64.c (aarch64_layout_frame):
3911 Ensure LR is always stored at the bottom of the callee-saves.
3912 Remove rarely used frame layout which saves callee-saves at top of
3913 frame, so the store of LR can be used as a valid probe in all cases.
3915 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3917 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3918 Improve unaligned TImode/TFmode base/offset split.
3920 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3921 Alan Hayward <alan.hayward@arm.com>
3922 David Sherwood <david.sherwood@arm.com>
3924 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3925 * combine.c (find_single_use_1): Likewise.
3926 (expand_field_assignment): Likewise.
3927 (move_deaths): Likewise.
3928 * lra-constraints.c (simplify_operand_subreg): Likewise.
3929 (curr_insn_transform): Likewise.
3930 * lra.c (collect_non_operand_hard_regs): Likewise.
3931 (add_regs_to_insn_regno_info): Likewise.
3932 * rtlanal.c (reg_referenced_p): Likewise.
3933 (covers_regno_no_parallel_p): Likewise.
3935 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3937 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3938 Don't print any bits outside the precision of the value.
3939 * wide-int.cc (test_printing): Add some new tests.
3941 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3943 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3944 supports -xbrace_comment option.
3945 * configure: Regenerate.
3946 * config.in: Regenerate.
3947 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3948 (ASM_CPU_SPEC): Use it.
3950 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3952 * target.def (static_rtx_alignment): New hook.
3953 * targhooks.h (default_static_rtx_alignment): Declare.
3954 * targhooks.c (default_static_rtx_alignment): New function.
3955 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3956 * doc/tm.texi: Regenerate.
3957 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3958 instead of targetm.constant_alignment. Remove call to
3960 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3961 (cris_preferred_mininum_alignment): New function, split out from...
3962 (cris_constant_alignment): ...here.
3963 (cris_static_rtx_alignment): New function.
3964 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3966 (ix86_constant_alignment): ...here.
3967 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3968 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3969 (mmix_static_rtx_alignment): New function.
3970 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3971 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3973 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3976 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3977 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3979 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3981 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3982 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3983 mark as a sign-extending load.
3984 (local_pic_load_u): Define.
3986 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3989 * fold-const.c (operand_equal_for_comparison_p): Also return true
3990 if ARG0 is a simple variant of ARG1 with narrower precision.
3991 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3993 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3995 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3997 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3998 scatter_static, scatter_per_elt.
3999 * x86-tune-costs.h: Add new cost entries.
4001 2017-10-25 Richard Biener <rguenther@suse.de>
4003 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4004 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4005 class pass_fre): Move to ...
4006 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4007 class pass_fre): ... here and adjust for statistics.
4009 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4012 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4013 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4014 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4015 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4018 2017-10-25 Richard Biener <rguenther@suse.de>
4020 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4021 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4022 eliminate_push_avail, eliminate_insert): Move inside...
4023 (class eliminate_dom_walker): ... this class in preparation
4025 (fini_eliminate): Remove by merging with ...
4026 (eliminate): ... this function. Adjust for class changes.
4027 (pass_pre::execute): Remove fini_eliminate call.
4028 (pass_fre::execute): Likewise.
4030 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4033 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4034 (VPERMI2, VPERMI2I): New mode iterators.
4035 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4036 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4038 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4039 mode iterator. Remove 3 old define_insn patterns.
4040 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4041 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4042 VPERMI2 mode iterator, remove the other two expanders.
4043 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4044 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4045 instructions, remove the other two patterns.
4046 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4047 mode iterator, remove the other two patterns.
4048 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4049 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4050 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4051 and adjust argument order accordingly.
4052 (ix86_expand_vec_perm): Adjust caller.
4053 (expand_vec_perm_1): Likewise.
4054 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4055 (expand_vec_perm_vpermt2_vpshub2): ... this.
4056 (ix86_expand_vec_perm_const_1): Adjust caller.
4057 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4060 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4061 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4062 (vec_shl_<mode>): Remove unused expander.
4063 (avx512bw_<shift_insn><mode>3): New define_insn.
4064 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4065 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4067 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4070 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4073 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4075 PR rtl-optimization/82396
4076 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4077 (autopref_multipass_init): Simplify initialization.
4078 (autopref_rank_data): Simplify sort order.
4079 * gcc/sched-int.h (autopref_multipass_data_): Remove
4080 multi_mem_insn_p, min_offset and max_offset.
4082 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4085 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4086 Check special value of flag_omit_frame_pointer.
4087 (aarch64_can_eliminate): Likewise.
4088 (aarch64_override_options_after_change_1): Simplify handling of
4089 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4091 2017-10-24 Richard Biener <rguenther@suse.de>
4093 PR tree-optimization/82697
4094 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4095 zero for conditional load and unconditional store.
4097 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4099 * doc/install.texi: Document bootstrap-cet.
4101 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4104 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4105 ENDBR instruction at function entrance if function is only
4108 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4111 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4112 patterns to better describe from which operation the CF is computed.
4113 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4114 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4115 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4116 is 0, use _0 suffixed expanders instead of emitting a comparison
4119 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4121 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4122 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4124 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4127 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4128 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4129 * loop-iv.c (iv_get_reaching_def): Likewise.
4130 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4131 variable is promoted and the partition contains undefined values.
4133 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4135 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4137 (nios2_address_cost): Define.
4138 (nios2_legitimize_address): Recognize (exp + constant) directly.
4139 (TARGET_ADDRESS_COST): Define.
4141 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4143 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4144 (nios2_symbolic_memory_operand_p): Declare.
4145 (nios2_split_large_constant): Declare.
4146 (nios2_split_symbolic_memory_operand): Declare.
4147 * config/nios2/nios2.c: Adjust includes.
4148 (nios2_symbolic_constant_allowed): New.
4149 (nios2_symbolic_constant_p): New.
4150 (nios2_plus_symbolic_constant_p): New.
4151 (nios2_valid_addr_expr_p): Recognize addresses involving
4153 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4154 (nios2_symbolic_memory_operand_p): New.
4155 (nios2_large_constant_p): New.
4156 (nios2_split_large_constant): New.
4157 (nios2_split_plus_large_constant): New.
4158 (nios2_split_symbolic_memory_operand): New.
4159 (nios2_legitimize_address): Code refactoring. Handle addresses
4160 involving symbolic constants.
4161 (nios2_emit_move_sequence): Likewise.
4162 (nios2_print_operand): Improve error output.
4163 (nios2_print_operand_address): Handle LO_SUM.
4164 (nios2_cdx_narrow_form_p): Likewise.
4165 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4166 operands involving symbolic constants.
4167 (movhi_internal, movsi_internal): Likewise.
4168 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4169 (extendhisi2, extendqi<mode>2): Likewise.
4171 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4173 * tree-pass.h (PROP_rtl_split_insns): Define.
4174 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4176 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4178 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4180 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4183 * target.def (const_not_ok_for_debug_p): Default to
4184 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4185 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4186 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4187 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4188 which targetm.const_not_ok_for_debug_p returned true.
4189 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4191 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4193 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4194 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4195 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4197 (ix86_const_not_ok_for_debug_p): New function.
4198 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4199 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4201 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4204 * system.h: Conditionally include "unique-ptr.h" if
4205 INCLUDE_UNIQUE_PTR is defined.
4206 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4207 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4209 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4211 * config/rl78/rl78.md: New define_expand "subdi3".
4213 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4216 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4217 DF_REF_INSN if DF_REF_INSN_INFO is false.
4219 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4221 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4222 xmm_move instead of sse_move.
4223 (sse_store_index): New function.
4224 (ix86_register_move_cost): Be more sensible about mismatch stall;
4225 model AVX moves correctly; make difference between sse->integer and
4227 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4228 moves; make difference between SSE and AVX.
4229 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4230 and zmm_move. Increase size of sse load and store tables;
4231 add unaligned load and store tables; add ssemmx_to_integer.
4232 * x86-tune-costs.h: Update all entries according to real
4233 move latencies from Agner Fog's manual and chip documentation.
4235 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4238 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4239 * config/i386/constraints.md (Wf): New constraint.
4240 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4241 (cmp<dwi>_doubleword): Removed.
4242 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4243 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4244 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4245 expand with cmp<dwi>_doubleword. For LTU and GEU use
4246 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4248 * common.opt (gcolumn-info): Enable by default.
4249 * doc/invoke.texi (gcolumn-info): Document new default.
4251 2017-10-23 Richard Biener <rguenther@suse.de>
4253 PR tree-optimization/82672
4254 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4255 Fold the stmt if we propagated into it.
4257 2017-10-23 Richard Biener <rguenther@suse.de>
4259 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4260 (bitmap_remove_expr_from_set): ... this. All callers call this
4261 for non-constant values.
4262 (bitmap_set_subtract): Rename to...
4263 (bitmap_set_subtract_expressions): ... this. Adjust and
4265 (bitmap_set_contains_value): Remove superfluous check.
4266 (bitmap_set_replace_value): Inline into single caller ...
4267 (bitmap_value_replace_in_set): ... here and simplify.
4268 (dependent_clean): Merge into ...
4269 (clean): ... this using an overload. Adjust.
4270 (prune_clobbered_mems): Adjust.
4271 (compute_antic_aux): Likewise.
4272 (compute_partial_antic_aux): Likewise.
4274 2017-10-23 Richard Biener <rguenther@suse.de>
4276 PR tree-optimization/82129
4278 2017-08-01 Richard Biener <rguenther@suse.de>
4280 PR tree-optimization/81181
4281 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4282 (compute_antic): ... end of iteration here.
4284 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4286 * target.def (starting_frame_offset): New hook.
4287 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4288 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4289 * doc/tm.texi.in: Regenerate.
4290 * hooks.h (hook_hwi_void_0): Declare.
4291 * hooks.c (hook_hwi_void_0): New function.
4292 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4293 STARTING_FRAME_OFFSET.
4294 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4295 * reload1.c (reload): Likewise.
4296 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4297 instead of STARTING_FRAME_OFFSET.
4298 * function.c (try_fit_stack_local): Likewise.
4299 (assign_stack_local_1): Likewise
4300 (instantiate_virtual_regs): Likewise.
4301 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4302 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4303 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4304 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4305 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4306 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4307 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4308 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4309 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4310 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4311 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4312 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4313 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4314 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4315 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4316 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4317 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4318 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4319 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4320 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4321 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4322 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4323 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4324 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4325 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4326 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4327 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4328 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4329 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4330 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4331 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4332 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4333 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4334 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4335 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4336 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4337 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4338 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4339 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4340 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4341 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4342 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4343 return a HOST_WIDE_INT.
4344 (avr_builtin_setjmp_frame_value): Use it instead of
4345 STARTING_FRAME_OFFSET.
4346 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4347 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4348 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4350 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4351 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4352 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4353 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4354 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4355 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4356 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4357 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4358 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4359 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4360 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4361 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4363 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4364 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4365 * config/mips/mips.c (mips_compute_frame_info): Refer to
4366 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4367 (mips_starting_frame_offset): New function.
4368 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4369 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4370 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4371 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4372 and return a HOST_WIDE_INT.
4373 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4374 (mmix_initial_elimination_offset): Refer to
4375 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4376 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4377 * config/pa/pa.c (pa_starting_frame_offset): New function.
4378 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4379 (pa_expand_prologue): Likewise.
4380 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4381 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4382 !FRAME_GROWS_DOWNWARD handling to...
4383 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4384 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4385 !FRAME_GROWS_DOWNWARD handling to...
4386 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4387 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4388 !FRAME_GROWS_DOWNWARD handling to...
4389 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4390 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4392 (rs6000_starting_frame_offset): New function.
4393 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4394 !FRAME_GROWS_DOWNWARD handling to...
4395 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4396 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4397 !FRAME_GROWS_DOWNWARD handling to...
4398 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4399 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4400 !FRAME_GROWS_DOWNWARD handling to...
4401 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4402 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4403 (rs6000_starting_frame_offset): New function.
4404 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4405 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4406 * config/vax/vax.c (vax_starting_frame_offset): New function.
4407 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4408 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4409 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4410 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4411 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4412 * system.h (STARTING_FRAME_OFFSET): Poison.
4414 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4416 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4417 SCALAR_TYPE_MODE instead of TYPE_MODE.
4419 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4420 Alan Hayward <alan.hayward@arm.com>
4421 David Sherwood <david.sherwood@arm.com>
4423 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4425 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4426 Alan Hayward <alan.hayward@arm.com>
4427 David Sherwood <david.sherwood@arm.com>
4429 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4431 2017-10-23 Richard Biener <rguenther@suse.de>
4433 PR tree-optimization/82129
4434 * tree-ssa-pre.c (bitmap_set_and): Remove.
4435 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4436 canonicalizing expressions in the set to those with lowest
4437 ID rather than taking that from the first edge.
4439 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4441 * combine.c (rtx_equal_for_field_assignment_p): Use
4442 byte_lowpart_offset.
4444 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4445 Alan Hayward <alan.hayward@arm.com>
4446 David Sherwood <david.sherwood@arm.com>
4448 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4449 to a SUBREG_PROMOTED_VAR.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4456 (expand_debug_source_expr): Likewise.
4457 * combine.c (combine_simplify_rtx): Likewise.
4458 * cse.c (fold_rtx): Likewise.
4459 * optabs.c (expand_float): Likewise.
4460 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4461 (simplify_binary_operation_1): Likewise.
4463 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4464 Alan Hayward <alan.hayward@arm.com>
4465 David Sherwood <david.sherwood@arm.com>
4467 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4468 (record_promoted_value): Likewise.
4469 * expr.c (expand_expr_real_2): Likewise.
4470 * ree.c (update_reg_equal_equiv_notes): Likewise.
4471 (combine_set_extension): Likewise.
4472 * rtlanal.c (low_bitmask_len): Likewise.
4473 * simplify-rtx.c (neg_const_int): Likewise.
4474 (simplify_binary_operation_1): Likewise.
4476 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4477 Alan Hayward <alan.hayward@arm.com>
4478 David Sherwood <david.sherwood@arm.com>
4480 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4481 * regcprop.c (maybe_mode_change): Likewise.
4482 * reload1.c (alter_reg): Likewise.
4484 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4486 * inchash.h (inchash::hash::add_wide_int): New function.
4487 * lto-streamer-out.c (hash_tree): Use it.
4489 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4491 * inchash.h (inchash::hash::add_wide_int): Rename to...
4492 (inchash::hash::add_hwi): ...this.
4493 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4494 (polymorphic_call_target_hasher::hash): Likewise.
4495 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4496 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4497 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4498 * lto-streamer-out.c (hash_tree): Likewise.
4499 * optc-save-gen.awk: Likewise.
4500 * tree.c (add_expr): Likewise.
4502 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4505 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4506 for ordered inequality comparisons even with TARGET_IEEE_FP.
4508 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4511 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4512 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4513 Expand with cmp<dwi>_doubleword.
4515 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4517 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4518 List CET intrinsics.
4519 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4520 specific to -fcf-protection option.
4522 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4524 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4525 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4526 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4527 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4528 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4529 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4530 (extra_objs): Add cet.o for Linux/x86 targets.
4531 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4532 * config/i386/cet.c: New file.
4533 * config/i386/cetintrin.h: Likewise.
4534 * config/i386/t-cet: Likewise.
4535 * config/i386/cpuid.h (bit_SHSTK): New.
4536 (bit_IBT): Likewise.
4537 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4538 pass IBT and SHSTK bits.
4539 * config/i386/i386-builtin-types.def
4540 (VOID_FTYPE_UNSIGNED_PVOID): New.
4541 (VOID_FTYPE_UINT64_PVOID): Likewise.
4542 * config/i386/i386-builtin.def: Add CET intrinsics.
4543 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4544 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4545 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4546 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4548 * config/i386/i386.c (rest_of_insert_endbranch): New.
4549 (pass_data_insert_endbranch): Likewise.
4550 (pass_insert_endbranch): Likewise.
4551 (make_pass_insert_endbranch): Likewise.
4552 (ix86_notrack_prefixed_insn_p): Likewise.
4553 (ix86_target_string): Add -mibt, -mshstk flags.
4554 (ix86_option_override_internal): Add flag_cf_protection
4556 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4557 (ix86_print_operand): Add 'notrack' prefix output.
4558 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4559 (ix86_expand_builtin): Expand CET intrinsics.
4560 (x86_output_mi_thunk): Add 'endbranch' instruction.
4561 * config/i386/i386.h (TARGET_IBT): New.
4562 (TARGET_IBT_P): Likewise.
4563 (TARGET_SHSTK): Likewise.
4564 (TARGET_SHSTK_P): Likewise.
4565 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4566 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4567 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4568 (builtin_setjmp_setup): New pattern.
4569 (builtin_longjmp): Likewise.
4570 (rdssp<mode>): Likewise.
4571 (incssp<mode>): Likewise.
4572 (saveprevssp): Likewise.
4573 (rstorssp): Likewise.
4574 (wrss<mode>): Likewise.
4575 (wruss<mode>): Likewise.
4576 (setssbsy): Likewise.
4577 (clrssbsy): Likewise.
4578 (nop_endbr): Likewise.
4579 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4581 * config/i386/immintrin.h: Include <cetintrin.h>.
4582 * config/i386/linux-common.h
4583 (file_end_indicate_exec_stack_and_cet): New prototype.
4584 (TARGET_ASM_FILE_END): New.
4586 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4588 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4589 latencies instead of having separate table; make difference between
4590 integer and float costs.
4591 * i386.h (processor_costs): Remove scalar_stmt_cost,
4592 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4593 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4595 * x86-tune-costs.h: Remove entries which has been removed in
4596 procesor_costs from all tables; make cond_taken_branch_cost
4597 and cond_not_taken_branch_cost COST_N_INSNS based.
4599 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4601 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4603 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4605 * config/i386/i386.md (isa): Remove fma_avx512f.
4606 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4607 <avx512>_fmadd_<mode>_mask3<round_name>,
4608 <avx512>_fmsub_<mode>_mask<round_name>,
4609 <avx512>_fmsub_<mode>_mask3<round_name>,
4610 <avx512>_fnmadd_<mode>_mask<round_name>,
4611 <avx512>_fnmadd_<mode>_mask3<round_name>,
4612 <avx512>_fnmsub_<mode>_mask<round_name>,
4613 <avx512>_fnmsub_<mode>_mask3<round_name>,
4614 <avx512>_fmaddsub_<mode>_mask<round_name>,
4615 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4616 <avx512>_fmsubadd_<mode>_mask<round_name>,
4617 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4618 (*vec_widen_umult_even_v16si<mask_name>,
4619 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4620 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4622 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4624 * extend.texi: Add 'nocf_check' documentation.
4625 * gimple.texi: Add second parameter to
4626 gimple_build_call_from_tree.
4627 * invoke.texi: Add -fcf-protection documentation.
4628 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4630 2017-10-20 Richard Biener <rguenther@suse.de>
4632 PR tree-optimization/82473
4633 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4634 the largest input type.
4636 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4638 * c-attribs.c (handle_nocf_check_attribute): New function.
4639 (c_common_attribute_table): Add 'nocf_check' handling.
4640 * gimple-parser.c: Add second argument NULL to
4641 gimple_build_call_from_tree.
4642 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4643 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4645 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4646 * common.opt: Add fcf-protection flag.
4647 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4648 * flag-types.h: Add enum cf_protection_level.
4649 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4650 Add 'nocf_check' attribute propagation to gimple call.
4651 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4652 (gimple_build_call_from_tree): Update prototype.
4653 (gimple_call_nocf_check_p): New function.
4654 (gimple_call_set_nocf_check): Likewise.
4655 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4656 * ipa-icf.c: Add nocf_check attribute in statement hash.
4657 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4658 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4659 * toplev.c (process_options): Add flag_cf_protection handling.
4661 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4663 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4665 2017-10-20 Richard Biener <rguenther@suse.de>
4667 PR tree-optimization/82603
4668 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4669 remove false predicated stores.
4671 2017-10-20 Richard Biener <rguenther@suse.de>
4673 * graphite-isl-ast-to-gimple.c
4674 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4675 Remove return value and simplify, dump copied stmt after lhs
4677 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4678 Reduce dump verbosity.
4679 (gsi_insert_earliest): Likewise.
4680 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4681 * graphite.c (print_global_statistics): Adjust dumping.
4682 (print_graphite_scop_statistics): Likewise.
4683 (print_graphite_statistics): Do not dump loops here.
4684 (graphite_transform_loops): But here.
4686 2017-10-20 Nicolas Roche <roche@adacore.com>
4688 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4689 * configure: Regenerate.
4691 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4694 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4695 functions when optimizing replace GIMPLE_RETURN stmts with
4696 calls to __builtin_unreachable ().
4699 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4700 for -fsanitize=thread link of executables.
4701 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4702 link of executables.
4705 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4707 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4708 define_insns for logical vector shifts to use VI248_AVX512BW
4709 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4710 condition, useless isa and prefix attributes. Change the first
4711 2 of these define_insns to ...
4712 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4713 define_insn for avx512vl.
4714 (<shift_insn><mode>3): ... and this, new define_insn without
4715 masking for non-avx512vl.
4718 * config/i386/sse.md (*andnot<mode>3,
4719 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4720 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4721 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4722 not applied use empty suffix even for TARGET_AVX512VL.
4723 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4724 is applied, supply evex,evex or evex,evex,evex instead of just
4727 2017-10-20 Julia Koval <julia.koval@intel.com>
4729 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4730 (OPTION_MASK_ISA_GFNI_UNSET): New.
4731 (ix86_handle_option): Handle OPT_mgfni.
4732 * config/i386/cpuid.h (bit_GFNI): New.
4733 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4734 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4735 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4736 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4737 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4738 * config/i386/i386.opt: Add mgfni.
4740 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4742 * config/msp430/msp430.c (msp430_option_override): Disable
4743 -fdelete-null-pointer-checks.
4744 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4746 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4748 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4749 of x87 and SSE instructions.
4751 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4753 * asan.c (create_cond_insert_point): Do not update edge count.
4754 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4755 (afdo_propagate_circuit): Likewise.
4756 (afdo_calculate_branch_prob): Likewise.
4757 (afdo_annotate_cfg): Likewise.
4758 * basic-block.h (struct edge_def): Remove count.
4759 (edge_def::count): New accessor.
4760 * bb-reorder.c (rotate_loop): Update.
4761 (find_traces_1_round): Update.
4762 (connect_traces): Update.
4763 (sanitize_hot_paths): Update.
4764 * cfg.c (unchecked_make_edge): Update.
4765 (make_single_succ_edge): Update.
4766 (check_bb_profile): Update.
4767 (dump_edge_info): Update.
4768 (update_bb_profile_for_threading): Update.
4769 (scale_bbs_frequencies_int): Update.
4770 (scale_bbs_frequencies_gcov_type): Update.
4771 (scale_bbs_frequencies_profile_count): Update.
4772 (scale_bbs_frequencies): Update.
4773 * cfganal.c (connect_infinite_loops_to_exit): Update.
4774 * cfgbuild.c (compute_outgoing_frequencies): Update.
4775 (find_many_sub_basic_blocks): Update.
4776 * cfgcleanup.c (try_forward_edges): Update.
4777 (try_crossjump_to_edge): Update
4778 * cfgexpand.c (expand_gimple_cond): Update
4779 (expand_gimple_tailcall): Update
4780 (construct_exit_block): Update
4781 * cfghooks.c (verify_flow_info): Update
4782 (redirect_edge_succ_nodup): Update
4783 (split_edge): Update
4784 (make_forwarder_block): Update
4785 (duplicate_block): Update
4786 (account_profile_record): Update
4787 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4788 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4789 * cfgloopmanip.c (scale_loop_profile): Update.
4791 (lv_adjust_loop_entry_edge): Update.
4792 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4793 (force_nonfallthru_and_redirect): Update.
4794 (purge_dead_edges): Update.
4795 (rtl_flow_call_edges_add): Update.
4796 * cgraphunit.c (init_lowered_empty_function): Update.
4797 (cgraph_node::expand_thunk): Update.
4798 * gimple-pretty-print.c (dump_probability): Update.
4799 (dump_edge_probability): Update.
4800 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4801 * haifa-sched.c (sched_create_recovery_edges): Update.
4802 * hsa-gen.c (convert_switch_statements): Update.
4803 * ifcvt.c (dead_or_predicable): Update.
4804 * ipa-inline-transform.c (inline_transform): Update.
4805 * ipa-split.c (split_function): Update.
4806 * ipa-utils.c (ipa_merge_profiles): Update.
4807 * loop-doloop.c (add_test): Update.
4808 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4809 * lto-streamer-in.c (input_cfg): Update.
4810 (input_function): Update.
4811 * lto-streamer-out.c (output_cfg): Update.
4812 * modulo-sched.c (sms_schedule): Update.
4813 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4814 * predict.c (maybe_hot_edge_p): Update.
4815 (unlikely_executed_edge_p): Update.
4816 (probably_never_executed_edge_p): Update.
4817 (dump_prediction): Update.
4818 (drop_profile): Update.
4819 (propagate_unlikely_bbs_forward): Update.
4820 (determine_unlikely_bbs): Update.
4821 (force_edge_cold): Update.
4822 * profile.c (compute_branch_probabilities): Update.
4823 * reg-stack.c (better_edge): Update.
4824 * shrink-wrap.c (handle_simple_exit): Update.
4825 * tracer.c (better_p): Update.
4826 * trans-mem.c (expand_transaction): Update.
4827 (split_bb_make_tm_edge): Update.
4828 * tree-call-cdce.c: Update.
4829 * tree-cfg.c (gimple_find_sub_bbs): Update.
4830 (gimple_split_edge): Update.
4831 (gimple_duplicate_sese_region): Update.
4832 (gimple_duplicate_sese_tail): Update.
4833 (gimple_flow_call_edges_add): Update.
4834 (insert_cond_bb): Update.
4835 (execute_fixup_cfg): Update.
4836 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4837 * tree-complex.c (expand_complex_div_wide): Update.
4838 * tree-eh.c (lower_resx): Update.
4839 (unsplit_eh): Update.
4840 (cleanup_empty_eh_move_lp): Update.
4841 * tree-inline.c (copy_edges_for_bb): Update.
4842 (freqs_to_counts): Update.
4843 (copy_cfg_body): Update.
4844 * tree-ssa-dce.c (remove_dead_stmt): Update.
4845 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4846 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4847 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4848 (unloop_loops): Update.
4849 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4850 * tree-ssa-loop-split.c (connect_loops): Update.
4851 (split_loop): Update.
4852 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4853 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4854 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4855 * tree-ssa-reassoc.c (branch_fixup): Update.
4856 * tree-ssa-tail-merge.c (replace_block_by): Update.
4857 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4858 (compute_path_counts): Update.
4859 (update_profile): Update.
4860 (recompute_probabilities): Update.
4861 (update_joiner_offpath_counts): Update.
4862 (estimated_freqs_path): Update.
4863 (freqs_to_counts_path): Update.
4864 (clear_counts_path): Update.
4865 (ssa_fix_duplicate_block_edges): Update.
4866 (duplicate_thread_path): Update.
4867 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4868 (case_bit_test_cmp): Update.
4869 (collect_switch_conv_info): Update.
4870 (gen_inbound_check): Update.
4871 (do_jump_if_equal): Update.
4872 (emit_cmp_and_jump_insns): Update.
4873 * tree-tailcall.c (decrease_profile): Update.
4874 (eliminate_tail_call): Update.
4875 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4876 (vect_do_peeling): Update.
4877 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4878 * ubsan.c (ubsan_expand_null_ifn): Update.
4879 (ubsan_expand_ptr_ifn): Update.
4880 * value-prof.c (gimple_divmod_fixed_value): Update.
4881 (gimple_mod_pow2): Update.
4882 (gimple_mod_subtract): Update.
4883 (gimple_ic): Update.
4884 (gimple_stringop_fixed_value): Update.
4886 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4889 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4891 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4893 PR rtl-optimization/82395
4894 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4895 based on non_spilled_static_chain_regno_p.
4897 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4899 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4900 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4901 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4903 2017-10-19 Martin Sebor <msebor@redhat.com>
4905 PR tree-optimization/82596
4906 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4908 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4910 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4911 (get_mem_refs_of_builtin_call): Likewise.
4912 * builtins.c (expand_builtin_apply): Adjust call to
4913 allocate_dynamic_stack_space.
4914 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4915 the third argument to allocate_dynamic_stack_space, otherwise -1.
4916 (expand_builtin): Deal with all alloca variants.
4917 (is_inexpensive_builtin): Likewise.
4918 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4919 * calls.c (special_function_p): Deal with all alloca variants.
4920 (initialize_argument_information): Adjust call to
4921 allocate_dynamic_stack_space.
4922 (expand_call): Likewise.
4923 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4924 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4925 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4926 use it for the stack usage computation.
4927 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4928 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4929 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4930 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4931 (in_loop_p): Remove first argument and useless check.
4932 (pass_walloca::execute): Remove useless test and adjust call to above.
4933 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4934 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4935 (gimplify_call_expr): Deal with all alloca variants.
4936 * hsa-gen.c (gen_hsa_alloca): Likewise.
4937 (gen_hsa_insns_for_call): Likewise.
4938 * ipa-pure-const.c (special_builtin_state): Likewise.
4939 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4940 * tree-object-size.c (alloc_object_size): Likewise.
4941 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4942 (call_may_clobber_ref_p_1): Likewise.
4943 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4944 (ccp_fold_stmt): Likewise.
4945 (optimize_stack_restore): Likewise.
4946 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4947 (mark_all_reaching_defs_necessary_1): Likewise.
4948 (propagate_necessity): Likewise.
4949 (eliminate_unnecessary_stmts): Likewise.
4950 * tree.c (build_common_builtin_nodes): Build
4951 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4952 (build_alloca_call_expr): New function.
4953 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4954 (CASE_BUILT_IN_ALLOCA): Likewise.
4955 (build_alloca_call_expr): Declare.
4956 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4958 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4961 * dwarf2out.c (new_die_raw): New static inline function.
4962 (new_die): Use it to create the DIE.
4963 (add_AT_external_die_ref): Likewise.
4964 (clone_die): Likewise.
4965 (clone_as_declaration): Likewise.
4966 (dwarf2out_vms_debug_main_pointer): Likewise.
4967 (base_type_die): Likewise. Remove early return for corner cases.
4968 Do not call add_pubtype on the DIE here.
4969 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4970 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4971 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4972 native order exists for base types, attach the DIE manually and call
4973 add_pubtype on it. Do not equate a reverse order DIE to the type.
4975 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4977 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4978 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4980 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4982 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4984 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4985 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4986 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4987 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4988 * opts.c (sanitizer_opts): Add builtin.
4989 * ubsan.c (instrument_builtin): New function.
4990 (pass_ubsan::execute): Call it.
4991 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4992 * doc/invoke.texi: Document -fsanitize=builtin.
4994 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4995 builtins, store max (log2 (align), 0) into uchar field instead of
4996 align into uptr field.
4997 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4998 store uchar 0 field instead of uptr 0 field.
4999 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5000 instead of passing one address of struct with 2 locations pass
5001 two addresses of structs with 1 location each.
5002 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5003 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5004 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5005 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5006 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5007 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5008 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5009 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5011 2017-10-19 Martin Liska <mliska@suse.cz>
5014 * file-find.c (remove_prefix): Remove.
5015 * file-find.h (remove_prefix): Likewise.
5016 * gcc-ar.c: Remove smartness of lookup.
5018 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5020 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5021 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5022 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5024 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5027 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5028 (setcc + and to xor + setcc): New peephole2.
5030 2017-10-19 Tom de Vries <tom@codesourcery.com>
5032 * doc/sourcebuild.texi (Test Directives, Variants of
5033 dg-require-support): Add dg-require-stack-size.
5035 2017-10-19 Martin Liska <mliska@suse.cz>
5038 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5039 that have a large alignment.
5040 (gimplify_target_expr): Likewise.
5042 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5044 PR rtl-optimization/82602
5045 * ira.c (rtx_moveable_p): Return false for volatile asm.
5047 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5050 * config/i386/i386-modes.def (CCGZ): New CC mode.
5051 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5052 * config/i386/predicates.md (ix86_comparison_operator):
5054 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5055 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5056 with double-word subtraction.
5057 (put_condition_code): Handle CCGZmode.
5059 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5061 * wide-int.cc (debug (const wide_int &)): New.
5062 (debug (const wide_int *)): New.
5063 (debug (const widest_int &)): New.
5064 (debug (const widest_int *)): New.
5066 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5069 * lra-constraints.c (curr_insn_transform): Use non-input operand
5070 instead of output one for matched reload.
5072 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5074 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5075 (tree-ssa-loop-ivopts.h): New header file.
5076 (struct builtin_info): New fields.
5077 (classify_builtin_1): Compute and record base and offset parts for
5078 memset builtin partition by calling strip_offset.
5079 (offset_cmp, fuse_memset_builtins): New functions.
5080 (finalize_partitions): Fuse adjacent memset partitions by calling
5082 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5083 Expose the interface.
5084 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5086 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5088 PR tree-optimization/82574
5089 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5090 that data reference must be executed exactly once per iteration
5091 against the outermost loop in nest.
5092 (classify_partition): Update call to above function.
5094 2017-10-18 Richard Biener <rguenther@suse.de>
5096 PR tree-optimization/82591
5097 * graphite.c (graphite_transform_loops): Move code gen message
5099 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5100 Here. Handle scop_to_isl_ast failing.
5101 (scop_to_isl_ast): Limit the number of ISL operations.
5103 2017-10-18 Richard Biener <rguenther@suse.de>
5105 * graphite-isl-ast-to-gimple.c
5106 (translate_isl_ast_to_gimple::set_rename): Simplify.
5107 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5108 (graphite_copy_stmts_from_block): ... here.
5109 (copy_bb_and_scalar_dependences): Simplify.
5110 (add_parameters_to_ivs_params): Canonicalize.
5111 (generate_entry_out_of_ssa_copies): Simplify.
5112 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5113 by passing in ISL dimension.
5114 (parameter_index_in_region_1): Rename to ...
5115 (parameter_index_in_region): ... this.
5116 (extract_affine): Adjust assert, pass down parameter index.
5117 (add_param_constraints): Use range-info when available.
5118 (build_scop_context): Adjust.
5119 * sese.c (new_sese_info): Adjust.
5120 (free_sese_info): Likewise.
5121 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5122 Remove unused typedefs.
5123 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5125 2017-10-18 Martin Liska <mliska@suse.cz>
5127 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5129 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5131 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5132 (s390_sched_init): Do not reset s390_sched_state if we entered the
5133 current basic block via a fallthru edge and all others are unlikely.
5135 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5137 * config/s390/s390.c (NUM_SIDES): New variable.
5138 (LONGRUNNING_THRESHOLD): New variable.
5139 (LATENCY_FACTOR): New variable.
5140 (s390_sched_score): Decrease score for long-running instructions on
5142 (s390_sched_variable_issue): Perform bookkeeping for long-running
5145 2017-10-18 Richard Biener <rguenther@suse.de>
5147 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5148 Simplify with removal of the parameter rename map.
5149 (set_rename): Likewise.
5150 (should_copy_to_new_region): Likewise.
5151 (graphite_copy_stmts_from_block): Likewise.
5152 (copy_bb_and_scalar_dependences): Remove initialization of
5153 unused copied_bb_map.
5155 (copy_internal_parameters): Likewise.
5156 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5157 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5158 Use INTEGRAL_TYPE_P.
5159 (parameter_index_in_region_1): Rename to ...
5160 (assign_parameter_index_in_region): ... this. Assert we have
5161 a parameter we handle.
5162 (scan_tree_for_params): Adjust.
5163 * sese.h (parameter_rename_map_t): Remove.
5164 (struct sese_info_t): Remove unused parameter_rename_map and
5165 copied_bb_map members.
5166 * sese.c (new_sese_info): Adjust.
5167 (free_sese_info): Likewise.
5169 2017-10-18 Martin Liska <mliska@suse.cz>
5172 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5173 on an abnormal edge.
5175 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5177 * doc/invoke.texi (ffunction-sections and fdata-sections):
5180 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5182 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5183 the use statement can throw internally.
5185 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5187 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5188 any RTX present on the RHS of a SET.
5189 * compare-elim.c (try_eliminate_compare): Restore comment.
5191 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5193 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5194 may be also called on tcc_type nodes.
5195 * langhooks.c (lhd_tree_size): Likewise.
5197 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5199 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5200 format_warning_at_substring.
5201 (maybe_warn): Convert source_range * param to a location_t. Pass
5202 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5203 (format_directive): Remove code to extract source_ranges and
5204 source_range * in favor of just a location_t.
5205 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5207 * substring-locations.c (format_warning_va): Convert
5208 source_range * param to a location_t.
5209 (format_warning_at_substring): Likewise.
5210 * substring-locations.h (format_warning_va): Likewise.
5211 (format_warning_at_substring): Likewise.
5213 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5215 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5217 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5218 and scatter/gather ops.
5220 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5221 vec_gather_load and vec_scatter_store.
5222 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5223 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5224 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5225 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5226 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5227 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5229 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5231 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5232 Detect FTST instruction and handle its register pops. Only pop
5233 second operand if can_pop_second_op is true.
5234 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5235 set can_pop_second_op to false in the compare_for_stack_reg call.
5237 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5238 output_fp_compare for stack register operands.
5239 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5240 instructions here. Do not emit stack register pops here. Assert
5241 that FCOMPP pops next to top stack register. Rewrite function.
5243 2017-10-17 Nathan Sidwell <nathan@acm.org>
5246 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5247 use DECL_ASSEMBLER_NAME_RAW.
5250 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5253 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5254 Wilco Dijkstra <wilco.dijkstra@arm.com>
5256 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5257 converstion to Pmode from the buf_addr.
5259 2017-10-17 Richard Biener <rguenther@suse.de>
5261 * graphite-dependences.c (scop_get_reads_and_writes): Change
5262 output parameters to references.
5264 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5266 PR 71026/tree-optimization
5267 * fold-const.c (distribute_real_division): Removed.
5268 (fold_binary_loc): Remove calls to distribute_real_divison.
5270 2017-10-17 Richard Biener <rguenther@suse.de>
5272 * graphite-scop-detection.c
5273 (scop_detection::stmt_has_simple_data_refs_p): Always use
5274 the full nest as region.
5275 (try_generate_gimple_bb): Likewise.
5276 * sese.c (scalar_evolution_in_region): Simplify now that
5277 SCEV can handle instantiation in regions.
5278 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5279 in the non-loop part of a function if requested.
5281 2017-10-17 Richard Biener <rguenther@suse.de>
5283 PR tree-optimization/82563
5284 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5286 (graphite_regenerate_ast_isl): Call it.
5287 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5289 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5291 PR tree-optimization/82549
5292 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5293 Formatting fixes. Instead of calling make_bit_field_ref with negative
5296 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5298 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5299 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5300 _mm_maskz_reduce_ss): New.
5301 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5302 __builtin_ia32_reducess_mask): Ditto..
5303 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5304 * config/i386/sse.md (reduces<mode>): Renamed to ...
5305 (reduces<mode><mask_scalar_name>): ... this.
5306 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5308 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5309 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5311 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5313 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5314 * selftest-run-tests.c (selftest::run_tests): Call
5315 selftest::unique_ptr_tests_cc_tests.
5316 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5317 * unique-ptr-tests.cc: New file.
5319 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5322 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5324 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5325 (make_hard_regno_born, make_hard_regno_dead): Update
5326 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5328 2017-10-16 Jeff Law <law@redhat.com>
5330 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5332 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5334 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5336 2017-10-16 Olivier Hainque <hainque@adacore.com>
5338 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5339 with_cpu if we were configured for an e500v2 target cpu name.
5341 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5343 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5344 * doc/invoke.texi: Document +nodsp as a valid extension for
5347 2017-10-16 Martin Liska <mliska@suse.cz>
5349 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5350 (test_set_range): Likewise.
5351 (test_range_functions): Rename to ...
5352 (test_bit_in_range): ... this.
5353 (sbitmap_c_tests): Add new test.
5355 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5357 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5359 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5360 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5362 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5364 * config/aarch64/aarch64-builtins.c
5365 (aarch64_types_quadopu_lane_qualifiers): New.
5366 (TYPES_QUADOPU_LANE): New.
5367 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5368 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5369 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5370 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5371 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5372 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5373 (Vdottype, DOTPROD): New.
5374 (sur): Add SDOT and UDOT.
5376 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5378 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5379 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5380 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5382 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5383 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5384 Enable TARGET_DOTPROD.
5385 (cortex-a75.cortex-a55): Likewise.
5386 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5388 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5390 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5391 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5393 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5395 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5396 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5397 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5398 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5399 * config/arm/types.md (neon_dot, neon_dot_q): New.
5400 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5402 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5404 * config/arm/arm.h (TARGET_DOTPROD): New.
5405 * config/arm/arm.c (arm_arch_dotprod): New.
5406 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5407 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5408 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5409 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5410 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5411 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5412 * doc/invoke.texi (armv8.2-a): Document dotprod
5414 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5416 * i386.c (ix86_vec_cost): New function.
5417 (ix86_rtx_costs): Handle vector operations better.
5418 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5419 * x86-tune-costs.h: Add new costs to all tables.
5421 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5423 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5425 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5426 divsd, sqrtss and sqrtsd
5427 * x86-tune-costs.h: Add new entries to all costs.
5428 (znver1_cost): Fix to match real instruction latencies.
5430 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5431 Michael Collison <michael.collison@arm.com>
5433 * compare-elim.c: Include emit-rtl.h.
5434 (can_merge_compare_into_arith): New function.
5435 (try_validate_parallel): Likewise.
5436 (try_merge_compare): Likewise.
5437 (try_eliminate_compare): Call the above when no previous clobber
5439 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5442 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5446 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5447 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5451 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5452 to be any operand_equal_p operands. For & (B - 1) require
5453 B to be power of 2. Recognize
5454 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5456 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5459 * optabs.c (expand_memory_blockage): Fix call of
5460 targetm.have_memory_blockage.
5462 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5465 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5466 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5467 objects to extra_objs instead of overwriting it.
5469 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5471 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5472 Use any_fp_register_operand as operand[3] predicate. Simplify
5473 equality test for operands[2] and operands[4] memory location.
5474 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5475 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5476 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5477 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5478 any_fp_register_operand as operand[1] predicate. Simplify
5479 equality test for operands[0] and operands[3] memory location.
5480 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5481 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5482 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5484 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5486 * target-insns.def: Add memory_blockage.
5487 * optabs.c (expand_memory_blockage): New function.
5488 (expand_asm_memory_barrier): Rename ...
5489 (expand_asm_memory_blockage): ... to this.
5490 (expand_mem_thread_fence): Call expand_memory_blockage
5491 instead of expand_asm_memory_barrier.
5492 (expand_mem_singnal_fence): Ditto.
5493 (expand_atomic_load): Ditto.
5494 (expand_atomic_store): Ditto.
5495 * doc/md.texi (Standard Pattern Names For Generation):
5496 Document memory_blockage instruction pattern.
5498 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5500 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5501 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5502 * config/rl78/rl78.md: New define_expand "adddi3".
5504 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5506 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5509 2017-10-13 Jeff Law <law@redhat.com>
5511 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5513 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5516 * internal-fn.c (expand_mul_overflow): If both operands have
5517 the same highpart of -1 or 0 and the topmost bit of lowpart
5518 is different, overflow is if res <= 0 rather than res < 0.
5520 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5522 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5523 TARGET_P9_VECTOR code for unaligned_load case.
5525 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5527 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5529 2017-10-13 Nathan Sidwell <nathan@acm.org>
5531 * tree-core.h (tree_contains_struct): Make bool.
5532 * tree.c (tree_contains_struct): Likewise.
5533 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5534 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5535 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5536 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5537 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5539 2017-10-13 Richard Biener <rguenther@suse.de>
5541 * graphite-isl-ast-to-gimple.c
5542 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5543 parameters and dominance check.
5544 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5545 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5546 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5547 Do not update SSA form here or do intermediate IL verification.
5548 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5549 (graphite_initialize): Remove check on the number of loops in
5550 the function and inline into graphite_transform_loops.
5551 (graphite_finalize): Inline into graphite_transform_loops.
5552 (graphite_transform_loops): Perform SSA update and IL verification
5554 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5556 2017-10-13 Richard Biener <rguenther@suse.de>
5558 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5559 graphite_expression_type_precision): Avoid global constructor
5561 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5562 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5563 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5564 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5565 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5566 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5568 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5571 * config/i386/i386.h (ix86_red_zone_size): New.
5572 * config/i386/i386.md (push peephole2s): Replace
5573 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5575 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5576 Alan Hayward <alan.hayward@arm.com>
5577 David Sherwood <david.sherwood@arm.com>
5579 * combine.c (can_change_dest_mode): Reject changes in
5580 REGMODE_NATURAL_SIZE.
5582 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5583 Alan Hayward <alan.hayward@arm.com>
5584 David Sherwood <david.sherwood@arm.com>
5586 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5587 (expand_debug_source_expr): Likewise.
5588 * combine.c (combine_simplify_rtx): Likewise.
5589 * cse.c (fold_rtx): Likewise.
5590 * fwprop.c (canonicalize_address): Likewise.
5591 * targhooks.c (default_shift_truncation_mask): 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 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5598 (widened_mode): Likewise.
5599 (expand_unop): Likewise.
5600 * ree.c (transform_ifelse): Likewise.
5601 (merge_def_and_ext): Likewise.
5602 (combine_reaching_defs): Likewise.
5603 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5605 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5606 Alan Hayward <alan.hayward@arm.com>
5607 David Sherwood <david.sherwood@arm.com>
5609 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5610 * combine.c (gen_lowpart_for_combine): Likewise.
5611 * dwarf2out.c (rtl_for_decl_location): Likewise.
5612 * final.c (alter_subreg): Likewise.
5613 * rtlhooks.c (gen_lowpart_general): Likewise.
5614 (gen_lowpart_if_possible): Likewise.
5616 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5617 Alan Hayward <alan.hayward@arm.com>
5618 David Sherwood <david.sherwood@arm.com>
5620 * calls.c (expand_call): Use subreg_lowpart_offset.
5621 * cse.c (cse_insn): Likewise.
5622 * regcprop.c (copy_value): Likewise.
5623 (copyprop_hardreg_forward_1): Likewise.
5625 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5628 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5629 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5630 =Q constraints to +Q and into insn condition add check
5631 that operands[0] and operands[1] are equal.
5632 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5633 =Q constraints to +Q and into insn condition add check
5634 that operands[0] is equal to either operands[1] or operands[2].
5637 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5638 instead of handling MINUS_EXPR twice (once for each argument),
5639 canonicalize operand order and handle just once, use rtype where
5640 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5643 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5644 any values of __C while still being pattern recognizable as a simple
5647 2017-10-13 Richard Biener <rguenther@suse.de>
5649 PR tree-optimization/82451
5651 2017-10-02 Richard Biener <rguenther@suse.de>
5653 PR tree-optimization/82355
5654 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5655 a mapping for the enclosing loop but avoid generating one for
5657 (copy_bb_and_scalar_dependences): Remove premature codegen
5658 error on PHIs in blocks duplicated into multiple places.
5659 * graphite-scop-detection.c
5660 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5661 in the region use it as loop and nest to analyze the DR in.
5662 (try_generate_gimple_bb): Likewise.
5663 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5664 (add_loop_constraints): For blocks in a loop not in the region
5665 create a dimension with a single iteration.
5666 * sese.h (gbb_loop_at_index): Remove assert.
5668 * cfgloop.c (loop_preheader_edge): For the loop tree root
5669 return the single successor of the entry block.
5670 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5671 Reset the SCEV hashtable and niters.
5672 * graphite-scop-detection.c
5673 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5674 assert that we only have POLYNOMIAL_CHREC that vary in loops
5675 contained in the region.
5676 (scop_detection::graphite_can_represent_expr): Adjust.
5677 (scop_detection::stmt_has_simple_data_refs_p): For loops
5678 not in the region set loop to NULL. The nest is now the
5679 entry edge to the region.
5680 (try_generate_gimple_bb): Likewise.
5681 * sese.c (scalar_evolution_in_region): Adjust for
5682 instantiate_scev change.
5683 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5684 Make nest parameter the edge into the region.
5685 (create_data_ref): Likewise.
5686 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5687 entry edge into a region and adjust instantiate_scev calls.
5688 (create_data_ref): Likewise.
5689 (graphite_find_data_references_in_stmt): Likewise.
5690 (find_data_references_in_stmt): Pass the loop preheader edge
5691 from the nest argument.
5692 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5693 parameter the edge into the region.
5694 (instantiate_parameters): Use the loop preheader edge as entry.
5695 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5697 (get_instantiated_value_entry): Make instantiate_below parameter
5698 the edge into the region.
5699 (instantiate_scev_name): Likewise. Adjust dominance checks,
5700 when we cannot use loop-based instantiation instantiate by
5701 walking use-def chains.
5702 (instantiate_scev_poly): Adjust.
5703 (instantiate_scev_binary): Likewise.
5704 (instantiate_scev_convert): Likewise.
5705 (instantiate_scev_not): Likewise.
5706 (instantiate_array_ref): Remove.
5707 (instantiate_scev_3): Likewise.
5708 (instantiate_scev_2): Likewise.
5709 (instantiate_scev_1): Likewise.
5710 (instantiate_scev_r): Do not blindly handle N-operand trees.
5711 Do not instantiate array-refs. Handle all constants and invariants.
5712 (instantiate_scev): Make instantiate_below parameter
5713 the edge into the region.
5714 (resolve_mixers): Use the loop preheader edge for the region
5715 parameter to instantiate_scev_r.
5716 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5718 2017-10-13 Richard Biener <rguenther@suse.de>
5720 PR tree-optimization/82525
5721 * graphite-isl-ast-to-gimple.c
5722 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5724 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5725 Fail code generation when we cannot represent the isl integer.
5726 (binary_op_to_tree): Elide modulo operations that are no-ops
5727 in the type we code generate. Remove now superfluous code
5730 2017-10-13 Richard Biener <rguenther@suse.de>
5732 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5733 (scop_detection::harmful_loop_in_region): Remove premature
5734 IV type restriction.
5735 (scop_detection::graphite_can_represent_scev): We can handle
5736 pointer IVs just fine.
5738 2017-10-13 Alan Modra <amodra@gmail.com>
5740 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5741 "Clobbers and Scratch Registers". Add paragraph on
5742 alternative to clobbers for scratch registers and OpenBLAS
5745 2017-10-13 Alan Modra <amodra@gmail.com>
5747 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5748 example of a memory input for a string of known length. Move
5749 commentary out of table. Add a number of new examples
5750 covering array memory inputs.
5752 2017-10-12 Martin Liska <mliska@suse.cz>
5754 PR tree-optimization/82493
5755 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5756 (test_range_functions): New function.
5757 (sbitmap_c_tests): Likewise.
5758 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5759 * selftest.h (sbitmap_c_tests): New function.
5761 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5763 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5765 * config/rs6000/amo.h: Fix spacing issue.
5767 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5770 * config/i386/i386.md (*ashl<mode>3_mask_1,
5771 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5772 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5775 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5777 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5778 (profile_probability): Set max_probability
5779 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5781 * profile-count.c (profile_probability::differs_from_p): Do not
5782 rely on max_probaiblity == 10000
5784 2017-10-12 Jeff Law <law@redhat.com>
5786 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5789 2017-10-12 Martin Sebor <msebor@redhat.com>
5793 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5794 (handle_alias_pairs): Call it.
5795 * common.opt (-Wattribute-alias): New option.
5796 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5797 * doc/invoke.texi (-Wattribute-alias): Document.
5799 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5802 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5804 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5806 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5807 (make_hard_regno_born, make_hard_regno_dead): Update
5808 bb_killed_pseudos and bb_gen_pseudos.
5810 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5812 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5814 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5816 * config/alpha/alpha.c (alpha_split_conditional_move):
5817 Use std::swap instead of manually swapping.
5818 (alpha_stdarg_optimize_hook): Ditto.
5819 (alpha_canonicalize_comparison): Ditto.
5821 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5823 * tree-loop-distribution.c (struct builtin_info): New struct.
5824 (struct partition): Refactor fields into struct builtin_info.
5825 (partition_free): Free struct builtin_info.
5826 (build_size_arg_loc, build_addr_arg_loc): Delete.
5827 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5828 information from struct builtin_info.
5829 (find_single_drs): New function refactored from classify_partition.
5830 Also moved builtin validity checks to this function.
5831 (compute_access_range, alloc_builtin): New functions.
5832 (classify_builtin_st, classify_builtin_ldst): New functions.
5833 (classify_partition): Refactor code into functions find_single_drs,
5834 classify_builtin_st and classify_builtin_ldst.
5835 (distribute_loop): Don't do runtime alias check when distributing
5837 (find_seed_stmts_for_distribution): New function.
5838 (pass_loop_distribution::execute): Refactor code finding seed
5839 stmts into above function. Support distribution for the innermost
5840 two-level loop nest. Adjust dump information.
5842 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5844 * tree-loop-distribution.c: Adjust the general comment.
5845 (NUM_PARTITION_THRESHOLD): New macro.
5846 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5847 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5848 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5849 in call to build_partition_graph.
5850 (finalize_partitions): New parameter. Make loop distribution more
5851 conservative by fusing more partitions.
5852 (distribute_loop): Don't do runtime alias check in case of loop nest
5854 (find_seed_stmts_for_distribution): New function.
5855 (prepare_perfect_loop_nest): New function.
5856 (pass_loop_distribution::execute): Refactor code finding seed stmts
5857 and loop nest into above functions. Support loop nest distribution.
5858 Adjust dump information accordingly.
5860 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5862 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5863 and set PTYPE_SEQUENTIAL for merged partition.
5865 2017-10-12 Richard Biener <rguenther@suse.de>
5867 PR tree-optimization/69728
5869 2017-09-19 Richard Biener <rguenther@suse.de>
5871 PR tree-optimization/69728
5872 * graphite-sese-to-poly.c (schedule_error): New global.
5873 (add_loop_schedule): Handle empty domain by failing the
5875 (build_original_schedule): Handle schedule_error.
5877 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5878 domain by returning an unchanged schedule.
5880 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5882 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5883 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5885 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5887 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5890 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5893 * expr.c (store_field): Don't optimize away bitsize == 0 store
5894 from CALL_EXPR with addressable return type.
5896 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5898 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5899 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5900 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5901 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5902 TARGET_ISEL instead of TARGET_ISEL<sel>.
5904 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5906 * config/rs6000/rs6000.c
5907 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5909 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5911 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5913 (reg_or_cint_operand): Fix comment.
5914 (reg_or_zero_operand): New predicate.
5915 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5916 * config/rs6000/rs6000.c (output_isel): Delete.
5917 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5918 instead of reg_or_cint_operand. Output instruction directly (not via
5920 (isel_unsigned_<mode>): Ditto.
5921 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5922 gpc_reg_operand. Add an instruction alternative for this. Output
5923 instruction directly.
5924 (*isel_reversed_unsigned_<mode>): Ditto.
5926 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5928 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5929 (TARGET_CANONICALIZE_COMPARISON): Define.
5931 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5934 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5935 Check whether the dest is REG before adding REG_EQUIV note.
5937 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5940 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5942 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5943 (make_hard_regno_born, make_hard_regno_dead): Update
5944 bb_killed_pseudos and bb_gen_pseudos.
5946 2017-10-11 Nathan Sidwell <nathan@acm.org>
5948 * incpath.h (enum incpath_kind): Name enum, prefix values.
5949 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5950 * incpath.c (heads, tails): Use INC_MAX.
5951 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5952 (merge_include_chains, split_quote_chain,
5953 register_include_chains): Update incpath_kind names.
5954 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5955 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5957 (add_framework_path, darwin_register_objc_includes): Likewise.
5958 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5960 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5962 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5963 Do not use float_operator operator predicate.
5964 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5965 * config/i386/predicates.md (float_operator): Remove predicate.
5967 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5969 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5970 (*jccxf_i387): Ditto.
5971 (*jcc<mode>_i387): Ditto.
5972 (*jccu<mode>_i387): Ditto.
5973 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5974 (*jcc_*_i387 splitters): Remove.
5975 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5976 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5977 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5980 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5982 * profile-count.h (slow_safe_scale_64bit): New function.
5983 (safe_scale_64bit): New inline.
5984 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5985 * profile-count.c: Include wide-int.h
5986 (slow_safe_scale_64bit): New.
5988 2017-10-11 Nathan Sidwell <nathan@acm.org>
5990 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5991 HAS_DECL_ASSEMBLER_NAME_P.
5992 * gimple-expr.c (gimple_decl_printable_name: Check
5993 HAS_DECL_ASSEMBLER_NAME_P too.
5994 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5995 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5996 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5997 * passes.c (rest_of_decl_compilation): Check
5998 HAS_DECL_ASSEMBLER_NAME_P too.
5999 * recog.c (verify_changes): Likewise.
6000 * tree-pretty-print.c (dump_decl_name): Likewise.
6001 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6003 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6004 (SET_DECL_ASSEMBLER_NAME): Use it.
6005 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6006 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6007 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6009 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6011 * config.gcc (i386, x86_64): Add extra objects.
6012 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6013 (ix86_min_insn_size): Declare.
6014 (ix86_issue_rate): Declare.
6015 (ix86_adjust_cost): Declare.
6016 (ia32_multipass_dfa_lookahead): Declare.
6017 (ix86_macro_fusion_p): Declare.
6018 (ix86_macro_fusion_pair_p): Declare.
6019 (ix86_bd_has_dispatch): Declare.
6020 (ix86_bd_do_dispatch): Declare.
6021 (ix86_core2i7_init_hooks): Declare.
6022 (ix86_atom_sched_reorder): Declare.
6023 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6024 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6025 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6026 (rip_relative_addr_p): Rename to ...
6027 (ix86_rip_relative_addr_p): ... this one; export.
6028 (memory_address_length): Update.
6029 (ix86_issue_rate): Move to x86-tune-sched.c.
6030 (ix86_flags_dependent): Move to x86-tune-sched.c.
6031 (ix86_agi_dependent): Move to x86-tune-sched.c.
6032 (exact_dependency_1): Move to x86-tune-sched.c.
6033 (exact_store_load_dependency): Move to x86-tune-sched.c.
6034 (ix86_adjust_cost): Move to x86-tune-sched.c.
6035 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6036 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6037 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6038 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6039 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6040 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6041 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6042 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6043 (min_insn_size): Rename to ...
6044 (ix86_min_insn_size): ... this one; export.
6045 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6046 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6047 (core2i7_first_cycle_multipass_backtrack): Move to
6048 x86-tune-sched-core.c.
6049 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6050 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6051 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6052 (ix86_avoid_jump_mispredicts): Update.
6053 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6054 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6055 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6056 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6057 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6058 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6059 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6060 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6061 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6062 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6063 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6064 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6065 (BIG): Move to ix86-tune-sched-bd.c.
6066 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6067 (enum insn_path): Move to ix86-tune-sched-bd.c.
6068 (get_mem_group): Move to ix86-tune-sched-bd.c.
6069 (is_cmp): Move to ix86-tune-sched-bd.c.
6070 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6071 (is_branch): Move to ix86-tune-sched-bd.c.
6072 (is_prefetch): Move to ix86-tune-sched-bd.c.
6073 (init_window): Move to ix86-tune-sched-bd.c.
6074 (allocate_window): Move to ix86-tune-sched-bd.c.
6075 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6076 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6077 (process_end_window): Move to ix86-tune-sched-bd.c.
6078 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6079 (find_constant): Move to ix86-tune-sched-bd.c.
6080 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6081 (has_immediate): Move to ix86-tune-sched-bd.c.
6082 (get_insn_path): Move to ix86-tune-sched-bd.c.
6083 (get_insn_group): Move to ix86-tune-sched-bd.c.
6084 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6085 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6086 (add_insn_window): Move to ix86-tune-sched-bd.c.
6087 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6088 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6089 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6090 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6091 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6092 (do_dispatch): Move to ix86-tune-sched-bd.c.
6093 (has_dispatch): Move to ix86-tune-sched-bd.c.
6094 * config/i386/t-i386: Add new object files.
6095 * config/i386/x86-tune-costs.h: New file.
6096 * config/i386/x86-tune-sched-atom.c: New file.
6097 * config/i386/x86-tune-sched-bd.c: New file.
6098 * config/i386/x86-tune-sched-core.c: New file.
6099 * config/i386/x86-tune-sched.c: New file.
6101 2017-10-11 Liu Hao <lh_mouse@126.com>
6103 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6104 the generic version below instead.
6105 (should_colorize): Recognize Windows consoles as terminals
6107 * pretty-print.c [__MINGW32__] (write_all): New function.
6108 [__MINGW32__] (find_esc_head): Likewise.
6109 [__MINGW32__] (find_esc_terminator): Likewise.
6110 [__MINGW32__] (eat_esc_sequence): Likewise.
6111 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6113 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6116 2017-10-11 Richard Biener <rguenther@suse.de>
6118 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6119 Properly call analyze_scalar_evolution with the loop of the stmt.
6121 2017-10-11 Richard Biener <rguenther@suse.de>
6123 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6124 * tree-core.h (tree_base): Add chrec_var union member.
6125 * tree.h (CHREC_VAR): Remove.
6126 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6127 * tree-chrec.h (build_polynomial_chrec): Adjust.
6128 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6129 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6131 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6133 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6134 * match.pd: ... here.
6135 ((T) X == (T) Y): Relax condition.
6137 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6139 PR tree-optimization/82472
6140 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6142 (break_alias_scc_partitions): Update postorder number.
6144 2017-10-11 Martin Liska <mliska@suse.cz>
6147 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6149 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6151 2017-10-11 Martin Liska <mliska@suse.cz>
6153 * print-rtl.c (print_insn): Move declaration of idbuf
6154 to same scope as name.
6156 2017-10-11 Martin Liska <mliska@suse.cz>
6161 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6164 2017-10-11 Martin Liska <mliska@suse.cz>
6167 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6170 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6172 * config.gcc (default_gnu_indirect_function): Default to yes for
6173 arm*-*-linux* with glibc.
6175 2017-10-11 Richard Biener <rguenther@suse.de>
6177 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6178 default-defs and types we do not want to analyze.
6179 (interpret_loop_phi): Replace unreachable code with an assert.
6180 (compute_scalar_evolution_in_loop): Remove and inline ...
6181 (analyze_scalar_evolution_1): ... here, replacing condition with
6182 what makes the intent clearer. Remove handling of cases
6183 get_scalar_evolution now handles.
6185 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6187 PR rtl-optimization/81434
6188 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6189 comment for main loop. In sched_group_found if, also add checks for
6190 pass and min_cost_group.
6192 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6194 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6195 (rs6000_insn_cost): New function.
6196 * config/rs6000/rs6000.md (cost): New attribute.
6198 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6199 H.J. Lu <hongjiu.lu@intel.com>
6203 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6204 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6205 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6206 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6207 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6208 OPTION_MASK_ISA_MMX and some other bit set require both
6209 mmx and the ISAs without the mmx bit.
6210 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6211 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6212 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6213 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6214 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6215 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6216 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6217 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6218 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6219 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6220 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6221 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6222 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6223 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6224 Add OPTION_MASK_ISA_MMX.
6226 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6228 * config.gcc (armv7*-*-freebsd*): New target.
6229 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6231 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6233 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6234 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6237 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6239 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6240 a template parameter.
6241 (WIDE_INT_REF_FOR): Update accordingly.
6242 * tree.h (wi::int_traits <const_tree>): Delete.
6243 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6244 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6245 (wi::tree_to_wide_ref): New typedef.
6246 (wi::to_wide): New function.
6247 * calls.c (get_size_range): Use wi::to_wide when operating on
6249 * cgraph.c (cgraph_node::create_thunk): Likewise.
6250 * config/i386/i386.c (ix86_data_alignment): Likewise.
6251 (ix86_local_alignment): Likewise.
6252 * dbxout.c (stabstr_O): Likewise.
6253 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6254 * expr.c (const_vector_from_tree): Likewise.
6255 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6256 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6257 (fold_negate_expr_1, int_const_binop_1, const_binop)
6258 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6259 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6260 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6261 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6262 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6263 (fold_not_const, round_up_loc): Likewise.
6264 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6265 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6266 (alloca_call_type): Likewise.
6267 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6268 * godump.c (go_output_typedef): Likewise.
6269 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6270 * internal-fn.c (get_min_precision): Likewise.
6271 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6272 * ipa-polymorphic-call.c
6273 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6274 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6275 (ipa_modify_call_arguments): Likewise.
6276 * match.pd: Likewise.
6277 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6278 * print-tree.c (print_node_brief, print_node): Likewise.
6279 * stmt.c (expand_case): Likewise.
6280 * stor-layout.c (layout_type): Likewise.
6281 * tree-affine.c (tree_to_aff_combination): Likewise.
6282 * tree-cfg.c (group_case_labels_stmt): Likewise.
6283 * tree-data-ref.c (dr_analyze_indices): Likewise.
6284 (prune_runtime_alias_test_list): Likewise.
6285 * tree-dump.c (dequeue_and_dump): Likewise.
6286 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6287 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6288 * tree-pretty-print.c (dump_generic_node): Likewise.
6289 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6290 (simple_iv_with_niters): Likewise.
6291 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6292 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6293 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6294 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6295 (refine_value_range_using_guard, number_of_iterations_ne_max)
6296 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6297 (get_cst_init_from_scev, record_nonwrapping_iv)
6298 (scev_var_range_cant_overflow): Likewise.
6299 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6300 * tree-ssa-pre.c (compute_avail): Likewise.
6301 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6302 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6303 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6304 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6305 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6306 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6307 (dump_case_nodes, try_switch_expansion): Likewise.
6308 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6309 (vect_do_peeling): Likewise.
6310 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6311 * tree-vect-stmts.c (vectorizable_load): Likewise.
6312 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6313 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6314 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6315 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6316 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6317 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6318 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6319 (evrp_dom_walker::before_dom_children): Likewise.
6320 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6321 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6322 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6323 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6324 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6325 (get_range_pos_neg): Likewise.
6326 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6327 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6328 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6329 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6330 * config/avr/avr.c (avr_fold_builtin): Likewise.
6331 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6332 * config/msp430/msp430.c (msp430_attr): Likewise.
6333 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6334 * config/powerpcspe/powerpcspe-c.c
6335 (altivec_resolve_overloaded_builtin): Likewise.
6336 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6337 (rs6000_expand_ternop_builtin): Likewise.
6338 * config/rs6000/rs6000-c.c
6339 (altivec_resolve_overloaded_builtin): Likewise.
6340 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6341 (rs6000_expand_ternop_builtin): Likewise.
6342 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6344 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6346 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6347 when copying loop nest with only one inner loop.
6349 2017-10-10 Richard Biener <rguenther@suse.de>
6351 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6352 blocks if SCEV is active.
6353 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6355 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6356 (scev_initialize): Assert we are not yet initialized.
6358 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6360 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6361 inner loop's exit stmt by making it always exit the loop, otherwise
6362 we would generate an infinite empty loop.
6364 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6366 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6367 renaming variables in new preheader if it's deleted.
6369 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6371 * tree-loop-distribution.c (struct partition): Remove unused field
6372 loops of the structure.
6373 (partition_alloc, partition_free): Ditto.
6374 (build_rdg_partition_for_vertex): Ditto.
6376 2017-10-09 Jeff Law <law@redhat.com>
6378 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6379 return type to match prototype and documentation.
6381 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6383 * config/rs6000/rs6000.c (processor_costs): Move to ...
6384 * config/rs6000/rs6000.h: ... here.
6385 (rs6000_cost): Declare.
6387 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6389 * except.c (setjmp_fn): New global variable.
6390 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6391 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6392 if DONT_USE_BUILTIN_SETJMP is defined.
6394 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6396 * target.def (insn_cost): New hook.
6397 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6398 * doc/tm.texi: Regenerate.
6399 * rtlanal.c (insn_cost): Use the new hook.
6401 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6403 * combine.c (combine_validate_cost): Compute the new insn_cost,
6404 not just pattern_cost.
6405 (try_combine): Adjust comment.
6407 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6409 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6411 * combine.c (uid_insn_cost): Adjust comment.
6412 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6414 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6415 * dse.c (find_shift_sequence): Ditto.
6416 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6417 (bb_valid_for_noce_process_p): Use pattern_cost.
6418 * rtl.h (insn_rtx_cost): Delete.
6419 (pattern_cost): New prototype.
6420 (insn_cost): New prototype.
6421 * rtlanal.c (insn_rtx_cost): Rename to...
6422 (pattern_cost): ... this.
6425 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6427 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6428 (*jcc<mode>_0_r_i387): Ditto.
6429 (*jccxf_r_i387): Ditto.
6430 (*jcc<mode>_r_i387): Ditto.
6431 (*jccu<mode>_r_i387): Ditto.
6432 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6433 (*jcc): Rename from *jcc_1.
6435 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6437 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6438 deferred rescans after the lvx/stvx recombination pre-pass.
6440 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6442 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6443 memory operation instruction support.
6444 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6445 (rs6000-ibm-aix[789]*): Likewise.
6446 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6447 Document new functions.
6449 2017-10-09 Richard Biener <rguenther@suse.de>
6451 PR tree-optimization/82397
6452 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6453 equality only for semantically equal trees.
6455 2017-10-09 Richard Biener <rguenther@suse.de>
6457 PR tree-optimization/82449
6458 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6459 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6460 allow constant addresses.
6461 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6464 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6466 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6469 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6472 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6475 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6478 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6480 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6483 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6484 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6486 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6488 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6489 (WI_BINARY_PREDICATE_RESULT): Likewise.
6490 (wi::binary_traits::operator_result): New type.
6491 (wi::binary_traits::predicate_result): Likewise.
6492 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6493 (generic_wide_int::operator==, generic_wide_int::operator!=)
6494 (generic_wide_int::operator&, generic_wide_int::and_not)
6495 (generic_wide_int::operator|, generic_wide_int::or_not)
6496 (generic_wide_int::operator^, generic_wide_int::operator+
6497 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6499 (operator~, unary operator-, operator==, operator!=, operator&)
6500 (operator|, operator^, operator+, binary operator-, operator*): New
6502 * expr.c (get_inner_reference): Use wi::bit_and_not.
6503 * fold-const.c (fold_binary_loc): Likewise.
6504 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6505 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6506 (bit_value_binop): Likewise.
6507 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6508 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6509 (extract_range_from_binary_expr_1): Likewise.
6510 (masked_increment): Likewise.
6511 (simplify_bit_ops_using_ranges): Likewise.
6513 2017-10-09 Martin Jambor <mjambor@suse.cz>
6516 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6517 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6518 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6520 (hsa_fixup_mov_insn_type): New function.
6521 (hsa_op_with_type::get_in_type): Use it.
6522 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6523 immediates in an assert.
6524 (hsa_op_with_type::extend_int_to_32bit): New method.
6525 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6526 types. Convert to dest type if necessary.
6527 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6528 (reg_for_gimple_ssa): Pass false as min32int to
6529 hsa_type_for_scalar_tree_type.
6530 (gen_hsa_addr): Fixup type when creating addresable temporary.
6531 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6532 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6533 necessary. Call hsa_fixup_mov_insn_type.
6534 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6535 extend operands and convert to dest type if necessary.
6536 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6537 to dest type if necessary.
6538 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6539 if conversion nt necessary and size matches.
6540 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6541 to dest type if necessary.
6542 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6543 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6545 (gen_hsa_clrsb): Likewise.
6546 (gen_hsa_ffs): Likewise.
6547 (gen_hsa_divmod): Extend operands and convert to dest type if
6549 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6551 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6553 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6554 Remove empty default arguments. Use a brace block as output
6556 (conditional return): Ditto.
6558 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6559 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6561 (group_ending_nop): Ditto.
6562 (doloop_end): Ditto.
6563 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6564 (splitters for those): Ditto.
6566 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6568 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6569 a conditional jump (and the compare for it) so that pc_rtx is the
6571 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6572 for the deleted and renamed ctr<mode>_internal[234] patterns.
6573 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6574 Delete second conditional return pattern.
6575 (ctr<mode>_internal2): Delete this second bdnz pattern.
6576 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6577 (ctr<mode>_internal4): Delete this second bdz pattern.
6579 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6581 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6582 (always_initialized_rtx_for_ssa_name_p): New predicate.
6583 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6584 (finish_out_of_ssa): Free new field of SA.
6585 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6586 * tree-ssa-coalesce.c: Include tree-ssa.h.
6587 (get_parm_default_def_partitions): Remove extern keyword.
6588 (get_undefined_value_partitions): New function.
6589 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6590 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6592 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6594 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6596 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6598 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6600 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6601 for targets that preffer 128bit.
6603 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6605 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6607 2017-10-08 Olivier Hainque <hainque@adacore.com>
6609 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6610 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6612 (thumb_set_return_address): Likewise.
6614 2017-10-08 Olivier Hainque <hainque@adacore.com>
6616 * common/config/arm/arm-common.c (arm_except_unwind_info):
6617 Handle DWARF2_UNWIND_INFO.
6619 2017-10-07 Michael Collison <michael.collison@arm.com>
6621 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6624 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6626 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6627 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6628 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6629 defined, force the creation of a new block for a dispatch label.
6631 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6633 * invoke.texi (Wsuggest-attribute=cold): Document.
6634 * common.opt (Wsuggest-attribute=cold): New
6635 * ipa-pure-const.c (warn_function_cold): New function.
6636 * predict.c (compute_function_frequency): Use it.
6637 * predict.h (warn_function_cold): Declare.
6639 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6641 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6644 2017-10-06 Martin Liska <mliska@suse.cz>
6646 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6647 keyword for member functions.
6648 (struct sanopt_tree_couple): New struct.
6649 (struct sanopt_tree_couple_hash): New function.
6650 (struct sanopt_ctx): Add new hash_map.
6651 (has_dominating_ubsan_ptr_check): New function.
6652 (record_ubsan_ptr_check_stmt): Likewise.
6653 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6654 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6655 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6657 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6660 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6661 aarch64_simd_valid_immediate on CONST_VECTORs.
6662 (aarch64_reg_or_bic_imm): Likewise.
6664 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6666 PR rtl-optimization/82396
6667 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6669 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6671 * graphite-dependences.c (scop_get_reads): Move code to...
6672 (scop_get_must_writes): Move code to...
6673 (scop_get_may_writes): Move code to...
6674 (scop_get_reads_and_writes): ... here.
6675 (scop_get_dependences): Call scop_get_reads_and_writes.
6677 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6679 PR tree-optimization/82434
6680 * fold-const.h (can_native_encode_type_p,
6681 can_native_encode_string_p): Remove.
6682 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6683 don't encode anything, just return what would be otherwise returned.
6684 (native_encode_fixed, native_encode_complex, native_encode_vector):
6686 (native_encode_string): Likewise. Inline by hand
6687 can_native_encode_string_p.
6688 (can_native_encode_type_p): Remove.
6689 (can_native_encode_string_p): Remove.
6690 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6691 STRING_CSTs using can_native_encode_string_p, test all
6692 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6693 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6694 argument from native_encode_expr.
6695 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6696 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6699 2017-10-06 Richard Biener <rguenther@suse.de>
6701 PR tree-optimization/82397
6702 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6703 operand_equal_p but rely on data_ref_compare_tree for detecting
6705 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6706 to match up with dr_group_sort_cmp.
6708 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6711 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6713 * config/s390/s390-builtin-types.def: Regenerate.
6715 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6718 * config/s390/s390-builtin-types.def: Regenerate.
6719 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6720 Change flag from B_VXE to B_VX.
6721 (s390_vec_min_dbl): Remove B_VXE flag.
6723 2017-10-06 Richard Biener <rguenther@suse.de>
6725 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6726 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6727 translate_isl_ast_to_gimple::is_valid_rename,
6728 translate_isl_ast_to_gimple::get_rename,
6729 translate_isl_ast_to_gimple::get_def_bb_for_const,
6730 translate_isl_ast_to_gimple::get_new_name,
6731 translate_isl_ast_to_gimple::collect_all_ssa_names,
6732 translate_isl_ast_to_gimple::copy_loop_phi_args,
6733 translate_isl_ast_to_gimple::collect_all_ssa_names,
6734 translate_isl_ast_to_gimple::copy_loop_phi_args,
6735 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6736 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6737 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6738 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6739 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6740 translate_isl_ast_to_gimple::copy_cond_phi_args,
6741 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6742 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6743 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6744 translate_isl_ast_to_gimple::rename_uses,
6745 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6746 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6747 (set_rename_for_each_def): Likewise.
6748 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6749 here. Handle rewriting SCEV analyzable uses here.
6750 (copy_bb_and_scalar_dependences): Generate code for PHI
6752 (graphite_regenerate_ast_isl): Adjust.
6753 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6754 (add_write, add_read): New functions.
6755 (build_cross_bb_scalars_def): Use it and simplify.
6756 (build_cross_bb_scalars_use): Likewise.
6757 (graphite_find_cross_bb_scalar_vars): Inline into...
6758 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6759 simulating out-of-SSA. Compute liveout and add dependencies.
6760 (build_scops): Force an empty entry block.
6761 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6763 (sese_build_liveouts): Declare.
6764 (sese_trivially_empty_bb_p): Likewise.
6765 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6766 compute liveout and debug_liveout.
6767 (sese_bad_liveouts_use): Remove.
6768 (sese_reset_debug_liveouts_bb): Likewise.
6769 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6770 (sese_build_liveouts): Build liveout and debug_liveout and store
6772 (new_sese_info): Adjust.
6773 (free_sese_info): Likewise.
6774 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6775 do not build liveout here.
6776 (move_sese_in_condition): Adjust region entry.
6777 (scev_analyzable_p): Match up with chrec_apply requirements.
6778 (sese_trivially_empty_bb_p): New.
6779 * tree-into-ssa.c (get_reaching_def): Properly support generating
6780 default-defs for incremental rewrite of anonymous names.
6782 2017-10-06 Richard Biener <rguenther@suse.de>
6784 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6785 precision do not perform modulo reduction.
6787 2017-10-06 Richard Biener <rguenther@suse.de>
6789 PR tree-optimization/82436
6790 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6791 conservatively choose the vectorization factor when checking
6792 whether we can perform the required load permutation.
6793 (vect_transform_slp_perm_load): Assert when we may not fail.
6795 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6797 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6798 message for incompatible -msdata=* and -mcall-* options.
6800 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6802 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6803 rate for post-reload scheduling.
6805 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6807 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6809 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6811 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6812 to improve monte carlo in scimark.
6814 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6816 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6817 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6818 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6819 pentium4_cost, nocona_cost): Set reassociation width to 1.
6820 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6821 width to 2 for fp operations and 1 otherwise.
6822 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6824 (atom_cost): Set reassociation width to 2.
6825 (slm_cost, generic_cost): Set fp reassociation width
6826 to 2 and 1 otherwise.
6827 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6828 (core_cost): Set fp reassociation width to 4 and vector to 2.
6829 (ix86_reassociation_width): Rewrite using cost table; special case
6830 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6831 and TARGET_AVX128_OPTIMAL.
6832 * config/i386/i386.h (processor_costs): Add
6833 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6834 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6835 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6836 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6837 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6838 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6840 2017-10-05 Nathan Sidwell <nathan@acm.org>
6842 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6844 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6846 * config/arm/arm.c (arm_test_fpu_data): New.
6847 (arm_run_selftests): Call arm_test_fpu_data.
6849 2017-10-04 Nathan Sidwell <nathan@acm.org>
6851 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6853 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6855 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6857 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6858 check type for aarch64_simd_valid_immediate.
6859 (aarch64_output_simd_mov_immediate): Update prototype.
6860 (aarch64_simd_valid_immediate): Update prototype.
6861 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6862 support for ORR-immediate.
6863 (and<mode>3): modified pattern to add support for BIC-immediate.
6864 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6865 now checks for valid immediate for BIC and ORR based on new enum
6867 (aarch64_output_simd_mov_immediate): Function now used to output
6868 BIC/ORR imm as well based on new enum argument.
6869 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6871 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6872 (aarch64_reg_or_bic_imm): Likewise.
6874 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6876 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6877 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6879 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6883 PR rtl-optimization/82396
6884 * haifa-sched.c (autopref_multipass_init): Simplify
6886 (autopref_rank_data): Simplify sort order.
6887 * sched-int.h (autopref_multipass_data_): Remove
6888 multi_mem_insn_p, min_offset and max_offset.
6890 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6892 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6894 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6896 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6899 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6901 * doc/sourcebuild.texi: Document vect_long_mult.
6903 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6905 PR tree-optimization/82413
6906 * fold-const.c (build_range_check): Use widest_int when comparing
6907 the maximum ETYPE value with HIGH.
6909 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6911 PR rtl-optimization/82396
6912 * haifa-sched.c (autopref_multipass_init): Simplify
6914 (autopref_rank_data): Simplify sort order.
6915 * sched-int.h (autopref_multipass_data_): Remove
6916 multi_mem_insn_p, min_offset and max_offset.
6918 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6920 PR tree-optimization/82381
6921 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6922 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6925 PR tree-optimization/82374
6926 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6927 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6928 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6929 current_function_decl to the new decl.
6931 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6933 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6934 helper macro for IEEE float128 hardware built-in functions.
6935 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6937 (TRUNCF128_ODD): Likewise.
6938 (ADDF128_ODD): Likewise.
6939 (SUBF128_ODD): Likewise.
6940 (MULF128_ODD): Likewise.
6941 (DIVF128_ODD): Likewise.
6942 (FMAF128_ODD): Likewise.
6943 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6944 UNSPEC_TRUNC_ROUND_TO_ODD.
6945 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6946 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6947 floating point round to odd instructions.
6948 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6949 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6950 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6951 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6952 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6953 (trunc<mode>sf2_hw): Change the truncate with round to odd
6954 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6955 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6956 to odd hardware instructions.
6957 (sub<mode>3_odd): Likewise.
6958 (mul<mode>3_odd): Likewise.
6959 (div<mode>3_odd): Likewise.
6960 (sqrt<mode>2_odd): Likewise.
6961 (fma<mode>4_odd): Likewise.
6962 (fms<mode>4_odd): Likewise.
6963 (nfma<mode>4_odd): Likewise.
6964 (nfms<mode>4_odd): Likewise.
6965 (trunc<mode>df2_odd): Change the truncate with round to odd
6966 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6968 * doc/extend.texi (PowerPC built-in functions): Update documentation
6969 for existing IEEE float128-bit built-in functions. Add built-in
6970 functions that generate the IEEE 128-bit floating point round to
6973 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6975 PR rtl-optimization/77729
6976 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6977 to (X&(C1&~C2))|C2 transformations.
6979 2017-10-03 Martin Jambor <mjambor@suse.cz>
6981 PR tree-optimization/82363
6982 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6983 mismatch, mark lacc written regardless of racc.
6985 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6987 PR tree-optimization/82381
6988 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6989 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6990 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6991 fallthrough into reassoc_stmt_dominates_stmt_p.
6994 * combine.c (combine_instructions): Don't combine in unreachable
6997 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7000 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7001 function to not use the have_cpu variable. Do not set cpu_index,
7002 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7004 (rs6000_valid_attribute_p): Remove duplicate initializations of
7005 old_optimize and func_optimize.
7006 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7007 (rs6000_activate_target_options): Make global.
7008 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7011 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7013 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7014 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7015 Set *poffset to 0 and *psize and *pmax_size to -1 if
7016 *poffset + *psize overflows in HOST_WIDE_INT.
7018 PR tree-optimization/82387
7019 PR tree-optimization/82388
7020 PR tree-optimization/82389
7021 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7022 instead of live_bytes non-NULL.
7024 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7027 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7030 2017-10-02 Richard Biener <rguenther@suse.de>
7032 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7033 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7034 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7036 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7038 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7039 requested precision matches the type's.
7040 * calls.c (alloc_max_size): Calculate the new candidate size as
7041 a widest_int and use wi::to_widest when comparing it with the
7042 current candidate size.
7043 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7044 zero rather than integer_zero_node.
7045 * match.pd: Check for a no-op conversion before using wi::add
7046 rather than after. Use tree_to_uhwi when summing small shift
7047 counts into an unsigned int.
7049 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7050 Alan Hayward <alan.hayward@arm.com>
7051 David Sherwood <david.sherwood@arm.com>
7054 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7055 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7056 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7057 POINTER_AND_FP_REGS.
7059 2017-10-02 Richard Biener <rguenther@suse.de>
7061 PR tree-optimization/82355
7062 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7063 a mapping for the enclosing loop but avoid generating one for
7065 (copy_bb_and_scalar_dependences): Remove premature codegen
7066 error on PHIs in blocks duplicated into multiple places.
7067 * graphite-scop-detection.c
7068 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7069 in the region use it as loop and nest to analyze the DR in.
7070 (try_generate_gimple_bb): Likewise.
7071 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7072 (add_loop_constraints): For blocks in a loop not in the region
7073 create a dimension with a single iteration.
7074 * sese.h (gbb_loop_at_index): Remove assert.
7076 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7078 * omp-expand.c (adjust_context_scope): New function.
7079 (expand_parallel_call): Call adjust_context_scope.
7081 2017-10-01 Jeff Law <law@redhat.com>
7083 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7084 dom_opt_dom_walker class with direct access to private members.
7085 Add comments. Call test_for_singularity.
7086 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7087 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7088 m_dummy_cond anymore.
7089 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7091 (pass_dominator:execute): Build the dummy_cond here and pass it
7092 to the dom_opt_dom_walker ctor.
7093 (test_for_singularity): New function.
7095 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7096 Maya Rashish <coypu@sdf.org>
7098 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7099 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7100 (alpha*-*-netbsd*) Use nbsd_tm_file.
7101 (arm*-*-netbsdelf*) Likewise.
7102 (i[34567]86-*-netbsdelf*) Likewise.
7103 (x86_64-*-netbsd*) Likewise.
7104 (mips*-*-netbsd*) Likewise.
7105 (powerpc-*-netbsd*) Likewise.
7106 (sh*-*-netbsd*) Likewise.
7107 (sparc-*-netbsdelf*) Likewise.
7108 (sparc64-*-netbsd*) Likewise.
7109 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7111 (vax-*-netbsdelf*) Likewise.
7112 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7113 (UINT_FAST8_TYPE) Likewise.
7114 (INT_FAST16_TYPE) Check CHAR_FAST16.
7115 (UINT_FAST16_TYPE) Likewise.
7117 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7120 * config/i386/i386.md
7121 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7122 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7123 *divmodsi4_zext_2): New define_insn_and_split.
7124 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7125 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7126 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7127 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7128 New define_insn_and_split.
7129 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7130 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7131 operands[1] having DImode when mode is SImode.
7133 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7134 always SImode for DIV and MOD in REG_EQUAL notes.
7136 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7139 * match.pd: Fix handling of NaNs in pattern.
7141 2017-09-29 Jeff Law <law@redhat.com>
7143 * sbitmap.c (bitmap_bit_in_range_p): New function.
7144 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7145 * tree-ssa-dse.c (live_bytes_read): New function.
7146 (dse_classify_store): Ignore reads of dead bytes.
7148 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7149 typos and whitespace errors.
7150 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7151 * config/s390/s390.c (s390_emit_prologue): Likewise.
7153 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7156 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7157 with a symbol for LRA.
7159 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7161 PR rtl-optimization/82338
7162 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7164 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7166 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7167 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7168 (qsort_chk): Declare.
7169 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7170 (qsort_chk): New function.
7172 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7174 PR tree-optimization/82337
7175 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7176 phi definition if the PHI result appears in an abnormal PHI.
7177 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7178 basis appears in an abnormal PHI.
7180 2017-09-29 Richard Biener <rguenther@suse.de>
7182 * graphite-isl-ast-to-gimple.c
7183 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7184 (binary_op_to_tree): Use it.
7185 (get_rename_from_scev): Likewise.
7186 (copy_loop_phi_nodes): Likewise.
7187 (copy_bb_and_scalar_dependences): Likewise.
7188 (translate_pending_phi_nodes): Likewise.
7190 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7193 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7194 for movabsq $(i32 << shift), r64.
7196 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7198 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7199 index when encoding %esp as %rsp to avoid 0x67 prefix.
7201 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7203 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7204 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7206 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7208 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7209 Extensions with more than 16 double VFP registers.
7210 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7211 to_clear_mask and all code related to it. Replace the remaining
7212 entry by a sbitmap and adapt code accordingly.
7214 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7216 * brig-builtins.def: Change pure attributes to const.
7218 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7220 * config.gcc (default_gnu_indirect_function): Default to yes for
7221 sparc*-*-linux* with glibc.
7223 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7225 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7226 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7227 when creating .init_array and .fini_array sections with priority
7230 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7233 * config/aarch64/aarch64.c
7234 (aarch64_builtin_support_vector_misalignment): Always return false
7235 when misalignment is unknown.
7237 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7239 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7240 this function to return false if the definition used by the swap
7241 instruction is artificial, or if the memory address from which the
7242 constant value is loaded is not represented by a base address held
7243 in a register or if the base address register is a frame or stack
7244 pointer. Additionally, return false if the base address of the
7245 loaded constant is a SYMBOL_REF but is not considered to be a
7247 (replace_swapped_load_constant): New function.
7248 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7249 loaded constant vector with a load of a swapped constant vector.
7251 2017-09-27 Carl Love <cel@us.ibm.com>
7253 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7254 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7255 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7258 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7260 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7261 first, always call autopref_rank_data otherwise.
7263 2017-09-27 Richard Biener <rguenther@suse.de>
7265 * graphite-scop-detection.c (find_scop_parameters): Move
7266 loop bound handling ...
7267 (gather_bbs::before_dom_children): ... here, avoiding the need
7268 to build scop_info->loop_nest.
7269 (record_loop_in_sese): Remove.
7270 * sese.h (sese_info_t::loop_nest): Remove.
7271 * sese.c (new_sese_info): Do not allocate loop_nest.
7272 (free_sese_info): Do not free loop_nest.
7274 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7277 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7278 lhs from calls if the lhs has addressable type.
7280 2017-09-27 Richard Biener <rguenther@suse.de>
7282 * graphite.h (scop::max_alias_set): New member.
7283 * graphite-scop-detection.c: Remove references to non-existing
7284 --param in comments.
7285 (build_alias_sets): Record the maximum alias set used for drs.
7286 (build_scops): Support zero as unlimited for
7287 --param graphite-max-arrays-per-scop.
7288 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7290 (build_poly_sr_1): ... here. Compute alias set based on the
7291 maximum alias set used for drs rather than
7292 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7294 2017-09-27 Richard Biener <rguenther@suse.de>
7296 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7297 --param loop-block-tile-size=0 to disable tiling.
7299 2017-09-27 Richard Biener <rguenther@suse.de>
7301 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7302 (graphite-max-nb-scop-params): Document special value zero.
7303 * domwalk.h (dom_walker::STOP): New symbolical constant.
7304 (dom_walker::dom_walker): Add optional parameter for bb to
7306 (dom_walker::~dom_walker): Declare.
7307 (dom_walker::before_dom_children): Document STOP return value.
7308 (dom_walker::m_user_bb_to_rpo): New member.
7309 (dom_walker::m_bb_to_rpo): Likewise.
7310 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7311 mapping here if not provided by the user.
7312 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7313 provided by the user.
7314 (dom_walker::STOP): Define.
7315 (dom_walker::walk): Do not compute bb to RPO mapping here.
7316 Support STOP return value from before_dom_children to stop
7318 * graphite-optimize-isl.c (optimize_isl): If the schedule
7319 is the same still generate code if -fgraphite-identity
7320 or -floop-parallelize-all are given.
7321 * graphite-scop-detection.c: Include cfganal.h.
7322 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7324 (gather_bbs::before_dom_children): Return STOP for BBs
7326 (build_scops): Compute bb to RPO mapping and pass it to
7327 the domwalk. Treat --param graphite-max-nb-scop-params=0
7328 as not limiting the number of params.
7329 * graphite.c (graphite_initialize): Remove limit on the
7330 number of basic-blocks in a function.
7331 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7332 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7333 default value of 10.
7335 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7337 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7338 Adjust code to eliminate needing to do the shift right 32-bits
7339 operation after XSCVDPSPN.
7341 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7343 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7344 ((X / Y) != 0 -> X >= Y): Likewise.
7346 2017-09-26 Carl Love <cel@us.ibm.com>
7348 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7349 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7350 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7351 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7352 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7353 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7354 definitions and overloading.
7355 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7356 statement for P9V_BUILTIN_XST_LEN_R.
7357 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7358 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7359 define_expand and define_insn for the instructions and builtins.
7360 * doc/extend.texi: Update the built-in documentation file for the new
7362 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7363 define_insn for the instructions
7365 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7368 * gcc/config/netbsd-protos.h: New file.
7369 * gcc/config/netbsd.c: New file.
7370 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7371 * gcc/config/t-netbsd: New file.
7372 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7373 (tmake_file) Add t-netbsd.
7374 (extra_objs) Add netbsd.o.
7376 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7380 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7382 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7384 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7385 sign extension from a vector register to a GPR by doing a 32-bit
7386 direct move and then an EXTSW.
7387 (extendsi<mode>2 splitter): Likewise.
7388 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7389 right or vector extract after doing XSCVDPSPN. Use
7390 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7392 (movdi_from_sf_zero_ext): Likewise.
7393 (reload_gpr_from_vsxsf): Likewise.
7394 (p8_mfvsrd_4_disf): Delete, no longer used.
7395 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7396 then needing to move the SFmode to a GPR to use the XSCVDPSP
7397 instruction instead of FRSP and XSCVDPSPN.
7398 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7399 it is adjacent to the other XSCVSPDP insns.
7400 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7401 SFmode to be in traditional Altivec registers.
7402 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7403 (vsx_xscvspdpn): Likewise.
7404 (vsx_xscvspdpn_scalar): Likewise.
7406 2017-09-26 Martin Jambor <mjambor@suse.cz>
7408 * tree-sra.c (compare_access_positions): Put integral types first,
7409 stabilize sorting of integral types, remove conditions putting
7410 non-full-precision integers last.
7411 (sort_and_splice_var_accesses): Disable scalarization if a
7412 non-integert would be represented by a non-full-precision integer.
7414 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7416 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7417 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7418 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7419 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7420 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7421 conditionals inside the function instead of around it. Call
7422 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7423 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7425 2017-09-26 Richard Biener <rguenther@suse.de>
7427 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7429 (scop_detection::build_scop_breadth): ... this. Removed.
7430 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7431 (scop_detection::harmful_stmt_in_bb): Likewise.
7432 (scop_detection::graphite_can_represent_stmt): Likewise.
7433 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7434 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7435 (scop_detection::can_represent_loop_1): ... this. Removed.
7436 (scop_detection::harmful_loop_in_region): Simplify after inlining
7437 the above and remove more quadraticness.
7438 (build_scops): Adjust.
7439 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7442 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7445 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7446 REGNO (base) == SP_REG if base is a REG.
7449 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7450 if it is different SSA_NAME.
7451 (optimize_range_tests_cmp_bitwise): New function.
7452 (optimize_range_tests): Call it.
7454 2017-09-26 Richard Biener <rguenther@suse.de>
7456 PR tree-optimization/82321
7457 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7458 for the def being inside the loop.
7460 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7462 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7464 * config/s390/s390-builtins.def: Fix constraint on op4.
7466 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7468 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7469 independent expanders.
7470 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7471 ("vec_ordered", "vec_unordered"): New expanders.
7473 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7475 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7478 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7480 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7481 vec_unpacks_lo_v16qi.
7482 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7484 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7486 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7487 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7488 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7490 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7492 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7494 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7496 ("*vec_slb<mode>"): New insn pattern.
7497 ("vec_shr_<mode>"): New expander.
7498 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7499 and force the shift count operand to V16QImode.
7500 ("vec_srb<mode>"): Set shift count mode to V16QI.
7502 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7504 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7505 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7506 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7508 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7511 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7513 2017-09-26 Richard Biener <rguenther@suse.de>
7515 PR tree-optimization/82320
7516 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7519 2017-09-25 Jeff Law <law@redhat.com>
7521 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7522 prototype for new argument.
7523 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7524 mostly extracted from rs6000_emit_allocate_stack.
7525 (rs6000_emit_probe_stack_range_stack_clash): New function.
7526 (rs6000_emit_allocate_stack): Call
7527 rs6000_emit_probe_stack_range_stack_clash as needed.
7528 (rs6000_emit_probe_stack_range): Add additional argument
7529 to call to gen_probe_stack_range{si,di}.
7530 (output_probe_stack_range): New.
7531 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7532 (output_probe_stack_range_stack_clash): New.
7533 (rs6000_emit_prologue): Emit notes into dump file as requested.
7534 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7535 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7536 Add additional operand and pass it to output_probe_stack_range.
7538 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7540 PR tree-optimization/82163
7541 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7542 (checking_verify_loop_closed_ssa): New parameter.
7543 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7544 (check_loop_closed_ssa_stmt): Delete.
7545 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7546 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7547 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7550 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7552 * brig-builtins.def: Treat HSAIL barrier builtins as
7553 setjmp/longjump style functions.
7555 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7557 * target.def (constant_alignment): New hook.
7558 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7559 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7560 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7561 * doc/tm.texi: Regenerate.
7562 * targhooks.h (default_constant_alignment): Declare.
7563 (constant_alignment_word_strings): Likewise.
7564 * targhooks.c (default_constant_alignment): New function.
7565 (constant_alignment_word_strings): Likewise.
7566 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7567 instead of CONSTANT_ALIGNMENT.
7568 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7569 (force_const_mem): Likewise.
7570 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7571 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7572 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7573 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7574 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7576 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7577 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7578 constant_alignment_word_strings.
7579 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7580 (CONSTANT_ALIGNMENT): Likewise.
7581 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7582 (arm_constant_alignment): New function.
7583 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7584 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7585 constant_alignment_word_strings.
7586 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7587 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7588 constant_alignment_word_strings.
7589 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7590 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7591 (cris_constant_alignment): New function.
7592 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7593 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7594 (epiphany_constant_alignment): New function.
7595 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7596 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7597 constant_alignment_word_strings.
7598 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7599 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7600 constant_alignment_word_strings.
7601 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7602 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7603 constant_alignment_word_strings.
7604 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7605 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7606 * config/i386/i386.c (ix86_constant_alignment): Make static.
7607 Use the same interface as the target hook.
7608 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7609 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7610 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7611 constant_alignment_word_strings.
7612 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7613 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7614 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7615 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7616 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7617 constant_alignment_word_strings.
7618 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7619 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7620 constant_alignment_word_strings.
7621 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7622 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7623 constant_alignment_word_strings.
7624 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7625 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7627 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7628 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7629 * config/mips/mips.c (mips_constant_alignment): New function.
7630 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7631 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7632 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7633 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7634 (mmix_constant_alignment): Make static. Use the same interface
7636 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7637 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7638 constant_alignment_word_strings.
7639 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7640 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7641 constant_alignment_word_strings.
7642 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7643 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7644 constant_alignment_word_strings.
7645 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7646 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7647 (rs6000_constant_alignment): New function.
7648 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7649 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7650 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7651 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7652 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7653 (rs6000_constant_alignment): New function.
7654 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7655 * config/s390/s390.c (s390_constant_alignment): New function.
7656 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7657 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7658 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7659 constant_alignment_word_strings.
7660 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7661 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7662 (sparc_constant_alignment): New function.
7663 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7664 * config/spu/spu.c (spu_constant_alignment): New function.
7665 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7666 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7667 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7668 constant_alignment_word_strings.
7669 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7670 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7671 constant_alignment_word_strings.
7672 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7673 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7674 constant_alignment_word_strings.
7675 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7676 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7677 (visium_constant_alignment): New function.
7678 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7679 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7680 (xtensa_constant_alignment): New function.
7681 * system.h (CONSTANT_ALIGNMENT): Poison.
7683 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7685 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7686 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7687 (rs6000_builtin_valid_without_lhs): New helper function.
7688 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7689 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7691 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7693 * target.h (vec_perm_indices): Use unsigned short rather than
7695 (auto_vec_perm_indices): Likewise.
7696 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7697 Use unsigned int rather than unsigned char.
7698 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7700 2017-09-25 Richard Biener <rguenther@suse.de>
7702 * cfgloop.h (sort_sibling_loops): Declare.
7703 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7704 (sort_sibling_loops): New function sorting the sibling loop list
7706 * graphite.c (graphite_transform_loops): Sort sibling loops.
7708 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7710 * target.def (vec_perm_const_ok): Change sel parameter to
7712 * optabs-query.c (can_vec_perm_p): Update accordingly.
7713 * doc/tm.texi: Regenerate.
7714 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7715 auto_vec_perm_indices and remove separate nelt field.
7716 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7717 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7718 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7719 (aarch64_expand_vec_perm_const): Update accordingly.
7720 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7721 to vec_perm_indices.
7722 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7723 auto_vec_perm_indices and remove separate nelt field.
7724 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7725 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7726 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7728 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7729 to vec_perm_indices.
7730 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7731 sel to vec_perm_indices.
7732 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7733 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7734 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7736 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7739 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7742 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7743 on the FUNCTION_DECL function context if it has a DIE that is a
7746 2017-09-25 Richard Biener <rguenther@suse.de>
7748 PR tree-optimization/82285
7749 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7752 2017-09-25 Tom de Vries <tom@codesourcery.com>
7756 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7759 2017-09-25 Richard Biener <rguenther@suse.de>
7761 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7762 ISL errors other than isl_error_quota happen. Dump if the
7763 schedule is the same.
7764 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7765 errors instead of aborting inside ISL.
7767 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7770 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7771 of libgcc_eh for m64.
7772 * config/i386/darwin64.h: Likewise.
7774 2017-09-25 Richard Biener <rguenther@suse.de>
7777 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7778 attribute for incomplete types nor twice for complete ones.
7780 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7783 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7784 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7787 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7790 * config/i386/i386.opt (mprefer-avx256): Use
7791 ix86_target_flags variable.
7792 * config/i386/i386.c (ix86_target_string): Move
7793 -mprefer-avx256 to flag2_opts.
7795 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7798 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7799 and x != -1 | y != -1 into (x & y) != -1.
7801 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7803 * config.gcc: Add new case statement to set
7804 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7805 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7806 s390x-*-linux* case statements. Added aarch64 to the list of
7807 supported architectures.
7809 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7811 PR tree-optimization/82289
7812 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7813 STMT_VINFO_RELEVANT_P.
7815 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7816 Alan Hayward <alan.hayward@arm.com>
7817 David Sherwood <david.sherwood@arm.com>
7819 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7820 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7821 (extract_range_from_binary_expr_1): Don't call
7822 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7824 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7825 Alan Hayward <alan.hayward@arm.com>
7826 David Sherwood <david.sherwood@arm.com>
7828 * target.def (preferred_vector_alignment): New hook.
7829 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7831 * doc/tm.texi: Regenerate.
7832 * targhooks.h (default_preferred_vector_alignment): Declare.
7833 * targhooks.c (default_preferred_vector_alignment): New function.
7834 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7836 (DR_TARGET_ALIGNMENT): New macro.
7837 (aligned_access_p): Update commentary.
7838 (vect_known_alignment_in_bytes): New function.
7839 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7841 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7842 Calculate the misalignment based on the target alignment rather than
7844 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7845 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7846 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7847 the target alignment, rather than masking the element misalignment
7848 with the number of elements in a vector. Also use the target
7849 alignment when calculating the maximum number of peels.
7850 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7851 instead of TYPE_ALIGN_UNIT.
7852 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7853 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7854 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7855 (vect_create_data_ref_ptr): Likewise.
7856 (vect_setup_realignment): Realign by ANDing with
7857 -DR_TARGET_MISALIGNMENT.
7858 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7859 the number of peels based on DR_TARGET_ALIGNMENT.
7860 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7861 with the guaranteed alignment boundary when deciding whether
7863 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7864 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7865 (ensure_base_align): Remove stmt_info parameter. Get the
7866 target base alignment from DR_TARGET_ALIGNMENT.
7867 (vectorizable_store): Update call accordingly. Interpret
7868 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7870 (vectorizable_load): Likewise.
7872 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7873 Alan Hayward <alan.hayward@arm.com>
7874 David Sherwood <david.sherwood@arm.com>
7876 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7877 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7878 (vect_enhance_data_refs_alignment): Likewise.
7880 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7882 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7883 error. Only quit immediately if parsing is complete.
7884 (BEGIN): Initialize fatal_err and parse_done.
7885 (begin fpu, end fpu): Check number of arguments.
7886 (begin arch, end arch): Likewise.
7887 (begin cpu, end cpu): Likewise.
7888 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7889 (optalias): Likewise.
7891 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7893 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7894 * config/arm/arm-isa.h: Delete. Move definitions to ...
7895 * arm-cpus.in: ... here. Use new feature and fgroup values.
7896 * config/arm/arm.c (arm_option_override): Use lower case for feature
7898 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7899 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7900 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7902 (print_isa_bits_for): New function.
7903 (gen_isa): New function.
7904 (gen_comm_data): Use print_isa_bits_for.
7905 (define feature): New keyword.
7906 (define fgroup): New keyword.
7907 * config/arm/t-arm (TM_H): Remove.
7908 (GTM_H): Add arm-isa.h.
7909 (arm-isa.h): Add rule to generate file.
7910 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7911 case for feature bit names.
7913 2017-09-22 Richard Biener <rguenther@suse.de>
7915 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7917 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7918 print of no dependency loops ...
7919 * graphite.c (graphite_transform_loops): ... here.
7920 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7922 (same_close_phi_node, remove_duplicate_close_phi,
7923 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7924 (canonicalize_loop_closed_ssa): ... here and simplify.
7925 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7926 (optimize_isl): Use dump_printf_loc to tell when we stopped
7927 optimizing because of an ISL timeout.
7929 2017-09-22 Richard Biener <rguenther@suse.de>
7931 PR tree-optimization/82291
7932 * tree-if-conv.c (predicate_mem_writes): Make sure to
7933 remove writes in blocks predicated with false.
7935 2017-09-22 Richard Biener <rguenther@suse.de>
7937 * sese.c: Include cfganal.h.
7938 (if_region_set_false_region): Remove.
7939 (create_if_region_on_edge): Likewise.
7940 (move_sese_in_condition): Re-implement without destroying
7943 2017-09-22 Richard Biener <rguenther@suse.de>
7945 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7946 Verify both BBs contain loop PHI nodes before dispatching to
7948 (graphite_regenerate_ast_isl): Do not recompute dominators,
7949 do not verify three times. Restructure for clarity.
7950 * graphite-scop-detection.c (same_close_phi_node,
7951 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7952 defined_in_loop_p, canonicalize_loop_closed_ssa,
7953 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7954 checking and SSA rewrite, move to ...
7955 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7956 (graphite_initialize): Do not pass in ctx, do not reset the
7957 SCEV cache, compute only dominators.
7958 (graphite_transform_loops): Allocate ISL ctx after
7959 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7960 Maintain post-dominators only around build_scops.
7961 * sese.c (if_region_set_false_region): Make static. Free
7962 and recompute dominators.
7963 (move_sese_in_condition): Assert we don't get called with
7964 post-dominators computed.
7965 * sese.h (if_region_set_false_region): Remove.
7967 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7969 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7970 mode attribute for TARGET_AVX512VL.
7972 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7974 * config/i386/i386.opt (mprefer-avx256): New option.
7975 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7977 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7978 for TARGET_PREFER_AVX256.
7979 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7981 2017-09-21 Jeff Law <law@redhat.com>
7983 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7984 Fix dump output if the only stack space is for pushed registers.
7986 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7988 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7991 2017-09-21 Martin Sebor <msebor@redhat.com>
7994 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7995 code (in C++) or code that triggers warnings.
7997 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7999 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8001 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8003 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8004 * sched-rgn.c: Ditto.
8005 * sel-sched-ir.c: Ditto.
8007 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8009 * toplev.h (set_random_seed): Adjust return type.
8010 * toplev.c (init_local_tick): Move eager initialization of random_seed
8011 to get_random_seed. Adjust comment.
8012 (init_random_seed): Inline to get_random_seed, delete.
8013 (get_random_seed): Initialize random_seed lazily.
8014 (set_random_seed): Do not return previous value.
8015 (print_switch_value): Do not call get_random_seed.
8017 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8019 * cgraph.c (delete_function_version): New, broken out from...
8020 (cgraph_node::delete_function_version): ...here. Rename to
8021 cgraph_node::delete_function_version_by_decl. Update all uses.
8022 (cgraph_node::remove): Call delete_function_version.
8024 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8027 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8028 VAR_DECLs to which addressable non-volatile parameters are mapped
8029 and for id->retvar after the return value assignment. Clear
8030 id->retval and id->retbnd after inlining.
8032 2017-09-21 Richard Biener <rguenther@suse.de>
8034 PR tree-optimization/82276
8035 PR tree-optimization/82244
8036 * tree-vrp.c (build_assert_expr_for): Set
8037 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8039 (remove_range_assertions): Revert earlier change.
8041 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8044 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8046 2017-09-21 Richard Biener <rguenther@suse.de>
8048 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8049 Restore valid IL after code generation errors.
8050 * graphite.c (graphite_transform_loops): Diagnose code
8051 generation issues as MSG_MISSED_OPTIMIZATION and continue
8052 with processing SCOPs.
8054 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8055 Alan Hayward <alan.hayward@arm.com>
8056 David Sherwood <david.sherwood@arm.com>
8058 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8059 rather than choosing between plus_constant and gen_rtx_<CODE>.
8060 * expr.c (emit_push_insn): Likewise.
8061 (expand_expr_real_2): Likewise.
8063 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8064 Alan Hayward <alan.hayward@arm.com>
8065 David Sherwood <david.sherwood@arm.com>
8067 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8069 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8070 Alan Hayward <alan.hayward@arm.com>
8071 David Sherwood <david.sherwood@arm.com>
8073 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8074 calling tree_to_uhwi.
8076 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8077 Alan Hayward <alan.hayward@arm.com>
8078 David Sherwood <david.sherwood@arm.com>
8080 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8081 INTEGER_CST rather than a negative test for ADDR_EXPR.
8083 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8084 Alan Hayward <alan.hayward@arm.com>
8085 David Sherwood <david.sherwood@arm.com>
8087 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8088 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8090 2017-09-21 Richard Biener <rguenther@suse.de>
8092 PR tree-optimization/71351
8093 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8094 graphite_create_new_loop_guard): Remove, fold remaining parts
8096 (translate_isl_ast_node_for): ... here and simplify.
8098 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8101 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8102 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8103 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8104 alternative always use QI mode, for -Os imov (=R,R) alternative
8105 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8108 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8109 Jeff Law <law@redhat.com>
8111 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8112 (allocate_stack_space): New function, partially extracted from
8114 (s390_emit_prologue): Track offset to most recent stack probe.
8115 Code to allocate space moved into allocate_stack_space.
8116 Dump actions when no stack is allocated.
8117 (s390_prologue_plus_offset): New function.
8118 (s390_emit_stack_probe): Likewise.
8120 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8122 * common.opt (Wa, Wl, Wp, g, gz=): Add
8124 (gno-column-info): Remove.
8125 (gcolumn-info): Drop RejectNegative.
8127 (gno-record-gcc-switches): Remove.
8128 (grecord-gcc-switches): Drop RejectNegative.
8129 (gno-split-dwarf): Remove.
8130 (gsplit-dwarf): Drop RejectNegative.
8131 (gno-strict-dwarf): Remove.
8132 (gstrict-dwarf): Drop RejectNegative.
8133 * config/darwin.opt (gfull, gused): Add RejectNegative.
8134 * dwarf2out.c (gen_producer_string): Drop
8135 gno-record-gcc-switches handler.
8136 * optc-gen.awk: Add g to prefixes with negative forms.
8137 * opts-common.c (remapping_prefix_p): New.
8138 (find_opt): Check it.
8139 (generate_canonical_option): Test g prefix.
8140 (option_map): Add -gno- mapping.
8141 (add_misspelling_candidates): Check remapping_prefix_p.
8143 2017-09-20 Jeff Law <law@redhat.com>
8145 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8146 thinko in stack clash protection support.
8148 * explow.c (compute_stack_clash_protection_loop_data): Use
8149 CONST_INT_P instead of explicit test. Verify object is a
8150 CONST_INT_P before looking at INTVAL.
8151 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8152 instead of explicit test.
8154 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8157 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8158 address instead of to r1 and r11.
8160 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8162 * config.gcc: Support "knm".
8163 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8164 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8166 * config/i386/i386.c (m_KNM): Define.
8167 (processor_target_table): Add "knm".
8169 (ix86_option_override_internal): Add "knm".
8170 (ix86_issue_rate): Add PROCESSOR_KNM.
8171 (ix86_adjust_cost): Ditto.
8172 (ia32_multipass_dfa_lookahead): Ditto.
8173 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8174 (fold_builtin_cpu): Add M_INTEL_KNM.
8175 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8176 (processor_type): Add PROCESSOR_KNM.
8177 * config/i386/x86-tune.def: Add m_KNM.
8178 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8180 2017-09-20 Richard Biener <rguenther@suse.de>
8182 PR tree-optimization/80213
8183 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8184 are allowed in empty BBs as well.
8185 (canonicalize_loop_closed_ssa): Also look for other complex
8187 (scop_detection::get_sese): Include the loop-closed PHI block
8189 (scop_detection::merge_sese): Remove code adding extra blocks.
8190 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8191 (build_scops): Assert the final returned scop is invalid.
8193 2017-09-20 Richard Biener <rguenther@suse.de>
8195 PR tree-optimization/82264
8196 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8198 (vn_phi_lookup): Likewise.
8199 (vn_phi_insert): Likewise.
8201 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8203 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8204 that fits into uhwi or shwi, add DW_AT_const_value regardless
8205 of early_dwarf without going through RTL, using add_AT_unsigned
8208 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8209 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8210 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8211 (DEBUG_MACRO_SECTION): Likewise.
8212 (DEBUG_LTO_MACRO_SECTION): Likewise.
8213 (DEBUG_STR_DWO_SECTION): Likewise.
8214 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8215 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8216 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8217 (DEBUG_LTO_LINE_STR_SECTION): Define.
8218 (init_sections_and_labels): Initialize debug_line_str_section
8219 variable. Initialize debug_loc_section for -gdwarf-5 to
8220 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8222 2017-09-20 Richard Biener <rguenther@suse.de>
8224 * graphite-sese-to-poly.c (extract_affine): Properly handle
8225 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8227 2017-09-20 Richard Biener <rguenther@suse.de>
8229 PR tree-optimization/81373
8230 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8231 Force SESE live-out defs to be handled even if they are
8234 2017-09-19 Jeff Law <law@redhat.com>
8236 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8237 nothing for stack adjustments with REG_STACK_CHECK.
8238 * sched-deps.c (parse_add_or_inc): Reject insns with
8239 REG_STACK_CHECK from dependency breaking.
8240 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8241 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8242 * reg-notes.def (STACK_CHECK): New note.
8244 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8245 (ix86_expand_prologue): Dump stack clash info as needed.
8246 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8248 * function.c (dump_stack_clash_frame_info): New function.
8249 * function.h (dump_stack_clash_frame_info): Prototype.
8250 (enum stack_clash_probes): New enum.
8252 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8253 flag_stack_clash_protection.
8254 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8255 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8256 (arm_frame_pointer_required): Likewise.
8257 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8258 (ia64_expand_prologue): Likewise.
8259 * config/mips/mips.c (mips_expand_prologue): Likewise.
8260 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8261 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8262 (sparc_flat_expand_prologue): Likewise.
8263 * config/spu/spu.c (spu_expand_prologue): Likewise.
8265 * explow.c: Include "params.h".
8266 (anti_adjust_stack_and_probe_stack_clash): New function.
8267 (get_stack_check_protect): Likewise.
8268 (compute_stack_clash_protection_loop_data): Likewise.
8269 (emit_stack_clash_protection_loop_start): Likewise.
8270 (emit_stack_clash_protection_loop_end): Likewise.
8271 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8272 Use anti_adjust_stack_and_probe_stack_clash.
8273 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8274 (emit_stack_clash_protection_loop_start): Likewise.
8275 (emit_stack_clash_protection_loop_end): Likewise.
8276 * rtl.h (get_stack_check_protect): Prototype.
8277 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8278 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8279 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8281 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8283 * doc/tm.texi: Rebuilt.
8284 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8285 get_stack_check_protect.
8286 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8287 * config/arm/arm.c (arm_expand_prologue): Likewise.
8288 (arm_frame_pointer_required): Likewise.
8289 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8290 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8291 * config/mips/mips.c (mips_expand_prologue): Likewise.
8292 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8293 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8294 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8295 (sparc_flat_expand_prologue): Likewise.
8297 * common.opt (-fstack-clash-protection): New option.
8298 * flag-types.h (enum stack_check_type): Note difference between
8299 -fstack-check= and -fstack-clash-protection.
8300 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8301 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8302 * toplev.c (process_options): Issue warnings/errors for cases
8303 not handled with -fstack-clash-protection.
8304 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8305 (-fstack-check): Note additional problem with -fstack-check=generic.
8306 Note that -fstack-check is primarily for Ada and refer users
8307 to -fstack-clash-protection for stack-clash-protection.
8308 Document new params for stack clash protection.
8310 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8312 * config/i386/i386.c (ix86_split_long_move): Do not handle
8313 address used for LEA in a special way.
8315 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8317 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8319 2017-09-19 Martin Sebor <msebor@redhat.com>
8322 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8323 of incompatible types.
8325 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8327 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8328 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8329 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8330 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8332 2017-09-19 Richard Biener <rguenther@suse.de>
8334 PR tree-optimization/82244
8335 * tree-vrp.c (remove_range_assertions): Do not propagate
8336 a constant to abnormals but replace the assert with a copy.
8338 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8340 PR rtl-optimization/57878
8341 PR rtl-optimization/68988
8342 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8343 avoidance test involving non_reload_pseudos. Move frequency test
8344 below the general fragmentation avoidance test.
8346 2017-09-19 Richard Biener <rguenther@suse.de>
8348 PR tree-optimization/69728
8349 * graphite-sese-to-poly.c (schedule_error): New global.
8350 (add_loop_schedule): Handle empty domain by failing the
8352 (build_original_schedule): Handle schedule_error.
8354 2017-09-19 Richard Biener <rguenther@suse.de>
8356 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8357 Do not iterate to sibling loops but only to siblings of inner
8360 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8363 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8364 that the registers are different.
8366 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8368 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8369 to processor_model and "amdfam17h" to arch_names_table.
8370 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8372 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8375 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8377 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8378 Alan Hayward <alan.hayward@arm.com>
8379 David Sherwood <david.sherwood@arm.com>
8381 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8383 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8385 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8386 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8387 vect_schedule_slp_instance.
8388 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8389 Update call to vect_slp_analyze_node_operations. Simplify return
8391 (vect_slp_analyze_bb_1): Update call accordingly.
8392 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8393 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8394 (vect_schedule_slp): Update call accordingly.
8396 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8397 Alan Hayward <alan.hayward@arm.com>
8398 David Sherwood <david.sherwood@arm.com>
8400 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8401 with types that aren't in fact scalar.
8403 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8405 * tree-vect-slp.c (vect_record_max_nunits): New function,
8407 (vect_build_slp_tree_1): ...here.
8408 (vect_build_slp_tree_2): Call it for phis too.
8410 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8412 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8413 to vect_get_vec_def_for_operand when getting the mask operand.
8415 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8416 Alan Hayward <alan.hayward@arm.com>
8417 David Sherwood <david.sherwood@arm.com>
8419 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8422 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8423 Alan Hayward <alan.hayward@arm.com>
8424 David Sherwood <david.sherwood@arm.com>
8426 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8427 calculation for vector booleans.
8429 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8430 Alan Hayward <alan.hayward@arm.com>
8431 David Sherwood <david.sherwood@arm.com>
8433 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8435 (vect_transform_stmt): ...here.
8436 (vect_analyze_stmt): Use it instead of calling
8437 vectorizable_live_operation directly.
8439 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8441 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8442 non-SIMT targets in acc vector loops.
8444 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8446 * configure.ac: Add arc and check if assembler supports gdwarf2.
8447 * configure: Regenerate.
8449 2017-09-18 Richard Biener <rguenther@suse.de>
8451 PR tree-optimization/82220
8452 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8453 epilogue niters from the min_profitable_iters compute.
8455 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8458 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8459 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8460 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8462 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8465 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8466 switching to a new text section.
8468 2017-09-18 Richard Biener <rguenther@suse.de>
8470 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8472 (build_alias_set): Reject aliases with no access function.
8474 2017-09-18 Richard Biener <rguenther@suse.de>
8476 PR tree-optimization/79622
8477 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8479 (build_cross_bb_scalars_use): Likewise.
8481 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8483 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8485 2017-09-18 Alan Modra <amodra@gmail.com>
8488 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8489 stack_pointer_rtx for count 0. Update comments. Break up
8490 large rtl expression.
8492 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8494 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8495 Increase to 20 bytes.
8496 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8497 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8498 or avx version of the stub.
8500 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8503 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8504 compute the minimum stack alignment. Also update preferred stack
8505 boundary for leaf functions.
8507 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8509 PR tree-optimization/82228
8510 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8513 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8515 * common/config/nds32/nds32-common.c
8516 (nds32_option_optimization_table): Refine formatting.
8517 (nds32_option_optimization_table): Use -fsched-pressure and
8518 -fomit-frame-pointer for specific optimization level.
8520 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8522 * config/nds32/nds32.c: Refine formatting and comments.
8523 * config/nds32/nds32.h: Likewise.
8524 * config/nds32/nds32.md: Likewise.
8525 * config/nds32/nds32-cost.c: Likewise.
8526 * config/nds32/nds32-isr.c: Likewise.
8527 * config/nds32/nds32-md-auxiliary.c: Likewise.
8528 * config/nds32/nds32-multiple.md: Likewise.
8529 * config/nds32/nds32-predicates.c: Likewise.
8531 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8532 Jakub Jelinek <jakub@redhat.com>
8534 Add support for -std=c++2a.
8535 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8537 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8539 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8542 * doc/extend.texi (Common Function Attributes): Add
8543 references to ARM, AArch64, and S/390 specific attributes.
8544 (Function Specific Option Pragmas): Add AArch64 and S/390
8545 to list of back ends that support the target pragma.
8547 2017-09-15 Nathan Sidwell <nathan@acm.org>
8549 * doc/standards.texi: Fix C++17 description. Update URLs for
8552 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8554 * common.opt (Wcast-align=strict): New warning option.
8555 * doc/invoke.texi: Document -Wcast-align=strict.
8557 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8559 * cgraph.h (cgraph_thunk_info): Add comments.
8560 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8561 assert for VIRTUAL_* arguments stricter.
8563 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8565 PR tree-optimization/71026
8566 * match.pd: Move RDIV patterns from fold-const.c
8567 * fold-const.c (distribute_real_division): Removed.
8568 (fold_binary_loc): Remove calls to distribute_real_divison.
8570 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8572 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8573 c++1z and gnu++1z as deprecated. Change other references to
8574 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8575 Change -Wc++1z-compat to -Wc++17-compat.
8576 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8577 * dwarf2out.c (highest_c_language): Handle C++17.
8578 (gen_compile_unit_die): Likewise.
8580 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8582 PR rtl-optimization/82192
8583 * combine.c (make_extraction): Don't look through non-paradoxical
8584 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8587 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8588 Alan Hayward <alan.hayward@arm.com>
8589 David Sherwood <david.sherwood@arm.com>
8591 * target.def (function_arg_offset): New hook.
8592 * targhooks.h (default_function_arg_offset): Declare.
8593 * targhooks.c (default_function_arg_offset): New function.
8594 * function.c (locate_and_pad_parm): Use
8595 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8596 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8597 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8598 * doc/tm.texi: Regenerate.
8599 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8600 * config/spu/spu.c (spu_function_arg_offset): New function.
8601 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8602 * system.h (FUNCTION_ARG_OFFSET): Poison.
8604 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8605 Alan Hayard <alan.hayward@arm.com>
8606 David Sherwood <david.sherwood@arm.com>
8608 * target.def (truly_noop_truncation): New hook.
8609 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8610 than TRULY_NOOP_TRUNCATION.
8611 * hooks.h (hook_bool_uint_uint_true): Declare.
8612 * hooks.c (hook_bool_uint_uint_true): New function.
8613 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8614 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8615 * doc/tm.texi: Regenerate.
8616 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8617 rather than TRULY_NOOP_TRUNCATION in comments.
8618 (simplify_comparison): Likewise.
8619 (record_truncated_value): Likewise.
8620 * expmed.c (extract_bit_field_1): Likewise.
8621 (extract_split_bit_field): Likewise.
8622 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8623 instead of TRULY_NOOP_TRUNCATION.
8624 * function.c (assign_parm_setup_block): Likewise.
8625 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8626 * rtlhooks.c: Include target.h.
8627 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8628 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8629 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8630 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8631 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8632 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8633 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8634 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8635 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8636 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8637 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8638 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8639 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8640 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8641 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8642 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8643 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8644 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8645 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8646 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8647 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8648 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8649 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8650 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8651 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8652 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8653 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8654 rather than TRULY_NOOP_TRUNCATION in comments.
8655 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8656 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8657 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8658 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8659 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8660 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8661 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8662 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8663 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8664 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8665 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8666 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8667 rather than TRULY_NOOP_TRUNCATION in comments.
8668 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8669 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8670 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8671 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8672 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8673 TRULY_NOOP_TRUNCATION condition.
8674 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8675 (TRULY_NOOP_TRUNCATION): Delete.
8676 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8677 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8678 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8679 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8680 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8681 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8682 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8683 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8684 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8685 rather than TRULY_NOOP_TRUNCATION in comments.
8686 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8687 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8688 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8689 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8690 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8691 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8693 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8696 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8697 (*cmp_ior): Likewise.
8698 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8699 (*ior_scc_scc_cmp): Likewise.
8700 (*and_scc_scc): Likewise.
8701 (*and_scc_scc_cmp): Likewise.
8703 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8704 Alan Hayard <alan.hayward@arm.com>
8705 David Sherwood <david.sherwood@arm.com>
8707 * target.def (can_change_mode_class): New hook.
8708 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8709 (hard_regno_nregs): Likewise.
8710 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8711 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8712 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8713 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8714 (LOAD_EXTEND_OP): Update accordingly.
8715 * doc/tm.texi: Regenerate.
8716 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8717 CANNOT_CHANGE_MODE_CLASS.
8718 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8719 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8720 * combine.c (simplify_set): Update accordingly.
8721 * emit-rtl.c (validate_subreg): Likewise.
8722 * recog.c (general_operand): Likewise.
8723 * regcprop.c (mode_change_ok): Likewise.
8724 * reload1.c (choose_reload_regs): Likewise.
8725 (inherit_piecemeal_p): Likewise.
8726 * rtlanal.c (simplify_subreg_regno): Likewise.
8727 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8728 instead of CANNOT_CHANGE_MODE_CLASS.
8729 (reload_cse_simplify_operands): Likewise.
8730 * reload.c (push_reload): Use targetm.can_change_mode_class
8731 instead of CANNOT_CHANGE_MODE_CLASS.
8732 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8733 REG_CANNOT_CHANGE_MODE_P.
8734 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8735 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8736 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8737 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8738 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8739 (arm_can_change_mode_class): New function.
8740 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8741 than CANNOT_CHANGE_MODE_CLASS in comments.
8742 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8743 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8744 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8745 (ix86_can_change_mode_class): ...this new function, inverting the
8746 sense of the return value.
8747 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8748 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8749 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8750 (ia64_can_change_mode_class): New function.
8751 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8752 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8753 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8754 (m32c_can_change_mode_class): ...this new function, inverting the
8755 sense of the return value.
8756 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8757 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8758 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8759 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8760 (mips_can_change_mode_class): ...this new function, inverting the
8761 sense of the return value.
8762 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8763 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8764 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8765 (msp430_can_change_mode_class): New function.
8766 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8767 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8768 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8769 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8770 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8771 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8772 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8773 (pa_cannot_change_mode_class): Replace with...
8774 (pa_can_change_mode_class): ...this new function, inverting the
8775 sense of the return value.
8776 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8777 than CANNOT_CHANGE_MODE_CLASS in comments.
8778 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8779 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8780 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8781 (pdp11_cannot_change_mode_class): Replace with...
8782 (pdp11_can_change_mode_class): ...this new function, inverting the
8783 sense of the return value.
8784 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8785 * config/powerpcspe/powerpcspe-protos.h
8786 (rs6000_cannot_change_mode_class_ptr): Delete.
8787 * config/powerpcspe/powerpcspe.c
8788 (rs6000_cannot_change_mode_class_ptr): Delete.
8789 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8790 (rs6000_option_override_internal): Assign to
8791 targetm.can_change_mode_class instead of
8792 rs6000_cannot_change_mode_class_ptr.
8793 (rs6000_cannot_change_mode_class): Replace with...
8794 (rs6000_can_change_mode_class): ...this new function, inverting the
8795 sense of the return value.
8796 (rs6000_debug_cannot_change_mode_class): Replace with...
8797 (rs6000_debug_can_change_mode_class): ...this new function.
8798 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8799 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8800 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8801 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8802 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8804 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8805 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8806 (rs6000_option_override_internal): Assign to
8807 targetm.can_change_mode_class instead of
8808 rs6000_cannot_change_mode_class_ptr.
8809 (rs6000_cannot_change_mode_class): Replace with...
8810 (rs6000_can_change_mode_class): ...this new function, inverting the
8811 sense of the return value.
8812 (rs6000_debug_cannot_change_mode_class): Replace with...
8813 (rs6000_debug_can_change_mode_class): ...this new function.
8814 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8815 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8816 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8817 (s390_can_change_mode_class): ...this new function, inverting the
8818 sense of the return value.
8819 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8820 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8821 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8822 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8823 (sh_cannot_change_mode_class): Replace with...
8824 (sh_can_change_mode_class): ...this new function, inverting the
8825 sense of the return value.
8826 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8827 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8828 (sparc_can_change_mode_class): New function.
8829 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8830 * config/spu/spu.c (spu_can_change_mode_class): New function.
8831 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8832 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8833 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8834 (visium_can_change_mode_class): New function.
8835 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8837 2017-09-15 Richard Biener <rguenther@suse.de>
8839 PR tree-optimization/82217
8840 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8841 but not undefined case.
8843 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8846 * postreload.c (reload_cse_simplify_operands): Skip
8847 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8849 2017-09-15 Richard Biener <rguenther@suse.de>
8851 PR tree-optimization/68823
8852 * graphite-scop-detection.c (build_alias_set): If we have a
8853 possible dependence check whether we can handle them by just
8854 looking at the DRs DR_ACCESS_FNs.
8855 (build_scops): If build_alias_set fails, fail the SCOP.
8857 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8859 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8860 to support float128 built-in functions that require the ISA 3.0
8862 (BU_FLOAT128_3_HW): Likewise.
8863 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8865 (FMAF128): Likewise.
8867 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8868 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8869 floating point instructions.
8870 (rs6000_invalid_builtin): Likewise.
8871 (rs6000_builtin_mask_names): Likewise.
8872 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8873 (RS6000_BTM_FLOAT128_HW): Likewise.
8874 (RS6000_BTM_COMMON): Likewise.
8875 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8877 * doc/extend.texi (RS/6000 built-in functions): Document the
8878 IEEE 128-bit floating point square root and fused multiply-add
8881 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8883 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8884 reg (r2) isn't in the set of registers defined in the prologue.
8886 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8887 Alan Hayward <alan.hayward@arm.com>
8888 David Sherwood <david.sherwood@arm.com>
8890 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8891 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8892 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8893 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8894 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8896 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8897 max_vectorization_factor.
8898 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8900 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8901 Alan Hayward <alan.hayward@arm.com>
8902 David Sherwood <david.sherwood@arm.com>
8904 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8905 (vect_worthwhile_without_simd_p): Declare.
8906 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8907 (vectorizable_reduction): Use it.
8908 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8909 (vectorizable_operation): Likewise.
8911 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8912 Alan Hayward <alan.hayward@arm.com>
8913 David Sherwood <david.sherwood@arm.com>
8915 * tree-vectorizer.h (vect_get_num_copies): New function.
8916 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8917 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8918 (vectorizable_induction): Likewise.
8919 (vectorizable_live_operation): Likewise.
8920 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8921 (vectorizable_bswap): Likewise.
8922 (vectorizable_call): Likewise.
8923 (vectorizable_conversion): Likewise.
8924 (vectorizable_assignment): Likewise.
8925 (vectorizable_shift): Likewise.
8926 (vectorizable_operation): Likewise.
8927 (vectorizable_store): Likewise.
8928 (vectorizable_load): Likewise.
8929 (vectorizable_condition): Likewise.
8930 (vectorizable_comparison): Likewise.
8931 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8933 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8934 Alan Hayward <alan.hayward@arm.com>
8935 David Sherwood <david.sherwood@arm.com>
8937 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8938 of vect_init_vector.
8940 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8941 Alan Hayward <alan.hayward@arm.com>
8942 David Sherwood <david.sherwood@arm.com>
8944 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8945 an inline wrapper that provides a location.
8946 (gimple_build_vector): Likewise.
8947 * gimple-fold.c (gimple_build_vector_from_val): New function.
8948 (gimple_build_vector): Likewise.
8949 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8950 functions to build the initial value. Always return a gimple value.
8951 (get_initial_defs_for_reduction): Likewise. Only compute
8953 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8954 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8955 (vectorizable_induction): Use gimple_build_vector rather than
8958 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8959 Alan Hayward <alan.hayward@arm.com>
8960 David Sherwood <david.sherwood@arm.com>
8962 * target.h (vec_perm_indices): New typedef.
8963 (auto_vec_perm_indices): Likewise.
8964 * optabs-query.h: Include target.h
8965 (can_vec_perm_p): Take a vec_perm_indices *.
8966 * optabs-query.c (can_vec_perm_p): Likewise.
8967 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8968 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8969 * tree-vect-generic.c (lower_vec_perm): Likewise.
8970 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8971 (vect_grouped_load_supported): Likewise.
8972 (vect_shift_permute_load_chain): Likewise.
8973 (vect_permute_store_chain): Use auto_vec_perm_indices.
8974 (vect_permute_load_chain): Likewise.
8975 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8976 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8977 Update uses of can_vec_perm_p.
8978 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8979 mode with a number of elements. Take a vec_perm_indices *.
8980 (vect_create_epilog_for_reduction): Update accordingly.
8981 Use auto_vec_perm_indices.
8982 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8983 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8984 (vect_transform_slp_perm_load): Likewise.
8985 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8986 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8987 (vect_gen_perm_mask_checked): Likewise.
8988 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8989 (vect_gen_perm_mask_checked): Likewise.
8990 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8991 (vectorizable_store): Likewise.
8992 (vectorizable_load): Likewise.
8993 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8994 (vectorizable_bswap): Likewise.
8996 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8997 Alan Hayward <alan.hayward@arm.com>
8998 David Sherwood <david.sherwood@arm.com>
9000 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9001 * tree.c (build_vector): Likewise.
9002 (build_vector_from_ctor): Update accordingly.
9003 (build_vector_from_val): Likewise.
9004 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9005 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9006 * tree-vect-generic.c (add_rshift): Likewise.
9007 (expand_vector_divmod): Likewise.
9008 (optimize_vector_constructor): Likewise.
9009 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9010 (vect_transform_slp_perm_load): Likewise.
9011 (vect_schedule_slp_instance): Likewise.
9012 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9013 (vectorizable_call): Likewise.
9014 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9015 * expmed.c (make_tree): Likewise.
9016 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9017 a vector passed to build_vector.
9018 (fold_convert_const): Likewise.
9019 (exact_inverse): Likewise.
9020 (fold_ternary_loc): Likewise.
9021 (fold_relational_const): Likewise.
9022 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9023 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9024 (const_unop): Likewise. Store the reduction accumulator in a
9025 variable rather than an array.
9026 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9027 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9028 the new vector, rather than constructing it after the input arrays.
9029 (native_interpret_vector): Use auto_vec<tree> when building
9030 a vector passed to build_vector. Add elements in order.
9031 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9032 auto_vec<tree> when building a vector passed to build_vector.
9033 (vect_create_epilog_for_reduction): Likewise.
9034 (vectorizable_induction): Likewise.
9035 (get_initial_def_for_reduction): Likewise. Fix indentation of
9037 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9039 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9040 passed to build_vector.
9042 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9043 Alan Hayward <alan.hayward@arm.com>
9044 David Sherwood <david.sherwood@arm.com>
9046 * tree-core.h (tree_base::u): Add an "nelts" field.
9047 (tree_vector): Use VECTOR_CST_NELTS as the length.
9048 * tree.c (tree_size): Likewise.
9049 (make_vector): Initialize VECTOR_CST_NELTS.
9050 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9051 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9052 TYPE_VECTOR_SUBPARTS.
9053 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9054 as the number of units, setting it from VECTOR_CST_NELTS.
9055 (const_vector_from_tree): Likewise.
9056 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9057 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9058 (fold_negate_expr_1): Likewise.
9059 (fold_convert_const): Likewise.
9060 (const_binop): Likewise. Differentiate the number of output and
9062 (const_unop): Likewise.
9063 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9064 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9065 in cases that did the opposite.
9067 2017-09-14 Richard Biener <rguenther@suse.de>
9069 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9072 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9074 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9076 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9079 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9080 if and where to split a bb, except for splitting before debug insn
9081 sequences followed by non-label real insn. Delete debug insns
9082 in between basic blocks.
9084 * combine.c (make_compound_operation_int): Formatting fixes.
9086 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9087 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9088 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9089 * config/sol2.h (LINK_EH_SPEC): Likewise.
9090 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9091 * config/s390/linux.h (LINK_SPEC): Likewise.
9092 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9093 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9094 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9095 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9096 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9097 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9098 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9099 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9101 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9103 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9105 (ENDFILE_LINUX_SPEC): Likewise.
9106 (LINK_EH_SPEC): Likewise.
9107 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9108 (LINK_OS_LINUX_SPEC32): Likewise.
9109 (LINK_OS_LINUX_SPEC64): Likewise.
9110 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9111 (LINK_OS_LINUX_SPEC): Likewise.
9113 2017-09-13 Martin Liska <mliska@suse.cz>
9116 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9117 CASE_HIGH is NULL_TREE.
9119 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9120 Alan Hayward <alan.hayward@arm.com>
9121 David Sherwood <david.sherwood@arm.com>
9123 * target.def (secondary_memory_needed): New hook.
9124 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9125 instead of SECONDARY_MEMORY_NEEDED.
9126 (secondary_memory_needed_mode): Likewise.
9127 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9128 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9129 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9130 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9131 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9132 * doc/tm.texi: Regenerate.
9133 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9134 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9135 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9136 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9137 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9138 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9139 mode argument first and change the reg_class arguments to reg_class_t.
9140 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9141 Make static. Update the call to inline_secondary_memory_needed.
9142 (ix86_register_move_cost): Update the call to
9143 inline_secondary_memory_needed.
9144 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9145 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9147 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9148 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9150 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9151 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9152 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9153 and match hook interface. Add comment from mips.h.
9154 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9155 * config/mmix/mmix.md (truncdfsf2): Refer to
9156 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9158 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9159 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9160 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9161 (pa_secondary_memory_needed): New function.
9162 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9163 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9164 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9165 (pdp11_secondary_memory_needed): Make static and match hook interface.
9166 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9167 * config/powerpcspe/powerpcspe-protos.h
9168 (rs6000_secondary_memory_needed_ptr): Delete.
9169 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9171 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9172 (rs6000_option_override_internal): Assign to
9173 targetm.secondary_memory_needed rather than
9174 rs6000_secondary_memory_needed_ptr.
9175 (rs6000_secondary_memory_needed): Match hook interface.
9176 (rs6000_debug_secondary_memory_needed): Likewise.
9177 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9178 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9179 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9180 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9181 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9182 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9184 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9185 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9186 (rs6000_option_override_internal): Assign to
9187 targetm.secondary_memory_needed rather than
9188 rs6000_secondary_memory_needed_ptr.
9189 (rs6000_secondary_memory_needed): Match hook interface.
9190 (rs6000_debug_secondary_memory_needed): Likewise.
9191 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9192 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9193 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9194 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9195 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9196 (sparc_secondary_memory_needed): New function.
9197 * lra-constraints.c (check_and_process_move): Refer to
9198 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9200 (curr_insn_transform): Likewise.
9201 (process_alt_operands): Use targetm.secondary_memory_needed
9202 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9203 (check_secondary_memory_needed_p): Likewise.
9204 (choose_split_class): Likewise.
9205 * reload.c: Unconditionally include code that was previously
9206 conditional on SECONDARY_MEMORY_NEEDED.
9207 (push_secondary_reload): Use targetm.secondary_memory_needed
9208 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9209 (push_reload): Likewise.
9210 * reload1.c: Unconditionally include code that was previously
9211 conditional on SECONDARY_MEMORY_NEEDED.
9212 (choose_reload_regs): Use targetm.secondary_memory_needed
9213 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9214 (gen_reload): Likewise.
9215 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9217 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9218 Alan Hayward <alan.hayward@arm.com>
9219 David Sherwood <david.sherwood@arm.com>
9221 * target.def (secondary_memory_needed_mode): New hook:
9222 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9223 * targhooks.h (default_secondary_memory_needed_mode): New function.
9224 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9225 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9226 * doc/tm.texi: Regenerate.
9227 * lra-constraints.c (check_and_process_move): Use
9228 targetm.secondary_memory_needed_mode instead of
9229 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9230 (curr_insn_transform): Likewise.
9231 * reload.c (get_secondary_mem): Likewise.
9232 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9233 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9235 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9236 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9237 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9238 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9239 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9241 * config/powerpcspe/powerpcspe-protos.h
9242 (rs6000_secondary_memory_needed_mode): Delete.
9243 * config/powerpcspe/powerpcspe.c
9244 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9245 (rs6000_secondary_memory_needed_mode): Make static.
9246 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9247 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9249 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9251 (rs6000_secondary_memory_needed_mode): Make static.
9252 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9253 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9254 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9255 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9256 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9258 (sparc_secondary_memory_needed_mode): New function.
9259 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9261 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9263 * config/aarch64/constraints.md (Umq): New constraint.
9264 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9266 (mov<mode>): Update condition.
9268 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9270 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9271 when bitposition is the same.
9273 2017-09-13 Richard Biener <rguenther@suse.de>
9275 * dwarf2out.c (output_die_symbol): Remove.
9276 (output_die): Do not output a DIEs symbol.
9278 2017-09-13 Richard Biener <rguenther@suse.de>
9281 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9282 default-def to avoid breaking iterator update with the weird
9283 interaction with cgraph_update_edges_for_call_stmt_node.
9285 2017-09-13 Richard Biener <rguenther@suse.de>
9287 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9288 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9289 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9290 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9292 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9294 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9295 Disable pc relative literal load irrespective of
9296 TARGET_FIX_ERR_A53_84341 for default.
9298 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9300 * config/sparc/sparc.c (output_return): Output the source location of
9301 the insn in the delay slot, if any.
9302 (output_sibcall): Likewise.
9304 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9307 * common.opt (-static-pie): New alias.
9308 (shared): Negate static-pie.
9309 (-no-pie): Update help text.
9311 (static-pie): New option.
9312 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9313 -static-pie support.
9314 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9315 (LINK_EH_SPEC): Likewise.
9316 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9317 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9318 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9319 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9320 (init_gcc_specs): Likewise.
9321 (init_spec): Likewise.
9322 (display_help): Update help message for -pie.
9323 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9326 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9328 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9329 (movdi_aarch64): Likewise.
9330 (movti_aarch64): Likewise.
9332 2017-09-12 Simon Wright <simon@pushface.org>
9335 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9336 calculation of the minor version, always output as 0.
9338 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9341 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9342 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9343 on it early, rather than manual conversion late. For
9344 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9345 instead of performing manual conversion.
9347 2017-09-12 Carl Love <cel@us.ibm.com>
9349 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9350 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9352 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9353 VMULOSW): Add definitions.
9354 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9355 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9356 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9357 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9358 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9360 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9362 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9364 (movti_aarch64): Likewise.
9365 (movdf_aarch64): Likewise.
9366 (movtf_aarch64): Likewise.
9367 (load_pairdi): Likewise.
9368 (store_pairdi): Likewise.
9369 (load_pairdf): Likewise.
9370 (store_pairdf): Likewise.
9371 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9372 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9373 (ldr_got_small_<mode>): Likewise.
9374 (ldr_got_small_28k_<mode>): Likewise.
9375 (ldr_got_tiny): Likewise.
9376 * config/aarch64/iterators.md (ldst_sz): New.
9377 (ldpstp_sz): Likewise.
9378 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9380 (thunderx_load): Split load_8 to load_16.
9381 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9383 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9384 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9385 (xgene1_store_pair): Split store_8 to store_16.
9386 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9387 (falkor_st_0_st_sd): Split store_8 to store_16.
9389 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9391 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9392 and store1/2/3/4 to store_4/8/12/16.
9393 * config/aarch64/aarch64.md: Update for rename.
9394 * config/arm/arm.md: Likewise.: Likewise.
9395 * config/arm/arm.c: Likewise.
9396 * config/arm/thumb1.md: Likewise.
9397 * config/arm/thumb2.md: Likewise.
9398 * config/arm/vfp.md: Likewise.
9399 * config/arm/arm-generic.md: Likewise.
9400 * config/arm/arm1020e.md: Likewise.
9401 * config/arm/arm1026ejs.md: Likewise.
9402 * config/arm/arm1136jfs.md: Likewise.
9403 * config/arm/arm926ejs.md: Likewise.
9404 * config/arm/cortex-a15.md: Likewise.
9405 * config/arm/cortex-a17.md: Likewise.
9406 * config/arm/cortex-a5.md: Likewise.
9407 * config/arm/cortex-a53.md: Likewise.
9408 * config/arm/cortex-a57.md: Likewise.
9409 * config/arm/cortex-a7.md: Likewise.
9410 * config/arm/cortex-a8.md: Likewise.
9411 * config/arm/cortex-a9.md: Likewise.
9412 * config/arm/cortex-m4.md: Likewise.
9413 * config/arm/cortex-m7.md: Likewise.
9414 * config/arm/cortex-r4.md: Likewise.
9415 * config/arm/exynos-m1.md: Likewise.
9416 * config/arm/fa526.md: Likewise.
9417 * config/arm/fa606te.md: Likewise.
9418 * config/arm/fa626te.md: Likewise.
9419 * config/arm/fa726te.md: Likewise.
9420 * config/arm/fmp626.md: Likewise.
9421 * config/arm/iwmmxt.md: Likewise.
9422 * config/arm/ldmstm.md: Likewise.
9423 * config/arm/marvell-pj4.md: Likewise.
9424 * config/arm/xgene1.md: Likewise.
9425 * config/aarch64/thunderx.md: Likewise.
9426 * config/aarch64/thunderx2t99.md: Likewise.
9427 * config/aarch64/falkor.md: Likewise.
9429 2017-09-12 Martin Liska <mliska@suse.cz>
9431 * attribs.c (private_lookup_attribute): New function.
9432 * attribs.h (private_lookup_attribute): Declared here.
9433 (lookup_attribute): Called from this place.
9435 2017-09-12 Richard Biener <rguenther@suse.de>
9437 PR tree-optimization/82157
9438 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9439 stmts with side-effects.
9441 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9442 Alan Hayward <alan.hayward@arm.com>
9443 David Sherwood <david.sherwood@arm.com>
9445 * target.def (hard_regno_nregs): New hook.
9446 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9447 * targhooks.h (default_hard_regno_nregs): Declare.
9448 * targhooks.c (default_hard_regno_nregs): New function.
9449 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9450 (TARGET_HARD_REGNO_NREGS): ...this hook.
9451 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9452 (CLASS_MAX_NREGS): Likewise.
9453 * doc/tm.texi: Regenerate.
9454 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9455 instead of HARD_REGNO_NREGS.
9456 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9457 HARD_REGNO_NREGS in the comment.
9458 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9459 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9460 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9461 Return an unsigned int.
9462 (TARGET_HARD_REGNO_NREGS): Redefine.
9463 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9464 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9465 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9466 (arc_hard_regno_nregs): New function.
9467 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9468 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9469 (arm_hard_regno_nregs): New function.
9470 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9471 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9472 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9473 (TARGET_HARD_REGNO_NREGS): Redefine.
9474 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9475 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9476 (HARD_REGNO_NREGS): Delete.
9477 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9478 (cr16_hard_regno_nregs): New function.
9479 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9480 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9481 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9482 (cris_hard_regno_nregs): New function.
9483 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9484 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9485 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9486 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9487 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9488 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9489 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9490 (frv_hard_regno_nregs): Make static. Take and return an
9492 (frv_class_max_nregs): Remove outdated copy of documentation.
9493 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9494 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9495 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9496 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9497 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9498 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9499 (TARGET_HARD_REGNO_NREGS): Redefine.
9500 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9501 (CLASS_MAX_NREGS): Update comment.
9502 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9503 (ia64_hard_regno_nregs): New function.
9504 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9505 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9506 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9507 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9508 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9510 (m32c_hard_regno_nregs): Likewise. Make static.
9511 (TARGET_HARD_REGNO_NREGS): Redefine.
9512 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9513 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9514 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9515 (m68k_hard_regno_nregs): New function.
9516 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9517 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9518 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9519 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9520 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9521 Take and return an unsigned int.
9522 (TARGET_HARD_REGNO_NREGS): Redefine.
9523 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9524 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9525 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9526 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9527 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9528 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9529 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9530 (msp430_hard_regno_nregs): Make static. Take and return an
9532 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9533 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9534 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9535 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9536 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9537 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9538 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9539 (TARGET_HARD_REGNO_NREGS): Redefine.
9540 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9541 (PA_HARD_REGNO_NREGS): ...this.
9542 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9543 (PA_HARD_REGNO_NREGS): ...this.
9544 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9545 (pa_hard_regno_nregs): New function.
9546 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9547 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9548 (pdp11_hard_regno_nregs): New function.
9549 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9550 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9551 (rs6000_hard_regno_nregs_hook): New function.
9552 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9553 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9554 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9555 Take and return an unsigned int. Move earlier in file.
9556 (TARGET_HARD_REGNO_NREGS): Redefine.
9557 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9558 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9559 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9560 (rl78_hard_regno_nregs): Make static. Take and return an
9562 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9563 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9564 (rs6000_hard_regno_nregs_hook): New function.
9565 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9566 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9567 (TARGET_HARD_REGNO_NREGS): Redefine.
9568 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9569 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9570 instead of HARD_REGNO_NREGS.
9571 (s390_hard_regno_nregs): New function.
9572 (s390_hard_regno_mode_ok): Add comment from s390.h.
9573 (TARGET_HARD_REGNO_NREGS): Redefine.
9574 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9575 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9576 (sh_hard_regno_nregs): New function.
9577 (sh_pass_in_reg_p): Use it.
9578 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9579 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9580 (sparc_hard_regno_nregs): New function.
9581 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9582 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9583 (spu_function_arg_advance): Use it, supplying a valid register number.
9584 (TARGET_HARD_REGNO_NREGS): Redefine.
9585 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9586 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9587 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9588 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9589 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9590 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9591 (CLASS_MAX_NREGS): Remove copy of old documentation.
9592 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9593 (visium_hard_regno_nregs): New function.
9594 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9595 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9596 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9597 xtensa_hard_regno_nregs): New function.
9598 * system.h (HARD_REGNO_NREGS): Poison.
9600 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9602 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9603 hard_regno_nregs instead of HARD_REGNO_NREGS.
9604 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9605 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9606 (c6x_expand_epilogue): Likewise.
9607 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9608 (frv_read_iacc_argument): Likewise.
9609 * config/sh/sh.c: Include regs.h.
9610 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9611 (regs_used): Likewise.
9612 (output_stack_adjust): Likewise.
9613 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9614 * expmed.c: Include regs.h.
9615 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9616 * ree.c: Include regs.h.
9617 (combine_reaching_defs): Use hard_regno_nregs instead of
9619 (add_removable_extension): Likewise.
9621 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9623 * regs.h (hard_regno_nregs): Turn into a function.
9624 (end_hard_regno): Update accordingly.
9625 * caller-save.c (setup_save_areas): Likewise.
9626 (save_call_clobbered_regs): Likewise.
9627 (replace_reg_with_saved_mem): Likewise.
9628 (insert_restore): Likewise.
9629 (insert_save): Likewise.
9630 * combine.c (can_change_dest_mode): Likewise.
9631 (move_deaths): Likewise.
9632 (distribute_notes): Likewise.
9633 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9634 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9635 (rs6000_split_multireg_move): Likewise.
9636 (rs6000_register_move_cost): Likewise.
9637 (rs6000_memory_move_cost): Likewise.
9638 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9639 (rs6000_split_multireg_move): Likewise.
9640 (rs6000_register_move_cost): Likewise.
9641 (rs6000_memory_move_cost): Likewise.
9642 * cselib.c (cselib_reset_table): Likewise.
9643 (cselib_lookup_1): Likewise.
9644 * emit-rtl.c (set_mode_and_regno): Likewise.
9645 * function.c (aggregate_value_p): Likewise.
9646 * ira-color.c (setup_profitable_hard_regs): Likewise.
9647 (check_hard_reg_p): Likewise.
9648 (calculate_saved_nregs): Likewise.
9649 (assign_hard_reg): Likewise.
9650 (improve_allocation): Likewise.
9651 (calculate_spill_cost): Likewise.
9652 * ira-emit.c (modify_move_list): Likewise.
9653 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9654 (ira_hard_reg_in_set_p): Likewise.
9655 * ira.c (setup_reg_mode_hard_regset): Likewise.
9656 (clarify_prohibited_class_mode_regs): Likewise.
9657 (check_allocation): Likewise.
9658 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9659 (lra_setup_reg_renumber): Likewise.
9660 (setup_try_hard_regno_pseudos): Likewise.
9661 (spill_for): Likewise.
9662 (assign_hard_regno): Likewise.
9663 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9664 * lra-constraints.c (in_class_p): Likewise.
9665 (lra_constraint_offset): Likewise.
9666 (simplify_operand_subreg): Likewise.
9667 (lra_constraints): Likewise.
9668 (split_reg): Likewise.
9669 (split_if_necessary): Likewise.
9670 (invariant_p): Likewise.
9671 (inherit_in_ebb): Likewise.
9672 * lra-lives.c (process_bb_lives): Likewise.
9673 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9674 (get_hard_regs): Likewise.
9675 (do_remat): Likewise.
9676 * lra-spills.c (assign_spill_hard_regs): Likewise.
9677 * mode-switching.c (create_pre_exit): Likewise.
9678 * postreload.c (reload_combine_recognize_pattern): Likewise.
9679 * recog.c (peep2_find_free_register): Likewise.
9680 * regcprop.c (kill_value_regno): Likewise.
9681 (set_value_regno): Likewise.
9682 (copy_value): Likewise.
9683 (maybe_mode_change): Likewise.
9684 (find_oldest_value_reg): Likewise.
9685 (copyprop_hardreg_forward_1): Likewise.
9686 * regrename.c (check_new_reg_p): Likewise.
9687 (regrename_do_replace): Likewise.
9688 * reload.c (push_reload): Likewise.
9689 (combine_reloads): Likewise.
9690 (find_dummy_reload): Likewise.
9691 (operands_match_p): Likewise.
9692 (find_reloads): Likewise.
9693 (find_equiv_reg): Likewise.
9694 (reload_adjust_reg_for_mode): Likewise.
9695 * reload1.c (count_pseudo): Likewise.
9696 (count_spilled_pseudo): Likewise.
9697 (find_reg): Likewise.
9698 (clear_reload_reg_in_use): Likewise.
9699 (free_for_value_p): Likewise.
9700 (allocate_reload_reg): Likewise.
9701 (choose_reload_regs): Likewise.
9702 (reload_adjust_reg_for_temp): Likewise.
9703 (emit_reload_insns): Likewise.
9704 (delete_output_reload): Likewise.
9705 * rtlanal.c (subreg_get_info): Likewise.
9706 * sched-deps.c (sched_analyze_reg): Likewise.
9707 * sel-sched.c (init_regs_for_mode): Likewise.
9708 (mark_unavailable_hard_regs): Likewise.
9709 (choose_best_reg_1): Likewise.
9710 (verify_target_availability): Likewise.
9711 * valtrack.c (dead_debug_insert_temp): Likewise.
9712 * var-tracking.c (track_loc_p): Likewise.
9713 (emit_note_insn_var_location): Likewise.
9714 * varasm.c (make_decl_rtl): Likewise.
9715 * reginfo.c (choose_hard_reg_mode): Likewise.
9716 (init_reg_modes_target): Refer directly to
9717 this_target_regs->x_hard_regno_nregs.
9719 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9721 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9722 instead of hard_regno_nregs.
9724 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9726 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9727 end_hard_regno instead of hard_regno_nregs.
9728 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9729 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9730 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9731 * ira-color.c (improve_allocation): Likewise.
9732 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9733 * lra-lives.c (mark_regno_live): Likewise.
9734 (mark_regno_dead): Likewise.
9735 * lra-remat.c (operand_to_remat): Likewise.
9736 * lra.c (collect_non_operand_hard_regs): Likewise.
9737 * postreload.c (reload_combine_note_store): Likewise.
9738 (move2add_valid_value_p): Likewise.
9739 * reload.c (regno_clobbered_p): Likewise.
9741 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9743 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9745 * config/v850/v850.c (v850_reorg): Likewise.
9746 * reload.c (refers_to_regno_for_reload_p): Likewise.
9747 (find_equiv_reg): Likewise.
9748 * reload1.c (reload_reg_reaches_end_p): Likewise.
9750 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9752 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9754 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9755 * config/arm/arm.c (output_move_neon): Likewise.
9756 (arm_attr_length_move_neon): Likewise.
9757 (neon_split_vcombine): Likewise.
9758 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9759 (c6x_mark_reg_written): Likewise.
9760 (c6x_dwarf_register_span): Likewise.
9761 * config/i386/i386.c (ix86_save_reg): Likewise.
9762 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9763 (rws_access_reg): Likewise.
9764 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9765 * mode-switching.c (create_pre_exit): Likewise.
9766 * ree.c (combine_reaching_defs): Likewise.
9767 (add_removable_extension): Likewise.
9768 * regcprop.c (find_oldest_value_reg): Likewise.
9769 (copyprop_hardreg_forward_1): Likewise.
9770 * reload.c (reload_inner_reg_of_subreg): Likewise.
9771 (push_reload): Likewise.
9772 (combine_reloads): Likewise.
9773 (find_dummy_reload): Likewise.
9774 (reload_adjust_reg_for_mode): Likewise.
9775 * reload1.c (find_reload_regs): Likewise.
9776 (forget_old_reloads_1): Likewise.
9777 (reload_reg_free_for_value_p): Likewise.
9778 (reload_adjust_reg_for_temp): Likewise.
9779 (emit_reload_insns): Likewise.
9780 (delete_output_reload): Likewise.
9781 * sel-sched.c (choose_best_reg_1): Likewise.
9782 (choose_best_pseudo_reg): Likewise.
9784 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9785 Alan Hayward <alan.hayward@arm.com>
9786 David Sherwood <david.sherwood@arm.com>
9788 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9789 * target.def (slow_unaligned_access): New hook.
9790 * targhooks.h (default_slow_unaligned_access): Declare.
9791 * targhooks.c (default_slow_unaligned_access): New function.
9792 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9793 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9794 * doc/tm.texi: Regenerate.
9795 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9796 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9797 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9799 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9800 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9802 (rs6000_slow_unaligned_access): New function.
9803 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9804 (expand_block_compare): Likewise.
9805 (expand_strn_compare): Likewise.
9806 (rs6000_rtx_costs): Likewise.
9807 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9808 (riscv_slow_unaligned_access): Likewise.
9809 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9810 (riscv_slow_unaligned_access_p): ...this and make static.
9811 (riscv_option_override): Update accordingly.
9812 (riscv_slow_unaligned_access): New function.
9813 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9814 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9815 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9816 (rs6000_slow_unaligned_access): New function.
9817 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9818 (rs6000_rtx_costs): Likewise.
9819 * config/rs6000/rs6000-string.c (expand_block_compare)
9820 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9821 of SLOW_UNALIGNED_ACCESS.
9822 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9823 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9824 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9825 of SLOW_UNALIGNED_ACCESS.
9826 * expmed.c (simple_mem_bitfield_p): Likewise.
9827 * expr.c (alignment_for_piecewise_move): Likewise.
9828 (emit_group_load_1): Likewise.
9829 (emit_group_store): Likewise.
9830 (copy_blkmode_from_reg): Likewise.
9831 (emit_push_insn): Likewise.
9832 (expand_assignment): Likewise.
9833 (store_field): Likewise.
9834 (expand_expr_real_1): Likewise.
9835 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9836 * lra-constraints.c (simplify_operand_subreg): Likewise.
9837 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9838 * gimple-ssa-store-merging.c: Likewise in block comment at start
9840 * tree-ssa-strlen.c: Include target.h.
9841 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9842 of SLOW_UNALIGNED_ACCESS.
9843 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9845 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9847 PR rtl-optimization/82185
9848 * expmed.c (emit_store_flag_int): Only test tem if it has been
9851 2017-09-12 Richard Biener <rguenther@suse.de>
9854 * match.pd ((FTYPE) N CMP CST): Fix typo.
9856 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9858 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9860 (mips_near_type_p): Add 'short_call' attribute as a synonym
9862 * doc/extend.texi (short_call): Document new function attribute.
9864 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9867 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9868 assertion check that in the condition.
9869 (get_atomic_generic_size): Likewise. Before testing if parameter
9870 has pointer type, if it has array type, call for C++
9871 default_conversion to perform array-to-pointer conversion.
9873 2017-09-12 Richard Biener <rguenther@suse.de>
9875 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9876 for operations we cannot scalarize.
9878 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9880 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9881 vectors heap vectors. Clean up comments.
9882 Make visited_bbs a reference.
9883 (profitable_jump_thread_path): Make GC
9884 vectors heap vectors. Clean up comments.
9886 (convert_and_register_jump_thread_path): Make GC vectors heap
9888 (check_subpath_and_update_thread_path): Same. Clean up comments.
9889 Make visited_bbs a reference.
9890 (handle_phi): Abstract common code to to
9891 register_jump_thread_path_if_profitable.
9892 Rename VAR_BB to DEF_BB.
9894 Make GC vectors heap vectors.
9895 Make visited_bbs a reference.
9896 (handle_assignment): Same.
9897 (register_jump_thread_path_if_profitable): New.
9898 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9900 Make GC vectors heap vectors. Clean up comments.
9901 Make visited_bbs a reference.
9902 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9903 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9906 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9909 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9910 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9912 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9914 Revert r251800 and r251799.
9916 2017-09-11 Martin Jambor <mjambor@suse.cz>
9919 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9920 arguments in advance.
9921 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9922 use it to find predecessor edges.
9923 (naive_outof_ssa): Collect vector of predecessors.
9925 2017-09-08 Jason Merrill <jason@redhat.com>
9927 PR c++/70029 - ICE with ref-qualifier and -flto
9928 * langhooks.h (struct lang_hooks_for_types): Add
9929 copy_lang_qualifiers.
9930 * attribs.c (build_type_attribute_qual_variant): Use it.
9931 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9933 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9934 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9936 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9939 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9940 (*mulsi3_sp64): New instruction.
9941 (mulsi3): New expander.
9943 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9945 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9947 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9949 * sancov.c: Include memmodel.h.
9951 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9954 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9957 2017-09-07 Carl Love <cel@us.ibm.com>
9959 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9960 the sldi instruction.
9962 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9964 * sancov.c: Include tm_p.h.
9966 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9969 * output.h (switch_to_other_text_partition): New declaration.
9970 * varasm.c (switch_to_other_text_partition): New function.
9971 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9972 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9973 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9974 to the other text partition before emitting LCL label and switch back
9975 after emitting the word after it.
9977 2017-09-07 Richard Biener <rguenther@suse.de>
9979 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9980 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9981 critical edges here, after loop init.
9982 (pass_data_pre): Remove PROP_no_crit_edges flags.
9983 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9984 for valueization of call args to avoid leaking VN_TOP.
9985 (visit_use): Assert we do not visit default defs.
9986 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9987 Use error_mark_node to more easily detect leaking VN_TOP.
9988 All default-defs are varying, not VN_TOP. Mark them visited.
9989 (run_scc_vn): Make code match comment.
9991 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9993 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9994 OPTION_MASK_FLOAT128_KEYWORD.
9995 (POWERPC_MASKS): Likewise.
9996 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9997 support for the -mfloat128-type option, and make -mfloat128
9998 default on PowerPC Linux systems. Define or undefine
9999 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10000 Define __float128 to be __ieee128 if IEEE 128-bit support is
10001 enabled, or undefine it.
10002 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10003 Delete defining __FLOAT128_TYPE__.
10004 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10005 -mfloat128-type option and make -mfloat128 default on PowerPC
10007 (TARGET_FLOAT128_TYPE): Likewise.
10008 (-mfloat128-type): Likewise.
10009 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10010 Delete the -mfloat128-type option and make -mfloat128 default on
10011 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10012 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10013 128-bit floating point is enabled. Change tests from using
10014 -mfloat128-type to -mfloat128.
10015 (rs6000_mangle_type): Use the correct mangling for the __float128
10016 type even if normal long double is restricted to 64-bits.
10017 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10019 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10020 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10021 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10023 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10024 documentation for -mfloat128.
10026 2017-09-06 Olivier Hainque <hainque@adacore.com>
10028 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10030 2017-09-06 Wish Wu <wishwu007@gmail.com>
10031 Jakub Jelinek <jakub@redhat.com>
10033 * asan.c (initialize_sanitizer_builtins): Add
10034 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10035 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10036 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10037 BT_FN_VOID_UINT64_PTR variables.
10038 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10039 (BT_FN_VOID_UINT16_UINT16): Likewise.
10040 (BT_FN_VOID_UINT32_UINT32): Likewise.
10041 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10042 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10043 (BT_FN_VOID_UINT64_PTR): Likewise.
10044 * common.opt (flag_sanitize_coverage): New variable.
10045 (fsanitize-coverage=trace-pc): Remove.
10046 (fsanitize-coverage=): Add.
10047 * flag-types.h (enum sanitize_coverage_code): New enum.
10048 * fold-const.c (fold_range_test): Disable non-short-circuit
10049 optimization if flag_sanitize_coverage.
10050 (fold_truth_andor): Likewise.
10051 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10052 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10053 (coverage_sanitizer_opts): New array.
10054 (get_closest_sanitizer_option): Add OPTS argument, handle also
10055 OPT_fsanitize_coverage_.
10056 (parse_sanitizer_options): Adjusted to also handle
10057 OPT_fsanitize_coverage_.
10058 (common_handle_option): Add OPT_fsanitize_coverage_.
10059 * sancov.c (instrument_comparison, instrument_switch): New function.
10060 (sancov_pass): Add trace-cmp support.
10061 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10062 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10063 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10064 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10065 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10066 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10067 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10068 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10069 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10070 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10072 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10074 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10075 error. Only quit immediately if parsing is complete.
10076 (BEGIN): Initialize fatal_err and parse_done.
10077 (begin fpu, end fpu): Check number of arguments.
10078 (begin arch, end arch): Likewise.
10079 (begin cpu, end cpu): Likewise.
10080 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10081 (optalias): Likewise.
10083 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10085 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10086 * config/arm/arm-isa.h: Delete. Move definitions to ...
10087 * arm-cpus.in: ... here. Use new feature and fgroup values.
10088 * config/arm/arm.c (arm_option_override): Use lower case for feature
10090 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10091 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10092 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10094 (print_isa_bits_for): New function.
10095 (gen_isa): New function.
10096 (gen_comm_data): Use print_isa_bits_for.
10097 (define feature): New keyword.
10098 (define fgroup): New keyword.
10099 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10100 (arm-isa.h): Add rule to generate file.
10101 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10102 case for feature bit names.
10104 2017-09-06 Richard Biener <rguenther@suse.de>
10106 * tree-ssa-pre.c (NECESSARY): Remove.
10107 (create_expression_by_pieces): Do not touch pass-local flags.
10108 (insert_into_preds_of_block): Likewise.
10109 (do_pre_regular_insertion): Likewise.
10110 (eliminate_insert): Likewise.
10111 (eliminate_dom_walker::before_dom_children): Likewise.
10112 (fini_eliminate): Do not look at inserted_exprs.
10113 (mark_operand_necessary): Remove.
10114 (remove_dead_inserted_code): Replace with simple work-list
10115 algorithm based on inserted_exprs and SSA uses.
10116 (pass_pre::execute): Re-order fini_eliminate and
10117 remove_dead_inserted_code.
10119 2017-09-06 Olivier Hainque <hainque@adacore.com>
10121 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10122 for VxWorks 7. Adjust surrounding comments.
10124 2017-09-06 Richard Biener <rguenther@suse.de>
10126 * gimple-ssa-strength-reduction.c
10127 (find_candidates_dom_walker::before_dom_children): Also allow
10130 2017-09-06 Richard Biener <rguenther@suse.de>
10132 PR tree-optimization/82108
10133 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10134 for gap in the non-permutation SLP case.
10136 2017-09-06 Martin Jambor <mjambor@suse.cz>
10138 PR tree-optimization/82078
10139 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10140 add_access_to_work_queue...
10141 (build_accesses_from_assign): ...here.
10142 (propagate_all_subaccesses): Make sure racc is the group
10143 representative, if there is one.
10145 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10147 PR middle-end/82095
10148 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10151 2017-09-06 Richard Biener <rguenther@suse.de>
10153 * gimple-ssa-strength-reduction.c
10154 (find_candidates_dom_walker::before_doom_children): Use a
10155 type and not a mode check.
10157 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10160 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10162 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10164 2017-09-05 Jeff Law <law@redhat.com>
10166 PR tree-optimization/64910
10167 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10168 cases where we have 3 or more operands.
10170 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10172 PR middle-end/81768
10173 * omp-low.c (lower_omp_for): Recompute tree invariant if
10174 gimple_omp_for_initial/final is ADDR_EXPR.
10176 PR middle-end/81768
10177 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10178 into gimple val before gimplification fo the COND_EXPR.
10180 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10182 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10183 REGION_COPY argument.
10184 (thread_through_all_blocks): Remove unused argument to
10185 duplicate_thread_path.
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_gen_adjusted_ldpstp):
10192 Take a scalar_mode rather than a machine_mode.
10193 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10194 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10195 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10196 (aarch64_gen_adjusted_ldpstp): Likewise.
10197 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10199 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10200 Alan Hayward <alan.hayward@arm.com>
10201 David Sherwood <david.sherwood@arm.com>
10203 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10204 Take a scalar_int_mode instead of a machine_mode.
10205 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10206 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10207 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10208 (aarch64_simd_attr_length_rglist): Delete.
10209 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10210 a scalar_int_mode instead of a machine_mode.
10211 (aarch64_add_offset): Likewise.
10212 (aarch64_internal_mov_immediate): Likewise
10213 (aarch64_add_constant_internal): Likewise.
10214 (aarch64_add_constant): Likewise.
10215 (aarch64_movw_imm): Likewise.
10216 (aarch64_rtx_arith_op_extract_p): Likewise.
10217 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10218 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10219 Remove assert that the mode isn't a vector.
10220 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10221 (aarch64_expand_mov_immediate): Update calls after above changes.
10222 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10223 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10224 (aarch64_move_imm): Likewise.
10225 (aarch64_can_const_movi_rtx_p): Likewise.
10226 (aarch64_strip_extend): Likewise.
10227 (aarch64_extr_rtx_p): Likewise.
10228 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10229 a CONST_INT when the mode parameter is VOIDmode.
10230 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10232 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10234 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10235 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10236 (bitwise_type_for_mode): Update accordingly.
10238 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10240 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10241 * stor-layout.c (mode_for_size_tree): Likewise.
10242 (mode_for_array): Update accordingly.
10243 (layout_decl): Likewise.
10244 (compute_record_mode): Likewise. Only set the mode once.
10246 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10248 * target.def (get_mask_mode): Change return type to opt_mode.
10250 * doc/tm.texi: Regenerate.
10251 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10252 * targhooks.c (default_get_mask_mode): Likewise.
10253 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10254 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10255 targetm.get_mask_mode.
10256 * tree.c (build_truth_vector_type): Likewise.
10258 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10260 * machmode.h (mode_for_vector): Return an opt_mode.
10261 * stor-layout.c (mode_for_vector): Likewise.
10262 (mode_for_int_vector): Update accordingly.
10263 (layout_type): Likewise.
10264 * config/i386/i386.c (emit_memmov): Likewise.
10265 (ix86_expand_set_or_movmem): Likewise.
10266 (ix86_expand_vector_init): Likewise.
10267 (ix86_get_mask_mode): Likewise.
10268 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10270 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10271 * expmed.c (extract_bit_field_1): Likewise.
10272 * expr.c (expand_expr_real_2): Likewise.
10273 * optabs-query.c (can_vec_perm_p): Likewise.
10274 (can_vec_mask_load_store_p): Likewise.
10275 * optabs.c (expand_vec_perm): Likewise.
10276 * targhooks.c (default_get_mask_mode): Likewise.
10277 * tree-vect-stmts.c (vectorizable_store): Likewise.
10278 (vectorizable_load): Likewise.
10279 (get_vectype_for_scalar_type_and_size): Likewise.
10281 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10283 * machmode.h (mode_for_int_vector): New function.
10284 * stor-layout.c (mode_for_int_vector): Likewise.
10285 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10286 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10287 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10288 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10289 (s390_expand_vcond): Likewise.
10291 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10293 * machmode.h (opt_machine_mode): New type.
10294 (opt_mode<T>): Allow construction from anything that can be
10296 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10297 (mode_for_size): Return an opt_machine_mode.
10298 * stor-layout.c (mode_for_size): Likewise.
10299 (mode_for_size_tree): Update call accordingly.
10300 (bitwise_mode_for_mode): Likewise.
10301 (make_fract_type): Likewise.
10302 (make_accum_type): Likewise.
10303 * caller-save.c (replace_reg_with_saved_mem): Update call
10305 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10306 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10307 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10308 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10309 * expmed.c (extract_bit_field_1): Likewise.
10310 * reload.c (get_secondary_mem): Likewise.
10311 * varasm.c (assemble_integer): Likewise.
10312 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10315 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10317 * machmode.h (decimal_float_mode_for_size): New function.
10318 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10319 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10320 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10321 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10322 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10323 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10325 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10327 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10328 (get_builtin_sync_mode): Likewise.
10329 (expand_ifn_atomic_compare_exchange): Likewise.
10330 (expand_builtin_atomic_clear): Likewise.
10331 (expand_builtin_atomic_test_and_set): Likewise.
10332 (fold_builtin_atomic_always_lock_free): Likewise.
10333 * calls.c (compute_argument_addresses): Likewise.
10334 (emit_library_call_value_1): Likewise.
10335 (store_one_arg): Likewise.
10336 * combine.c (combine_instructions): Likewise.
10337 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10338 * config/arm/arm.c (arm_function_value): Likewise.
10339 (aapcs_allocate_return_reg): Likewise.
10340 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10341 * config/i386/i386.c (construct_container): Likewise.
10342 (ix86_gimplify_va_arg): Likewise.
10343 (ix86_expand_sse_cmp): Likewise.
10344 (emit_memmov): Likewise.
10345 (emit_memset): Likewise.
10346 (expand_small_movmem_or_setmem): Likewise.
10347 (ix86_expand_pextr): Likewise.
10348 (ix86_expand_pinsr): Likewise.
10349 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10350 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10352 * config/mips/mips.c (mips_function_value_1) Likewise.
10353 (mips_block_move_straight): Likewise.
10354 (mips_expand_ins_as_unaligned_store): Likewise.
10355 * config/powerpcspe/powerpcspe.c
10356 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10357 (rs6000_darwin64_record_arg_flush): Likewise.
10358 * config/rs6000/rs6000.c
10359 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10360 (rs6000_darwin64_record_arg_flush): Likewise.
10361 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10362 (sparc_function_value_1): Likewise.
10363 * config/spu/spu.c (adjust_operand): Likewise.
10364 (spu_emit_branch_or_set): Likewise.
10365 (arith_immediate_p): Likewise.
10366 * emit-rtl.c (gen_lowpart_common): Likewise.
10367 * expr.c (expand_expr_real_1): Likewise.
10368 * function.c (assign_parm_setup_block): Likewise.
10369 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10370 * reload1.c (alter_reg): Likewise.
10371 * stor-layout.c (mode_for_vector): Likewise.
10372 (layout_type): Likewise.
10374 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10376 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10377 (spu_convert_move): Likewise.
10378 * lower-subreg.c (resolve_simple_move): Likewise.
10380 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10383 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10384 define_insn to a define_expand.
10385 (altivec_vsum2sws_direct): New define_insn.
10386 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10388 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10390 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10393 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10398 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10399 realign stack if stack alignment needed is less than incoming
10402 2017-09-05 Marek Polacek <polacek@redhat.com>
10405 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10408 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10410 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10412 2017-09-05 Richard Biener <rguenther@suse.de>
10414 PR tree-optimization/82084
10415 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10417 2017-09-05 Richard Biener <rguenther@suse.de>
10419 PR tree-optimization/82102
10420 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10422 2017-09-05 Martin Liska <mliska@suse.cz>
10424 PR tree-optimization/82032
10425 * tree-cfg.c (generate_range_test): New function.
10426 * tree-cfg.h (generate_range_test): Declared here.
10427 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10428 (cleanup_control_expr_graph): Use it.
10429 * tree-switch-conversion.c (try_switch_expansion): Remove
10431 (emit_case_nodes): Use generate_range_test.
10433 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10436 * config/i386/i386.md (*<btsc><mode>_mask): Add
10437 TARGET_USE_BT to insn constraint.
10438 (*btr<mode>_mask): Ditto.
10440 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10442 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10443 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10445 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10448 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10449 TARGET_NEON and TARGET_IWMMXT.
10450 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10451 TARGET_NEON and TARGET_IWMMXT.
10452 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10454 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10456 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10457 (ix86_rewrite_tls_address): Ditto.
10458 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10459 (ix86_rewrite_tls_address_1): Ditto.
10460 (ix86_rewrite_tls_address): Ditto.
10461 * config/i386/predicates.md (tls_address_pattern): New predicate.
10462 * config/i386/i386.md (TLS address splitter): New splitter.
10464 2017-09-04 Richard Biener <rguenther@suse.de>
10466 PR tree-optimization/82084
10467 * fold-const.h (can_native_encode_string_p): Declare.
10468 * fold-const.c (can_native_encode_string_p): Factor out from ...
10469 (native_encode_string): ... here.
10470 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10471 vectorizing stores from constants we later cannot handle.
10473 2017-09-04 Marek Polacek <polacek@redhat.com>
10476 * doc/invoke.texi: Update -Wtautological-compare documentation.
10478 2017-09-04 Jeff Law <law@redhat.com>
10480 PR tree-optimization/64910
10481 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10482 swap the first and last operand if the last is a constant.
10484 2017-09-04 Marek Polacek <polacek@redhat.com>
10487 * convert.c (do_narrow): When sanitizing signed integer overflows,
10488 bail out for signed types.
10489 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10491 2017-09-04 Richard Biener <rguenther@suse.de>
10493 PR tree-optimization/82060
10494 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10495 Move devirtualization after stmt folding and before EH/AB/noreturn
10496 cleanup to get the stmt refs canonicalized. Use a bool instead
10497 of gimple_modified_p since that doesn't work for NOPs. Schedule
10498 NOPs generated by folding for removal.
10500 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10501 Alan Hayward <alan.hayward@arm.com>
10502 David Sherwood <david.sherwood@arm.com>
10504 * coretypes.h (pad_direction): New enum.
10505 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10506 (FUNCTION_ARG_PADDING): Likewise.
10507 * target.def (function_arg_padding): New hook.
10508 * targhooks.h (default_function_arg_padding): Declare.
10509 * targhooks.c (default_function_arg_padding): New function.
10510 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10511 (TARGET_FUNCTION_ARG_PADDING): ...this.
10512 * doc/tm.texi: Regenerate.
10513 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10514 instead of direction.
10515 (compute_argument_addresses): Likewise.
10516 (load_register_parameters): Likewise.
10517 (emit_library_call_value_1): Likewise.
10518 (store_one_arg): Use targetm.calls.function_arg_padding instead
10519 of FUNCTION_ARG_PADDING.
10520 (must_pass_in_stack_var_size_or_pad): Likewise.
10521 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10522 (emit_group_store): Likewise.
10523 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10524 instead of FUNCTION_ARG_PADDING.
10525 (emit_push_insn): Likewise, and propagate enum change throughout
10527 * function.h (direction): Delete.
10528 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10530 * function.c (assign_parm_find_stack_rtl): Likewise.
10531 (assign_parm_setup_block_p): Likewise.
10532 (assign_parm_setup_block): Likewise.
10533 (gimplify_parameters): Likewise.
10534 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10535 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10537 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10538 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10539 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10540 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10541 (aarch64_function_arg_padding): ...this new function.
10542 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10543 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10544 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10545 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10546 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10547 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10548 (arm_pad_arg_upward): Replace with...
10549 (arm_function_arg_padding): ...this new function.
10550 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10552 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10553 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10554 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10555 (ia64_hpux_function_arg_padding): Replace with...
10556 (ia64_function_arg_padding): ...this new function. Use pad_direction
10557 instead of direction. Check for TARGET_HPUX.
10558 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10559 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10560 (iq2000_function_arg_padding): New function.
10561 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10562 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10563 (mips_function_arg_padding): ...this new function.
10564 (mips_pad_reg_upward): Update accordingly.
10565 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10566 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10567 targetm.calls.function_arg_padding.
10568 (FUNCTION_ARG_PADDING): Delete.
10569 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10570 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10571 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10572 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10573 (nios2_block_reg_padding): Return pad_direction instead of direction.
10574 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10575 instead of direction.
10576 (nios2_function_arg_padding): Likewise. Make static.
10577 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10578 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10579 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10580 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10581 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10582 (pa_function_arg_padding): Make static. Return pad_direction instead
10584 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10585 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10586 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10587 instead of direction. Use targetm.calls.function_arg_padding.
10588 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10589 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10590 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10591 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10592 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10594 (function_arg_padding): Rename to...
10595 (rs6000_function_arg_padding): ...this. Make static. Return
10596 pad_direction instead of direction.
10597 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10598 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10599 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10600 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10601 instead of direction. Use targetm.calls.function_arg_padding.
10602 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10603 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10604 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10605 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10606 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10607 (function_arg_padding): Rename to...
10608 (rs6000_function_arg_padding): ...this. Make static. Return
10609 pad_direction instead of direction.
10610 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10611 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10612 * config/s390/s390.c (s390_function_arg_padding): New function.
10613 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10614 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10615 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10616 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10617 (function_arg_padding): Rename to...
10618 (sparc_function_arg_padding): ...this. Make static. Return
10619 pad_direction instead of direction.
10620 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10621 * config/spu/spu.c (spu_function_arg_padding): New function.
10622 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10623 * system.h (FUNCTION_ARG_PADDING): Poison.
10625 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10626 Alan Hayward <alan.hayward@arm.com>
10627 David Sherwood <david.sherwood@arm.com>
10629 * target.def (modes_tieable_p): New hook.
10630 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10631 (TARGET_MODES_TIEABLE_P): ...this.
10632 * doc/tm.texi.in: Regenerate.
10633 * hooks.h (hook_bool_mode_mode_true): Declare.
10634 * hooks.c (hook_bool_mode_mode_true): New function.
10635 * combine.c (subst): Use targetm.modes_tieable_p instead of
10637 * dse.c (find_shift_sequence): Likewise.
10638 * expmed.c (extract_low_bits): Likewise.
10639 * lower-subreg.c: Include target.h.
10640 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10642 * rtlanal.c (rtx_cost): Likewise.
10643 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10644 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10645 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10646 (TARGET_MODES_TIEABLE_P): Redefine.
10647 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10648 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10649 (TARGET_MODES_TIEABLE_P): Redefine.
10650 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10651 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10652 (arc_modes_tieable_p): New function.
10653 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10654 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10655 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10656 (arm_modes_tieable_p): Make static.
10657 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10658 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10659 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10660 (TARGET_MODES_TIEABLE_P): Redefine.
10661 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10662 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10663 (TARGET_MODES_TIEABLE_P): Redefine.
10664 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10665 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10666 (cr16_modes_tieable_p): New function.
10667 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10668 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10669 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10670 (TRULY_NOOP_TRUNCATION): Update comment.
10671 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10672 (TRULY_NOOP_TRUNCATION): Update comment.
10673 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10674 (frv_modes_tieable_p): New function.
10675 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10676 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10677 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10678 (TARGET_MODES_TIEABLE_P): Redefine.
10679 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10680 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10681 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10682 (TARGET_MODES_TIEABLE_P): Redefine.
10683 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10684 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10685 (ia64_modes_tieable_p): New function.
10686 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10687 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10688 (iq2000_modes_tieable_p): New function.
10689 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10690 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10691 (lm32_modes_tieable_p): New function.
10692 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10693 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10694 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10695 (TARGET_MODES_TIEABLE_P): Redefine.
10696 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10697 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10698 (m32r_modes_tieable_p): New function.
10699 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10700 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10701 (m68k_modes_tieable_p): New function.
10702 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10703 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10704 (mcore_modes_tieable_p): New function.
10705 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10706 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10708 (TARGET_MODES_TIEABLE_P): Redefine.
10709 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10710 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10711 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10712 (TARGET_MODES_TIEABLE_P): Redefine.
10713 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10714 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10715 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10716 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10717 (mn10300_modes_tieable_p): ...this and make static.
10718 (TARGET_MODES_TIEABLE_P): Redefine.
10719 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10720 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10721 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10722 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10723 (msp430_modes_tieable_p): Make static.
10724 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10725 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10726 (TARGET_MODES_TIEABLE_P): Redefine.
10727 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10728 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10729 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10730 (TARGET_MODES_TIEABLE_P): Redefine.
10731 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10732 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10733 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10734 (TARGET_MODES_TIEABLE_P): Redefine.
10735 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10736 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10737 (pdp11_modes_tieable_p): New function.
10738 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10739 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10740 (rs6000_modes_tieable_p): New function.
10741 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10742 * config/powerpcspe/powerpcspe.md: Update comment.
10743 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10744 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10745 (TARGET_MODES_TIEABLE_P): Redefine.
10746 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10747 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10748 (rl78_modes_tieable_p): New function.
10749 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10750 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10751 (rs6000_modes_tieable_p): New function.
10752 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10753 * config/rs6000/rs6000.md: Update comment.
10754 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10755 * config/rx/rx.c (rx_modes_tieable_p): New function.
10756 (TARGET_MODES_TIEABLE_P): Redefine.
10757 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10758 * config/s390/s390.c (s390_modes_tieable_p): New function.
10759 (TARGET_MODES_TIEABLE_P): Redefine.
10760 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10761 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10762 (sh_modes_tieable_p): New function.
10763 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10764 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10765 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10766 (sparc_modes_tieable_p): Make static.
10767 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10768 * config/spu/spu.c (spu_modes_tieable_p): New function.
10769 (TARGET_MODES_TIEABLE_P): Redefine.
10770 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10771 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10772 (TARGET_MODES_TIEABLE_P): Redefine.
10773 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10774 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10775 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10776 * config/v850/v850.c (v850_modes_tieable_p): New function.
10777 (TARGET_MODES_TIEABLE_P): Redefine.
10778 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10779 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10780 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10781 (visium_modes_tieable_p): New function.
10782 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10783 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10784 (xtensa_modes_tieable_p): New function.
10785 * system.h (MODES_TIEABLE_P): Poison.
10787 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10788 Alan Hayward <alan.hayward@arm.com>
10789 David Sherwood <david.sherwood@arm.com>
10791 * target.def (hard_regno_mode_ok): New hook.
10792 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10793 (TARGET_HARD_REGNO_MODE_OK): ...this.
10794 * doc/tm.texi.in: Regenerate.
10795 * hooks.h (hook_bool_uint_mode_true): Declare.
10796 * hooks.c (hook_bool_uint_mode_true): New function.
10797 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10798 HARD_REGNO_MODE_OK.
10799 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10800 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10801 instead of HARD_REGNO_MODE_OK.
10802 * caller-save.c: Include target.h.
10803 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10804 HARD_REGNO_MODE_OK.
10805 * combine.c (can_combine_p): Likewise.
10806 (combinable_i3pat): Likewise.
10807 (can_change_dest_mode): Likewise.
10808 * expr.c (init_expr_target): Likewise.
10809 (convert_move): Likewise.
10810 (convert_modes): Likewise.
10811 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10812 (setup_prohibited_mode_move_regs): Likewise.
10813 * ira.h (target_ira): Likewise.
10814 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10815 * lra-constraints.c (process_alt_operands): Likewise.
10816 (split_reg): Likewise.
10817 * recog.c (peep2_find_free_register): Likewise.
10818 * ree.c (combine_reaching_defs): Likewise.
10819 * regcprop.c (maybe_mode_change): Likewise.
10820 * reginfo.c (init_reg_sets_1): Likewise.
10821 (choose_hard_reg_mode): Likewise.
10822 (simplifiable_subregs): Likewise.
10823 * regrename.c (check_new_reg_p): Likewise.
10824 * reload.c (find_valid_class): Likewise.
10825 (find_valid_class_1): Likewise.
10826 (reload_inner_reg_of_subreg): Likewise.
10827 (push_reload): Likewise.
10828 (combine_reloads): Likewise.
10829 (find_dummy_reload): Likewise.
10830 (find_reloads): Likewise.
10831 * reload1.c (find_reg): Likewise.
10832 (set_reload_reg): Likewise.
10833 (allocate_reload_reg): Likewise.
10834 (choose_reload_regs): Likewise.
10835 (reload_adjust_reg_for_temp): Likewise.
10836 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10837 (simplify_subreg_regno): Likewise.
10838 * sel-sched.c (init_regs_for_mode): Likewise.
10839 * varasm.c (make_decl_rtl): Likewise.
10840 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10841 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10842 HARD_REGNO_MODE_OK.
10843 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10844 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10846 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10847 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10848 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10849 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10850 (arc_mode_class): Delete.
10851 (HARD_REGNO_MODE_OK): Delete.
10852 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10853 (arc_hard_regno_mode_ok): Rename old array to...
10854 (arc_hard_regno_mode_ok_modes): ...this.
10855 (arc_conditional_register_usage): Update accordingly.
10856 (arc_mode_class): Make static.
10857 (arc_hard_regno_mode_ok): New function.
10858 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10859 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10860 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10861 (arm_hard_regno_mode_ok): Make static.
10862 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10863 HARD_REGNO_MODE_OK.
10864 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10865 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10866 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10868 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10869 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10870 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10871 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10872 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10873 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10874 * config/bfin/predicates.md (valid_reg_operand): Use
10875 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10876 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10877 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10878 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10879 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10880 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10881 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10882 (cr16_hard_regno_mode_ok): Make static and return a bool.
10883 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10884 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10885 (cris_hard_regno_mode_ok): New function.
10886 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10887 (epiphany_mode_class): Delete.
10888 (HARD_REGNO_MODE_OK): Delete.
10889 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10890 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10891 (hard_regno_mode_ok): Rename to...
10892 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10893 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10894 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10895 HARD_REGNO_MODE_OK.
10896 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10897 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10898 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10899 (frv_hard_regno_mode_ok): Make static and return a bool.
10900 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10901 HARD_REGNO_MODE_OK.
10902 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10903 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10904 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10905 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10907 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10908 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10909 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10910 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10912 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10913 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10914 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10915 (ia64_hard_regno_mode_ok): New function.
10916 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10917 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10918 (iq2000_hard_regno_mode_ok): New function.
10919 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10920 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10921 (lm32_hard_regno_mode_ok): New function.
10922 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10923 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10924 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10925 instead of HARD_REGNO_MODE_OK.
10926 (m32c_hard_regno_ok): Rename to...
10927 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10928 (m32c_cannot_change_mode_class): Update accordingly.
10929 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10930 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10931 (m32r_mode_class): Delete.
10932 (HARD_REGNO_MODE_OK): Delete.
10933 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10934 (m32r_hard_regno_mode_ok): Rename to...
10935 (m32r_hard_regno_modes): ...this.
10936 (m32r_mode_class): Make static.
10937 (m32r_hard_regno_mode_ok): New function.
10938 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10939 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10940 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10941 (m68k_hard_regno_mode_ok): Make static.
10942 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10943 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10944 (mcore_hard_regno_mode_ok): New function.
10945 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10946 (HARD_REGNO_MODE_OK): Delete.
10947 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10949 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10950 (microblaze_hard_regno_mode_ok): New function.
10951 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10952 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10953 (mips_hard_regno_mode_ok): Delete.
10954 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10955 (mips_hard_regno_mode_ok_p): ...this and make static.
10956 (mips_hard_regno_mode_ok_p): Rename to...
10957 (mips_hard_regno_mode_ok_uncached): ...this.
10958 (mips_hard_regno_mode_ok): New function.
10959 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10960 of HARD_REGNO_MODE_OK.
10961 (mips_option_override): Update after above name changes.
10962 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10963 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10964 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10965 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10966 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10967 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10968 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10969 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10970 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10971 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10972 (msp430_hard_regno_mode_ok): Make static and return a bool.
10973 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10974 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10975 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10977 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10978 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10979 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10980 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10981 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10982 (PA_HARD_REGNO_MODE_OK): ...this
10983 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10984 (PA_HARD_REGNO_MODE_OK): ...this.
10985 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10986 (pa_hard_regno_mode_ok): New function.
10987 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10988 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10989 (pdp11_hard_regno_mode_ok): New function.
10990 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10991 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10993 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10995 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10996 (rs6000_hard_regno_mode_ok): Rename to...
10997 (rs6000_hard_regno_mode_ok_uncached): ...this.
10998 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10999 (rs6000_hard_regno_mode_ok): New function.
11000 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11001 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11002 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11003 (riscv_hard_regno_mode_ok): ...this and make static.
11004 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11005 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11006 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11007 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11008 (rl78_hard_regno_mode_ok): Make static and return bool.
11009 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11010 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11012 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11013 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11014 (rs6000_hard_regno_mode_ok): Rename to...
11015 (rs6000_hard_regno_mode_ok_uncached): ...this.
11016 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11017 (rs6000_hard_regno_mode_ok): New function.
11018 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11019 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11020 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11021 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11022 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11023 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11025 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11026 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11027 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11028 (sh_hard_regno_mode_ok): Make static.
11029 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11030 instead of HARD_REGNO_MODE_OK.
11031 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11032 (sparc_mode_class): Delete.
11033 (HARD_REGNO_MODE_OK): Delete.
11034 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11035 (hard_regno_mode_classes): Make static.
11036 (sparc_mode_class): Likewise.
11037 (sparc_hard_regno_mode_ok): New function.
11038 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11039 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11040 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11042 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11043 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11044 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11045 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11046 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11048 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11049 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11050 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11051 (visium_hard_regno_mode_ok): New function.
11052 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11053 instead of HARD_REGNO_MODE_OK.
11054 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11055 (HARD_REGNO_MODE_OK): Delete.
11056 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11057 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11058 (xtensa_option_override): Update accordingly.
11059 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11060 (xtensa_hard_regno_mode_ok): New function.
11061 * system.h (HARD_REGNO_MODE_OK): Poison.
11063 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11064 Alan Hayward <alan.hayward@arm.com>
11065 David Sherwood <david.sherwood@arm.com>
11067 * target.def (hard_regno_call_part_clobbered): New hook.
11068 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11069 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11070 * doc/tm.texi: Regenerate.
11071 * hooks.h (hook_bool_uint_mode_false): Declare.
11072 * hooks.c (hook_bool_uint_mode_false): New function.
11073 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11074 * cselib.c (cselib_process_insn): Use
11075 targetm.hard_regno_call_part_clobbered instead of
11076 HARD_REGNO_CALL_PART_CLOBBERED.
11077 * ira-conflicts.c (ira_build_conflicts): Likewise.
11078 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11079 * lra-constraints.c (need_for_call_save_p): Likewise.
11080 * lra-lives.c: Include target.h.
11081 (check_pseudos_live_through_calls): Use
11082 targetm.hard_regno_call_part_clobbered instead of
11083 HARD_REGNO_CALL_PART_CLOBBERED.
11084 * regcprop.c: Include target.h.
11085 (copyprop_hardreg_forward_1): Use
11086 targetm.hard_regno_call_part_clobbered instead of
11087 HARD_REGNO_CALL_PART_CLOBBERED.
11088 * reginfo.c (choose_hard_reg_mode): Likewise.
11089 * regrename.c (check_new_reg_p): Likewise.
11090 * reload.c (find_equiv_reg): Likewise.
11091 * reload1.c (emit_reload_insns): Likewise.
11092 * sched-deps.c (deps_analyze_insn): Likewise.
11093 * sel-sched.c (init_regs_for_mode): Likewise.
11094 (mark_unavailable_hard_regs): Likewise.
11095 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11096 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11097 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11099 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11100 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11101 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11103 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11105 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11106 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11107 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11109 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11110 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11111 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11113 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11114 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11116 * config/powerpcspe/powerpcspe.c
11117 (rs6000_hard_regno_call_part_clobbered): New function.
11118 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11119 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11120 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11122 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11123 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11124 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11126 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11127 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11128 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11130 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11131 Alan Hayward <alan.hayward@arm.com>
11132 David Sherwood <david.sherwood@arm.com>
11134 * rtl.h (subreg_memory_offset): Declare.
11135 * emit-rtl.c (subreg_memory_offset): New function.
11136 * expmed.c (store_bit_field_1): Use it.
11137 * expr.c (undefined_operand_subword_p): Likewise.
11138 * simplify-rtx.c (simplify_subreg): Likewise.
11140 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11142 PR rtl-optimization/57448
11145 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11146 using atomic_load pattern.
11147 (expand_atomic_store): Likewise.
11149 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11152 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11153 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11154 handling. Use replace_call_with_value with NULL instead of
11155 gsi_replace, unlink_stmt_vdef and release_defs.
11157 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11160 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11162 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11165 * rtl.h (emit_library_call_value_1): Declare.
11166 (emit_library_call): Replace declaration with a series of overloads.
11167 Remove the parameter count argument.
11168 (emit_library_call_value): Likewise.
11169 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11170 with an "rtx_mode_t *".
11171 (emit_library_call_value): Delete.
11172 (emit_library_call): Likewise.
11173 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11174 (asan_emit_allocas_unpoison): Likewise.
11175 * builtins.c (expand_builtin_powi): Likewise.
11176 (expand_asan_emit_allocas_unpoison): Likewise.
11177 * cfgexpand.c (expand_main_function): Likewise.
11178 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11179 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11180 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11181 * config/arm/arm.c (arm_trampoline_init): Likewise.
11182 (arm_call_tls_get_addr): Likewise.
11183 (arm_expand_divmod_libfunc): Likewise.
11184 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11185 (smulsi3_highpart): Likewise.
11186 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11187 (c6x_expand_compare): Likewise.
11188 (c6x_expand_movmem): Likewise.
11189 * config/frv/frv.c (frv_trampoline_init): Likewise.
11190 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11191 (ix86_expand_divmod_libfunc): Likewise.
11192 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11193 (ia64_expand_compare): Likewise.
11194 (ia64_profile_hook): Likewise.
11195 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11196 (nonlocal_goto): Likewise.
11197 (restore_stack_nonlocal): Likewise.
11198 * config/m32r/m32r.c (block_move_call): Likewise.
11199 (m32r_trampoline_init): Likewise.
11200 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11201 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11202 (m68k_call_m68k_read_tp): Likewise.
11203 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11204 (microblaze_expand_divide): Likewise.
11205 * config/mips/mips.h (mips_args): Likewise.
11206 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11207 (MIPS_ICACHE_SYNC): Likewise.
11208 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11209 (nios2_trampoline_init): Likewise.
11210 * config/pa/pa.c (hppa_tls_call): Likewise.
11211 (pa_trampoline_init): Likewise.
11212 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11213 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11214 (expand_strn_compare): Likewise.
11215 (rs6000_generate_compare): Likewise.
11216 (rs6000_expand_float128_convert): Likewise.
11217 (output_profile_hook): Likewise.
11218 (rs6000_trampoline_init): Likewise.
11219 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11220 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11221 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11222 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11223 (rs6000_generate_compare): Likewise.
11224 (rs6000_expand_float128_convert): Likewise.
11225 (output_profile_hook): Likewise.
11226 (rs6000_trampoline_init): Likewise.
11227 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11228 * config/sh/sh.c (sh_trampoline_init): Likewise.
11229 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11230 (sparc_emit_float_lib_cmp): Likewise.
11231 (sparc32_initialize_trampoline): Likewise.
11232 (sparc64_initialize_trampoline): Likewise.
11233 (sparc_profile_hook): Likewise.
11234 * config/spu/spu.c (ea_load_store): Likewise.
11235 * config/spu/spu.md (floatunssidf2): Likewise.
11236 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11237 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11238 * config/visium/visium.c (expand_block_move_4): Likewise.
11239 (expand_block_move_2): Likewise.
11240 (expand_block_move_1): Likewise.
11241 (expand_block_set_4): Likewise.
11242 (expand_block_set_2): Likewise.
11243 (expand_block_set_1): Likewise.
11244 (visium_trampoline_init): Likewise.
11245 (visium_profile_hook): Likewise.
11246 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11247 (xtensa_setup_frame_addresses): Likewise.
11248 (xtensa_trampoline_init): Likewise.
11249 * except.c (sjlj_emit_function_enter): Likewise.
11250 (sjlj_emit_function_exit): Likewise.
11251 * explow.c (allocate_dynamic_stack_space): Likewise.
11252 (probe_stack_range): Likewise.
11253 * expr.c (convert_mode_scalar): Likewise.
11254 * optabs.c (expand_binop): Likewise.
11255 (expand_twoval_binop_libfunc): Likewise.
11256 (expand_unop): Likewise.
11257 (prepare_cmp_insn): Likewise.
11258 (prepare_float_lib_cmp): Likewise.
11259 (expand_float): Likewise.
11260 (expand_fix): Likewise.
11261 (expand_fixed_convert): Likewise.
11262 (maybe_emit_sync_lock_test_and_set): Likewise.
11263 (expand_atomic_compare_and_swap): Likewise.
11264 (expand_mem_thread_fence): Likewise.
11265 (expand_atomic_fetch_op): Likewise.
11267 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11269 * doc/generic.texi (OpenACC): Adjust URL.
11270 * doc/invoke.texi (C Dialect Options): Ditto.
11272 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11274 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11275 predicate for operand 1. Add (m,<S>) constraint.
11276 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11277 Prevent memory operand 1 with register operand 2.
11279 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11281 PR rtl-optimization/82024
11282 * combine.c (try_combine): If the combination result is a PARALLEL,
11283 and we only need to retain the SET in there that would be placed
11284 at I2, check that we can place that at I3 instead, before doing so.
11286 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11289 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11291 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11292 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11295 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11296 Jeff Law <law@redhat.com>
11298 * varasm.c (bss_initializer_p): Do not put constants into .bss
11299 (categorize_decl_for_section): Handle bss_initializer_p returning
11300 false when DECL_INITIAL is NULL.
11302 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11305 * config/s390/s390.c (s390_can_inline_p): New function.
11307 2017-09-01 Jeff Law <law@redhat.com>
11309 PR tree-optimization/82052
11310 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11311 Always initialize the returned slot after a hash table miss
11312 when INSERT is true.
11314 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11316 * config/s390/s390.md (mem_signal_fence): Remove.
11317 * doc/md.texi (mem_signal_fence): Remove.
11318 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11320 * target-insns.def (mem_signal_fence): Remove.
11322 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11325 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11328 * asan.c (create_odr_indicator): Strip name encoding from assembler
11329 name before appending it after __odr_asan_.
11331 2017-09-01 Martin Liska <mliska@suse.cz>
11333 PR tree-optimization/82059
11334 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11335 frequency only when an edge is redirected.
11337 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11339 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11340 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11341 (arc_conditional_register_usage): Remove ARC600 lp_count
11343 (arc_file_start): Emit Tag_ARC_CPU_variation.
11344 (arc_can_use_doloop_p): New conditions to use ZOLs.
11345 (hwloop_fail): New function.
11346 (hwloop_optimize): Likewise.
11347 (hwloop_pattern_reg): Likewise.
11348 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11349 (arc_reorg_loops): New function, calls reorg_loops.
11350 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11351 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11353 (arc_loop_hazard): Remove function, functionality moved into
11355 (arc_hazard): Remove arc_loop_hazard call.
11356 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11357 into hwloop_optimize.
11358 (arc_label_align): Remove ZOL handling.
11359 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11360 * config/arc/arc.md (doloop_begin): Remove pattern.
11361 (doloop_begin_i): Likewise.
11362 (doloop_end_i): Likewise.
11363 (doloop_fallback): Likewise.
11364 (doloop_fallback_m): Likewise.
11365 (doloop_end): Reimplement expand.
11366 (arc_lp): New pattern for LP instruction.
11367 (loop_end): New pattern.
11368 (loop_fail): Likewise.
11369 (decrement_and_branch_until_zero): Likewise.
11370 * config/arc/arc.opt (mlpc-width): New option.
11371 * doc/invoke.texi (mlpc-width): Document option.
11373 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11375 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11376 (arc_ccfsm_advance): Fix checking for delay slots.
11377 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11379 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11381 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11383 (movhi_insn): Update store instruction constraint which are saving
11384 6-bit short immediates.
11385 (movsi_insn): Consider also short scaled load operations.
11386 (zero_extendhisi2_i): Use Usd constraint instead of T.
11387 (extendhisi2_i): Add q constraint.
11388 (arc_clzsi2): Add type and length attributes.
11389 (arc_ctzsi2): Likewise.
11390 * config/arc/constraints.md (Usc): Update constraint, the
11391 assembler can parse two relocations for a single instruction.
11393 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11395 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11396 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11398 2017-08-31 Olivier Hainque <hainque@adacore.com>
11400 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11401 match as powerpc-wrs-vxworks*.
11403 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11405 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11406 register constraint for by-element operand.
11407 (aarch64_mls_elt_merge<mode>): Likewise.
11409 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11411 * config/arc/arc.c (arc_can_follow_jump): Check for short
11414 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11416 * config.gcc: Use g.opt for arc.
11417 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11418 functionality moved to ...
11419 (legitimate_scaled_address_p): New function, ...here.
11420 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11421 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11422 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11424 (arc_override_options): Handle G option.
11425 (arc_output_pic_addr_const): Correct function definition.
11426 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11427 (arc_decl_anon_ns_mem_p): Delete.
11428 (arc_in_small_data_p): Overhaul this function to take into
11429 consideration the value given via G option.
11430 (arc_rewrite_small_data_1): Renamed and corrected old
11431 arc_rewrite_small_data function.
11432 (arc_rewrite_small_data): New function.
11433 (small_data_pattern): Don't use pic_offset_table_rtx.
11434 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11435 * config/arc/simdext.md (movmisalignv2hi): Use
11436 prepare_move_operands function.
11438 (movmisalign*): Likewise.
11439 * doc/invoke.texi (ARC options): Document -G option.
11441 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11443 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11445 * config/arc/arc.c (arc_print_operand): Output scalled address for
11446 sdata whenever is possible.
11447 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11448 load/stores are available.
11449 (compact_sda_memory_operand): Check for the alignment required by
11450 code density instructions.
11451 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11453 * config/arc/constraints.md (Usd): Update constraint.
11454 (Us0): New constraint.
11455 (Usc): Update constraint.
11457 2017-08-31 Richard Biener <rguenther@suse.de>
11459 PR middle-end/82054
11460 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11461 function only once.
11463 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11465 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11466 Resize type_signature.
11468 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11469 Alan Hayward <alan.hayward@arm.com>
11470 David Sherwood <david.sherwood@arm.com>
11472 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11473 subregs whose inner modes can be stored in GPRs.
11474 (aarch64_classify_index): Likewise.
11476 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11477 Alan Hayward <alan.hayward@arm.com>
11478 David Sherwood <david.sherwood@arm.com>
11480 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11481 (V_INT_EQUIV): ...this.
11482 (v_cmp_result): Rename to...
11483 (v_int_equiv): ...this.
11484 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11485 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11486 (copysign<mode>3): Likewise.
11487 (aarch64_simd_bsl<mode>_internal): Likewise.
11488 (aarch64_simd_bsl<mode>): Likewise.
11489 (vec_cmp<mode><mode>): Likewise.
11490 (vcond<mode><mode>): Likewise.
11491 (vcond<v_cmp_mixed><mode>): Likewise.
11492 (vcondu<mode><v_cmp_mixed>): Likewise.
11493 (aarch64_cm<optab><mode>): Likewise.
11494 (aarch64_cmtst<mode>): Likewise.
11495 (aarch64_fac<optab><mode>): Likewise.
11496 (vec_perm_const<mode>): Likewise.
11497 (vcond_mask_<mode><v_cmp_result>): Rename to...
11498 (vcond_mask_<mode><v_int_equiv>): ...this.
11499 (vec_cmp<mode><v_cmp_result>): Rename to...
11500 (vec_cmp<mode><v_int_equiv>): ...this.
11502 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11503 Alan Hayward <alan.hayward@arm.com>
11504 David Sherwood <david.sherwood@arm.com>
11506 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11508 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11509 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11510 (UNSPEC_LD4_DREG): New unspecs.
11511 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11512 (aarch64_ld2<mode>_dreg_be): Replace with...
11513 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11515 (aarch64_ld3<mode>_dreg_le)
11516 (aarch64_ld3<mode>_dreg_be): Replace with...
11517 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11519 (aarch64_ld4<mode>_dreg_le)
11520 (aarch64_ld4<mode>_dreg_be): Replace with...
11521 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11524 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11526 PR tree-optimization/81987
11527 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11528 insert an initializer in a location not dominated by the stride
11531 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11533 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11534 on the entire header of the finally block in the fallthru case.
11536 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11538 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11540 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11542 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11543 rs6000_emit_move_from_cr and call renamed function.
11544 (rs6000_emit_prologue): Call renamed functions.
11545 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11546 movesi_from_cr, remove volatile CRs.
11548 2017-08-30 Jon Beniston <jon@beniston.com>
11549 Richard Biener <rguenther@suse.de>
11551 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11552 of VECTOR_MODE_P check.
11553 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11554 element vector types.
11556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11558 * df.h (df_read_modify_subreg_p): Remove in favor of...
11559 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11560 const_rtx instead of an rtx.
11561 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11562 * df-problems.c (df_word_lr_mark_ref): Likewise.
11563 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11564 (mark_pseudo_reg_dead): Likewise.
11565 (mark_ref_dead): Likewise.
11566 * reginfo.c (init_subregs_of_mode): Likewise.
11567 * sched-deps.c (sched_analyze_1): Likewise.
11568 * df-scan.c (df_def_record_1): Likewise.
11569 (df_uses_record): Likewise.
11570 (df_read_modify_subreg_p): Remove in favor of...
11571 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11572 const_rtx instead of an rtx.
11574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11575 Alan Hayward <alan.hayward@arm.com>
11576 David Sherwood <david.sherwood@arm.com>
11578 * rtl.h (partial_subreg_p): New function.
11579 * caller-save.c (save_call_clobbered_regs): Use it.
11580 * calls.c (expand_call): Likewise.
11581 * combine.c (combinable_i3pat): Likewise.
11582 (simplify_set): Likewise.
11583 (make_extraction): Likewise.
11584 (make_compound_operation_int): Likewise.
11585 (gen_lowpart_or_truncate): Likewise.
11586 (force_to_mode): Likewise.
11587 (make_field_assignment): Likewise.
11588 (reg_truncated_to_mode): Likewise.
11589 (record_truncated_value): Likewise.
11590 (move_deaths): Likewise.
11591 * cse.c (record_jump_cond): Likewise.
11592 (cse_insn): Likewise.
11593 * cselib.c (cselib_lookup_1): Likewise.
11594 * expmed.c (extract_bit_field_using_extv): Likewise.
11595 * function.c (assign_parm_setup_reg): Likewise.
11596 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11597 * ira-build.c (create_insn_allocnos): Likewise.
11598 * lra-coalesce.c (merge_pseudos): Likewise.
11599 * lra-constraints.c (match_reload): Likewise.
11600 (simplify_operand_subreg): Likewise.
11601 (curr_insn_transform): Likewise.
11602 * lra-lives.c (process_bb_lives): Likewise.
11603 * lra.c (new_insn_reg): Likewise.
11604 (lra_substitute_pseudo): Likewise.
11605 * regcprop.c (mode_change_ok): Likewise.
11606 (maybe_mode_change): Likewise.
11607 (copyprop_hardreg_forward_1): Likewise.
11608 * reload.c (push_reload): Likewise.
11609 (find_reloads): Likewise.
11610 (find_reloads_subreg_address): Likewise.
11611 * reload1.c (alter_reg): Likewise.
11612 (eliminate_regs_1): Likewise.
11613 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11615 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11617 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11618 back to if statements, including unpack.
11620 2017-08-30 Martin Liska <mliska@suse.cz>
11622 PR inline-asm/82001
11623 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11625 (func_checker::compare_asm_inputs_outputs): ... this function.
11626 (func_checker::compare_gimple_asm): Use the function to compare
11627 also ASM constrains.
11628 * ipa-icf-gimple.h: Rename the function.
11630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11631 Alan Hayward <alan.hayward@arm.com>
11632 David Sherwood <david.sherwood@arm.com>
11634 * coretypes.h (complex_mode): New type.
11635 * gdbhooks.py (build_pretty_printer): Handle it.
11636 * machmode.h (complex_mode): New class.
11637 (complex_mode::includes_p): New function.
11638 (is_complex_int_mode): Likewise.
11639 (is_complex_float_mode): Likewise.
11640 * genmodes.c (get_mode_class): Handle complex mode classes.
11641 * function.c (expand_function_end): Use is_complex_int_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 * coretypes.h (scalar_mode_pod): New typedef.
11648 * gdbhooks.py (build_pretty_printer): Handle it.
11649 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11650 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11651 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11652 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11653 as_a <scalar_mode>.
11655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11656 Alan Hayward <alan.hayward@arm.com>
11657 David Sherwood <david.sherwood@arm.com>
11659 * machmode.h (mode_for_vector): Take a scalar_mode instead
11661 * stor-layout.c (mode_for_vector): Likewise.
11662 * explow.c (promote_mode): Use as_a <scalar_mode>.
11663 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11666 Alan Hayward <alan.hayward@arm.com>
11667 David Sherwood <david.sherwood@arm.com>
11669 * target.def (preferred_simd_mode): Take a scalar_mode
11670 instead of a machine_mode.
11671 * targhooks.h (default_preferred_simd_mode): Likewise.
11672 * targhooks.c (default_preferred_simd_mode): Likewise.
11673 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11674 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11675 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11676 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11677 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11678 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11679 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11680 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11682 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11683 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11684 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11685 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11686 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11687 * doc/tm.texi: Regenerate.
11688 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11692 Alan Hayward <alan.hayward@arm.com>
11693 David Sherwood <david.sherwood@arm.com>
11695 * target.def (scalar_mode_supported_p): Take a scalar_mode
11696 instead of a machine_mode.
11697 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11698 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11699 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11700 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11701 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11702 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11703 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11704 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11705 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11706 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11707 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11708 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11709 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11710 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11712 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11713 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11714 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11715 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11716 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11718 * doc/tm.texi: Regenerate.
11720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11721 Alan Hayward <alan.hayward@arm.com>
11722 David Sherwood <david.sherwood@arm.com>
11724 * coretypes.h (opt_scalar_mode): New typedef.
11725 * gdbhooks.py (build_pretty_printers): Handle it.
11726 * machmode.h (mode_iterator::get_2xwider): Add overload for
11728 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11730 * expr.c (convert_mode_scalar): Likewise.
11731 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11732 * optabs.c (expand_float): Likewise.
11733 (expand_fix): Likewise.
11734 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11737 Alan Hayward <alan.hayward@arm.com>
11738 David Sherwood <david.sherwood@arm.com>
11740 * optabs.c (expand_float): Explicitly check for scalars before
11741 using a branching expansion.
11742 (expand_fix): Likewise.
11744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11745 Alan Hayward <alan.hayward@arm.com>
11746 David Sherwood <david.sherwood@arm.com>
11748 * expr.c (convert_mode): Split scalar handling out into...
11749 (convert_mode_scalar): ...this new function. Treat the modes
11752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11753 Alan Hayward <alan.hayward@arm.com>
11754 David Sherwood <david.sherwood@arm.com>
11756 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11758 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11761 Alan Hayward <alan.hayward@arm.com>
11762 David Sherwood <david.sherwood@arm.com>
11764 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11765 rather than a machine_mode.
11766 (fixed_from_string): Likewise.
11767 (fixed_convert): Likewise.
11768 (fixed_convert_from_int): Likewise.
11769 (fixed_convert_from_real): Likewise.
11770 (real_convert_from_fixed): Likewise.
11771 * fixed-value.c (fixed_from_double_int): Likewise.
11772 (fixed_from_string): Likewise.
11773 (fixed_convert): Likewise.
11774 (fixed_convert_from_int): Likewise.
11775 (fixed_convert_from_real): Likewise.
11776 (real_convert_from_fixed): Likewise.
11777 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11780 Alan Hayward <alan.hayward@arm.com>
11781 David Sherwood <david.sherwood@arm.com>
11783 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11784 of separate mode class checks. Do not allow vector modes here.
11785 (immed_wide_int_const): Use as_a <scalar_mode>.
11786 * explow.c (trunc_int_for_mode): Likewise.
11787 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11788 (wi::shwi): Likewise.
11789 (wi::min_value): Likewise.
11790 (wi::max_value): Likewise.
11791 * dwarf2out.c (loc_descriptor): Likewise.
11792 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11793 for CONST_WIDE_INT.
11795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11796 Alan Hayward <alan.hayward@arm.com>
11797 David Sherwood <david.sherwood@arm.com>
11799 * tree.h (SCALAR_TYPE_MODE): New macro.
11800 * expr.c (expand_expr_addr_expr_1): Use it.
11801 (expand_expr_real_2): Likewise.
11802 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11803 (fold_convert_const_fixed_from_int): Likewise.
11804 (fold_convert_const_fixed_from_real): Likewise.
11805 (native_encode_fixed): Likewise
11806 (native_encode_complex): Likewise
11807 (native_encode_vector): Likewise.
11808 (native_interpret_fixed): Likewise.
11809 (native_interpret_real): Likewise.
11810 (native_interpret_complex): Likewise.
11811 (native_interpret_vector): Likewise.
11812 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11813 (simd_clone_adjust_argument_types): Likewise.
11814 (simd_clone_init_simd_arrays): Likewise.
11815 (simd_clone_adjust): Likewise.
11816 * stor-layout.c (layout_type): Likewise.
11817 * tree.c (build_minus_one_cst): Likewise.
11818 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11819 * tree-inline.c (estimate_move_cost): Likewise.
11820 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11821 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11822 (vectorizable_reduction): Likewise.
11823 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11824 (vect_recog_mixed_size_cond_pattern): Likewise.
11825 (check_bool_pattern): Likewise.
11826 (adjust_bool_pattern): Likewise.
11827 (search_type_for_mask_1): Likewise.
11828 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11829 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11830 (vectorizable_load): Likewise.
11831 (vectorizable_store): Likewise.
11832 * ubsan.c (ubsan_encode_value): Likewise.
11833 * varasm.c (output_constant): Likewise.
11835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11836 Alan Hayward <alan.hayward@arm.com>
11837 David Sherwood <david.sherwood@arm.com>
11839 * coretypes.h (scalar_mode): New class.
11840 * machmode.h (scalar_mode): Likewise.
11841 (scalar_mode::includes_p): New function.
11842 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11843 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11844 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11845 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11846 * expmed.c (store_bit_field_1): Likewise.
11847 (extract_bit_field_1): Likewise.
11848 * expr.c (write_complex_part): Likewise.
11849 (read_complex_part): Likewise.
11850 (emit_move_complex_push): Likewise.
11851 (expand_expr_real_2): Likewise.
11852 * function.c (assign_parm_setup_reg): Likewise.
11853 (assign_parms_unsplit_complex): Likewise.
11854 * optabs.c (expand_binop): Likewise.
11855 * rtlanal.c (subreg_get_info): Likewise.
11856 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11857 * varasm.c (output_constant_pool_2): Likewise.
11859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11860 Alan Hayward <alan.hayward@arm.com>
11861 David Sherwood <david.sherwood@arm.com>
11863 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11865 (expmed_mult_highpart_optab): Likewise.
11866 (expmed_mult_highpart): Likewise.
11868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11869 Alan Hayward <alan.hayward@arm.com>
11870 David Sherwood <david.sherwood@arm.com>
11872 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11873 instead of a machine_mode.
11874 (builtin_memset_read_str): Likewise.
11875 * builtins.c (c_readstr): Likewise.
11876 (builtin_memcpy_read_str): Likewise.
11877 (builtin_strncpy_read_str): Likewise.
11878 (builtin_memset_read_str): Likewise.
11879 (builtin_memset_gen_str): Likewise.
11880 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11881 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11882 instead of a machine_mode.
11883 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11885 (make_extraction): Likewise.
11886 (try_widen_shift_mode): Take and return scalar_int_modes instead
11888 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11889 a scalar_int_mode instead of a machine_mode.
11890 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11891 (avr_addr_space_pointer_mode): Likewise.
11892 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11893 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11894 (msp430_unwind_word_mode): Likewise.
11895 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11896 (spu_addr_space_pointer_mode): Likewise.
11897 (spu_addr_space_address_mode): Likewise.
11898 (spu_libgcc_cmp_return_mode): Likewise.
11899 (spu_libgcc_shift_count_mode): Likewise.
11900 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11901 (rl78_addr_space_pointer_mode): Likewise.
11902 (fl78_unwind_word_mode): Likewise.
11903 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11905 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11906 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11907 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11908 (mips_valid_pointer_mode): Likewise.
11909 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11910 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11911 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11913 (ft32_addr_space_address_mode): Likewise.
11914 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11915 scalar_int_mode instead of a machine_mode.
11916 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11918 (m32c_addr_space_address_mode): Likewise.
11919 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11920 (rs6000_eh_return_filter_mode): Likewise.
11921 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11922 (rs6000_eh_return_filter_mode): Likewise.
11923 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11924 (s390_libgcc_shift_count_mode): Likewise.
11925 (s390_unwind_word_mode): Likewise.
11926 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11928 * target.def (mode_rep_extended): Likewise.
11929 (valid_pointer_mode): Likewise.
11930 (addr_space.valid_pointer_mode): Likewise.
11931 (eh_return_filter_mode): Return a scalar_int_mode rather than
11933 (libgcc_cmp_return_mode): Likewise.
11934 (libgcc_shift_count_mode): Likewise.
11935 (unwind_word_mode): Likewise.
11936 (addr_space.pointer_mode): Likewise.
11937 (addr_space.address_mode): Likewise.
11938 * doc/tm.texi: Regenerate.
11939 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11941 (do_jump): Use scalar_int_mode for local variables.
11942 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11943 rather than a machine_mode.
11944 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11945 (scompare_loc_descriptor_wide): Likewise.
11946 (scompare_loc_descriptor_narrow): Likewise.
11947 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11949 * except.c (sjlj_emit_dispatch_table): Likewise.
11950 (expand_builtin_eh_copy_values): Likewise.
11951 * explow.c (convert_memory_address_addr_space_1): Likewise.
11952 Take a scalar_int_mode rather than a machine_mode.
11953 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11954 than a machine_mode.
11955 (memory_address_addr_space): Use scalar_int_mode for local variables.
11956 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11957 rather than a machine_mode.
11958 * expmed.c (mask_rtx): Likewise.
11959 (init_expmed_one_conv): Likewise.
11960 (expand_mult_highpart_adjust): Likewise.
11961 (extract_high_half): Likewise.
11962 (expmed_mult_highpart_optab): Likewise.
11963 (expmed_mult_highpart): Likewise.
11964 (expand_smod_pow2): Likewise.
11965 (expand_sdiv_pow2): Likewise.
11966 (emit_store_flag_int): Likewise.
11967 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11969 (extract_low_bits): Likewise.
11970 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11972 * expr.c (pieces_addr::adjust): Likewise.
11973 (can_store_by_pieces): Likewise.
11974 (store_by_pieces): Likewise.
11975 (clear_by_pieces_1): Likewise.
11976 (expand_expr_addr_expr_1): Likewise.
11977 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11978 (expand_expr_real_1): Likewise.
11979 (try_casesi): Likewise.
11980 * final.c (shorten_branches): Likewise.
11981 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11982 type of "mode" to machine_mode.
11983 * internal-fn.c (expand_arith_overflow_result_store): Take a
11984 scalar_int_mode rather than a machine_mode.
11985 (expand_mul_overflow): Use scalar_int_mode for local variables.
11986 * loop-doloop.c (doloop_modify): Likewise.
11987 (doloop_optimize): Likewise.
11988 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11989 than a machine_mode.
11990 (expand_doubleword_shift_condmove): Likewise.
11991 (expand_doubleword_shift): Likewise.
11992 (expand_doubleword_clz): Likewise.
11993 (expand_doubleword_popcount): Likewise.
11994 (expand_doubleword_parity): Likewise.
11995 (expand_absneg_bit): Use scalar_int_mode for local variables.
11996 (prepare_float_lib_cmp): Likewise.
11997 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11998 rather than a machine_mode.
11999 (convert_memory_address_addr_space): Likewise.
12000 (get_mode_bounds): Likewise.
12001 (get_address_mode): Return a scalar_int_mode rather than a
12003 * rtlanal.c (get_address_mode): Likewise.
12004 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12005 than a machine_mode.
12006 * targhooks.c (default_mode_rep_extended): Likewise.
12007 (default_valid_pointer_mode): Likewise.
12008 (default_addr_space_valid_pointer_mode): Likewise.
12009 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12010 than a machine_mode.
12011 (default_libgcc_cmp_return_mode): Likewise.
12012 (default_libgcc_shift_count_mode): Likewise.
12013 (default_unwind_word_mode): Likewise.
12014 (default_addr_space_pointer_mode): Likewise.
12015 (default_addr_space_address_mode): Likewise.
12016 * targhooks.h (default_eh_return_filter_mode): Likewise.
12017 (default_libgcc_cmp_return_mode): Likewise.
12018 (default_libgcc_shift_count_mode): Likewise.
12019 (default_unwind_word_mode): Likewise.
12020 (default_addr_space_pointer_mode): Likewise.
12021 (default_addr_space_address_mode): Likewise.
12022 (default_mode_rep_extended): Take a scalar_int_mode rather than
12024 (default_valid_pointer_mode): Likewise.
12025 (default_addr_space_valid_pointer_mode): Likewise.
12026 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12028 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12029 rather than a machine_mode.
12030 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12031 for local variables.
12032 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12033 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12034 than a machine_mode.
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 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12041 the mode argument to scalar_int_mode.
12042 (do_jump_by_parts_zero_rtx): Likewise.
12043 (do_jump_by_parts_equality_rtx): Likewise.
12044 (do_jump_by_parts_greater): Take a mode argument.
12045 (do_jump_by_parts_equality): Likewise.
12046 (do_jump_1): Update calls accordingly.
12048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12049 Alan Hayward <alan.hayward@arm.com>
12050 David Sherwood <david.sherwood@arm.com>
12052 * is-a.h (safe_dyn_cast): New function.
12053 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12054 (jump_table_for_label): Likewise.
12055 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12056 instead of an rtx_insn *.
12057 (shorten_branches): Use dyn_cast instead of LABEL_P and
12058 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12059 rtx_jump_table_data::get_data_mode.
12060 (final_scan_insn): Likewise.
12062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12063 Alan Hayward <alan.hayward@arm.com>
12064 David Sherwood <david.sherwood@arm.com>
12066 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12067 trying to combine a full-register integer set with a subreg
12070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12071 Alan Hayward <alan.hayward@arm.com>
12072 David Sherwood <david.sherwood@arm.com>
12074 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12075 that is always either address_mode or pointer_mode.
12077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12078 Alan Hayward <alan.hayward@arm.com>
12079 David Sherwood <david.sherwood@arm.com>
12081 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12082 when the two are known to be equal.
12084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12085 Alan Hayward <alan.hayward@arm.com>
12086 David Sherwood <david.sherwood@arm.com>
12088 * simplify-rtx.c (simplify_const_unary_operation): Use
12089 is_a <scalar_int_mode> instead of checking for a nonzero
12090 precision. Forcibly convert op_mode to a scalar_int_mode
12091 in that case. More clearly differentiate the operand and
12092 result modes and use the former when deciding what the value
12093 of a count-bits operation should be. Use is_int_mode instead
12094 of checking for a MODE_INT. Remove redundant check for whether
12095 this mode has a zero precision.
12097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12098 Alan Hayward <alan.hayward@arm.com>
12099 David Sherwood <david.sherwood@arm.com>
12101 * optabs.c (widen_leading): Change the type of the mode argument
12102 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12103 (widen_bswap): Likewise.
12104 (expand_parity): Likewise.
12105 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12106 (expand_ffs): Likewise.
12107 (epand_unop): Check for scalar integer modes before calling the
12110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12111 Alan Hayward <alan.hayward@arm.com>
12112 David Sherwood <david.sherwood@arm.com>
12114 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12116 (expand_expr_real_1): Update call accordingly.
12118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12119 Alan Hayward <alan.hayward@arm.com>
12120 David Sherwood <david.sherwood@arm.com>
12122 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12123 value_mode arguments. Use scalar_int_mode internally.
12124 (store_bit_field_1): Rename the new integer mode from imode
12125 to op0_mode and use it instead of GET_MODE (op0). Update calls
12126 to store_split_bit_field, store_bit_field_using_insv and
12127 store_fixed_bit_field.
12128 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12129 Use scalar_int_mode internally. Use a bit count rather than a mode
12130 when calculating the largest bit size for get_best_mode.
12131 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12132 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12133 Remove assertion that OP0 has a scalar integer mode.
12134 (store_split_bit_field): Add op0_mode and value_mode arguments.
12135 Update calls to extract_fixed_bit_field.
12136 (extract_bit_field_using_extv): Add an op0_mode argument.
12137 Use scalar_int_mode internally.
12138 (extract_bit_field_1): Rename the new integer mode from imode to
12139 op0_mode and use it instead of GET_MODE (op0). Update calls to
12140 extract_split_bit_field, extract_bit_field_using_extv and
12141 extract_fixed_bit_field.
12142 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12143 to extract_split_bit_field and extract_fixed_bit_field_1.
12144 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12145 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12146 on the target mode.
12147 (extract_split_bit_field): Add an op0_mode argument. Update call
12148 to extract_fixed_bit_field.
12150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12151 Alan Hayward <alan.hayward@arm.com>
12152 David Sherwood <david.sherwood@arm.com>
12154 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12155 * explow.c (hard_function_value): Likewise.
12156 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12157 convert_to_mode call outside the loop.
12158 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12159 for the mode iterator. Require the mode specified by max_pieces
12161 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12163 (copy_blkmode_to_reg): Likewise.
12164 (set_storage_via_setmem): Likewise.
12165 * optabs.c (prepare_cmp_insn): Likewise.
12166 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12167 * stor-layout.c (finish_bitfield_representative): Likewise.
12169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12170 Alan Hayward <alan.hayward@arm.com>
12171 David Sherwood <david.sherwood@arm.com>
12173 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12174 * expr.c (convert_move): Use them.
12175 (convert_modes): Likewise.
12176 (store_expr_with_bounds): Likewise.
12178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12179 Alan Hayward <alan.hayward@arm.com>
12180 David Sherwood <david.sherwood@arm.com>
12182 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12183 parameter for the mode of "x". Remove the "known_x", "known_mode"
12184 and "known_ret" arguments. Change the type of the mode argument
12185 to scalar_int_mode.
12186 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12187 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12188 (reg_num_sign_bit_copies_for_combine): Likewise.
12189 * rtlanal.c (nonzero_bits1): Likewise.
12190 (num_sign_bit_copies1): Likewise.
12191 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12192 (reg_num_sign_bit_copies_general): Likewise.
12193 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12194 (reg_nonzero_bits_general): Likewise.
12196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12197 Alan Hayward <alan.hayward@arm.com>
12198 David Sherwood <david.sherwood@arm.com>
12200 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12201 than in subroutines. Return 1 for non-integer modes.
12202 (cached_num_sign_bit_copies): Change the type of the mode parameter
12203 to scalar_int_mode.
12204 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12205 classes. Handle CONST_INT_P first and then check whether X also
12206 has a scalar integer mode. Check the same thing for inner registers
12207 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12214 in subroutines. Return the mode mask for non-integer modes.
12215 (cached_nonzero_bits): Change the type of the mode parameter
12216 to scalar_int_mode.
12217 (nonzero_bits1): Likewise. Remove early exit for other mode
12218 classes. Handle CONST_INT_P first and then check whether X
12219 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12226 Return a scalar_int_mode and assert that the size is greater than
12228 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12229 (op_by_pieces_d::op_by_pieces_d): Likewise.
12230 (op_by_pieces_d::run): Likewise.
12231 (can_store_by_pieces): Likewise.
12233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12234 Alan Hayward <alan.hayward@arm.com>
12235 David Sherwood <david.sherwood@arm.com>
12237 * combine.c (extract_left_shift): Add a mode argument and update
12239 (make_compound_operation_int): Change the type of the mode parameter
12240 to scalar_int_mode and update the call to extract_left_shift.
12242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12243 Alan Hayward <alan.hayward@arm.com>
12244 David Sherwood <david.sherwood@arm.com>
12246 * combine.c (simplify_and_const_int): Change the type of the mode
12247 parameter to scalar_int_mode.
12248 (simplify_and_const_int_1): Likewise. Update recursive call.
12250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12251 Alan Hayward <alan.hayward@arm.com>
12252 David Sherwood <david.sherwood@arm.com>
12254 * combine.c (simplify_compare_const): Check that the mode is a
12255 scalar_int_mode (rather than VOIDmode) before testing its
12257 (simplify_comparison): Move COMPARISON_P handling out of the
12258 loop and restrict the latter part of the loop to scalar_int_modes.
12259 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12260 and when considering SUBREG_REGs. Use is_int_mode instead of
12261 checking GET_MODE_CLASS against MODE_INT.
12263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12264 Alan Hayward <alan.hayward@arm.com>
12265 David Sherwood <david.sherwood@arm.com>
12267 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12268 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12269 shift_unit_mode and for modes involved in scalar shifts.
12271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12272 Alan Hayward <alan.hayward@arm.com>
12273 David Sherwood <david.sherwood@arm.com>
12275 * combine.c (force_int_to_mode): New function, split out from...
12276 (force_to_mode): ...here. Keep xmode up-to-date and use it
12277 instead of GET_MODE (x).
12279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12280 Alan Hayward <alan.hayward@arm.com>
12281 David Sherwood <david.sherwood@arm.com>
12283 * optabs-query.h (extraction_insn::struct_mode): Change type to
12284 opt_scalar_int_mode and update comment.
12285 (extraction_insn::field_mode): Change type to scalar_int_mode.
12286 (extraction_insn::pos_mode): Likewise.
12287 * combine.c (make_extraction): Update accordingly.
12288 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12289 (get_optab_extraction_insn): Likewise.
12290 * recog.c (simplify_while_replacing): Likewise.
12291 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12292 parameter to opt_scalar_int_mode.
12294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12295 Alan Hayward <alan.hayward@arm.com>
12296 David Sherwood <david.sherwood@arm.com>
12298 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12299 to a scalar_int_mode instead of a machine_mode.
12300 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12301 (get_best_mode): Return a boolean and use a pointer argument to store
12302 the selected mode. Replace the limit mode parameter with a bit limit.
12303 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12304 for the values returned by bit_field_mode_iterator::next_mode.
12305 (store_bit_field): Update call to get_best_mode.
12306 (store_fixed_bit_field): Likewise.
12307 (extract_fixed_bit_field): Likewise.
12308 * expr.c (optimize_bitfield_assignment_op): Likewise.
12309 * fold-const.c (optimize_bit_field_compare): Likewise.
12310 (fold_truth_andor_1): Likewise.
12311 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12312 Update for new type of m_mode.
12313 (get_best_mode): As above.
12315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12316 Alan Hayward <alan.hayward@arm.com>
12317 David Sherwood <david.sherwood@arm.com>
12319 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12320 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12321 (store_bit_field): Check is_a <scalar_int_mode> before calling
12322 strict_volatile_bitfield_p.
12323 (extract_bit_field): Likewise.
12325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12326 Alan Hayward <alan.hayward@arm.com>
12327 David Sherwood <david.sherwood@arm.com>
12329 * target.def (cstore_mode): Return a scalar_int_mode.
12330 * doc/tm.texi: Regenerate.
12331 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12332 * targhooks.h (default_cstore_mode): Likewise.
12333 * targhooks.c (default_cstore_mode): Likewise, using a forced
12335 * expmed.c (emit_cstore): Expect the target of the cstore to be
12338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12339 Alan Hayward <alan.hayward@arm.com>
12340 David Sherwood <david.sherwood@arm.com>
12342 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12344 (niter_desc): Likewise mode.
12345 (iv_analyze): Add a mode parameter.
12347 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12348 and change its type to scalar_int_mode.
12349 * loop-iv.c: Update commentary at head of file.
12350 (iv_constant): Pass the mode paraeter before the rtx it describes
12351 and change its type to scalar_int_mode. Remove VOIDmode handling.
12352 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12353 (iv_extend): Likewise.
12354 (shorten_into_mode): Likewise.
12355 (iv_add): Use scalar_int_mode.
12356 (iv_mult): Likewise.
12357 (iv_shift): Likewise.
12358 (canonicalize_iv_subregs): Likewise.
12359 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12360 it describes and change its mode to scalar_int_mode. Also change
12361 the type of the returned inner_mode to scalar_int_mode.
12362 (get_biv_step): Likewise, turning outer_mode from a pointer
12363 into a direct parameter. Update call to get_biv_step_1.
12364 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12365 iv_constant and get_biv_step.
12366 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12367 and change its type to scalar_int_mode. Don't initialise iv->mode
12368 to VOIDmode and remove later checks for its still being VOIDmode.
12369 Update calls to iv_analyze_op and iv_analyze_expr. Check
12370 is_a <scalar_int_mode> when changing the mode under consideration.
12371 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12372 Update call to iv_analyze_expr.
12373 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12374 inner register is not also a scalar_int_mode. Update call to
12376 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12377 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12378 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12379 separate mode class checks. Update calls to iv_analyze. Remove
12380 fix-up of VOIDmodes after iv_analyze_biv.
12381 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12382 don't have a scalar_int_mode. Update call to biv_p.
12384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12385 Alan Hayward <alan.hayward@arm.com>
12386 David Sherwood <david.sherwood@arm.com>
12388 * cfgexpand.c (convert_debug_memory_address): Use
12389 as_a <scalar_int_mode>.
12390 * combine.c (expand_compound_operation): Likewise.
12391 (make_extraction): Likewise.
12392 (change_zero_ext): Likewise.
12393 (simplify_comparison): Likewise.
12394 * cse.c (cse_insn): Likewise.
12395 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12396 (mem_loc_descriptor): Likewise.
12397 (loc_descriptor): Likewise.
12398 * expmed.c (init_expmed_one_mode): Likewise.
12399 (synth_mult): Likewise.
12400 (emit_store_flag_1): Likewise.
12401 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12402 of a comparison with size.
12403 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12404 (reduce_to_bit_field_precision): Likewise.
12405 * function.c (expand_function_end): Likewise.
12406 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12407 * loop-doloop.c (doloop_modify): Likewise.
12408 * optabs.c (expand_binop): Likewise.
12409 (expand_unop): Likewise.
12410 (expand_copysign_absneg): Likewise.
12411 (prepare_cmp_insn): Likewise.
12412 (maybe_legitimize_operand): Likewise.
12413 * recog.c (const_scalar_int_operand): Likewise.
12414 * rtlanal.c (get_address_mode): Likewise.
12415 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12416 (simplify_cond_clz_ctz): Likewise.
12417 * tree-nested.c (get_nl_goto_field): Likewise.
12418 * tree.c (build_vector_type_for_mode): Likewise.
12419 * var-tracking.c (use_narrower_mode): Likewise.
12421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12422 Alan Hayward <alan.hayward@arm.com>
12423 David Sherwood <david.sherwood@arm.com>
12425 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12426 * builtins.c (expand_builtin_signbit): Use it.
12427 * cfgexpand.c (expand_debug_expr): Likewise.
12428 * dojump.c (do_jump): Likewise.
12429 (do_compare_and_jump): Likewise.
12430 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12431 * expmed.c (make_tree): Likewise.
12432 * expr.c (expand_expr_real_2): Likewise.
12433 (expand_expr_real_1): Likewise.
12434 (try_casesi): Likewise.
12435 * fold-const-call.c (fold_const_call_ss): Likewise.
12436 * fold-const.c (unextend): Likewise.
12437 (extract_muldiv_1): Likewise.
12438 (fold_single_bit_test): Likewise.
12439 (native_encode_int): Likewise.
12440 (native_encode_string): Likewise.
12441 (native_interpret_int): Likewise.
12442 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12443 * internal-fn.c (expand_addsub_overflow): Likewise.
12444 (expand_neg_overflow): Likewise.
12445 (expand_mul_overflow): Likewise.
12446 (expand_arith_overflow): Likewise.
12447 * match.pd: Likewise.
12448 * stor-layout.c (layout_type): Likewise.
12449 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12450 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12451 * tree-ssanames.c (get_range_info): Likewise.
12452 * tree-switch-conversion.c (array_value_type) Likewise.
12453 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12454 (vect_recog_divmod_pattern): Likewise.
12455 (vect_recog_mixed_size_cond_pattern): Likewise.
12456 * tree-vrp.c (extract_range_basic): Likewise.
12457 (simplify_float_conversion_using_ranges): Likewise.
12458 * tree.c (int_fits_type_p): Likewise.
12459 * ubsan.c (instrument_bool_enum_load): Likewise.
12460 * varasm.c (mergeable_string_section): Likewise.
12461 (narrowing_initializer_constant_valid_p): Likewise.
12462 (output_constant): 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 * machmode.h (NARROWEST_INT_MODE): New macro.
12469 * expr.c (alignment_for_piecewise_move): Use it instead of
12470 GET_CLASS_NARROWEST_MODE (MODE_INT).
12471 (push_block): Likewise.
12472 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12474 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12477 Alan Hayward <alan.hayward@arm.com>
12478 David Sherwood <david.sherwood@arm.com>
12480 * postreload.c (move2add_valid_value_p): Change the type of the
12481 mode parameter to scalar_int_mode.
12482 (move2add_use_add2_insn): Add a mode parameter and use it instead
12484 (move2add_use_add3_insn): Likewise.
12485 (reload_cse_move2add): Update accordingly.
12487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12488 Alan Hayward <alan.hayward@arm.com>
12489 David Sherwood <david.sherwood@arm.com>
12491 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12493 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12494 * optabs.c (expand_unop): Likewise.
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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12501 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12502 (popcount_loc_descriptor): Likewise.
12503 (bswap_loc_descriptor): Likewise.
12504 (rotate_loc_descriptor): Likewise.
12505 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12506 calling the functions above.
12508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12509 Alan Hayward <alan.hayward@arm.com>
12510 David Sherwood <david.sherwood@arm.com>
12512 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12514 (try_combine): Likewise.
12515 (simplify_if_then_else): Likewise.
12516 * cse.c (cse_insn): Likewise.
12517 * dwarf2out.c (mem_loc_descriptor): Likewise.
12518 * emit-rtl.c (gen_lowpart_common): Likewise.
12519 * simplify-rtx.c (simplify_truncation): Likewise.
12520 (simplify_binary_operation_1): Likewise.
12521 (simplify_const_relational_operation): Likewise.
12522 (simplify_ternary_operation): Likewise.
12523 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12526 Alan Hayward <alan.hayward@arm.com>
12527 David Sherwood <david.sherwood@arm.com>
12529 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12530 * reload.c (push_reload): Likewise.
12531 (find_reloads): Likewise.
12533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12534 Alan Hayward <alan.hayward@arm.com>
12535 David Sherwood <david.sherwood@arm.com>
12537 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12538 (make_compound_operation_int): Likewise.
12539 (change_zero_ext): Likewise.
12540 * expr.c (convert_move): Likewise.
12541 (convert_modes): Likewise.
12542 * fwprop.c (forward_propagate_subreg): Likewise.
12543 * loop-iv.c (get_biv_step_1): Likewise.
12544 * optabs.c (widen_operand): Likewise.
12545 * postreload.c (move2add_valid_value_p): Likewise.
12546 * recog.c (simplify_while_replacing): Likewise.
12547 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12548 (simplify_binary_operation_1): Likewise. Remove redundant
12549 mode equality check.
12551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12552 Alan Hayward <alan.hayward@arm.com>
12553 David Sherwood <david.sherwood@arm.com>
12555 * combine.c (combine_simplify_rtx): Add checks for
12556 is_a <scalar_int_mode>.
12557 (simplify_if_then_else): Likewise.
12558 (make_field_assignment): Likewise.
12559 (simplify_comparison): Likewise.
12560 * ifcvt.c (noce_try_bitop): Likewise.
12561 * loop-invariant.c (canonicalize_address_mult): Likewise.
12562 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12565 Alan Hayward <alan.hayward@arm.com>
12566 David Sherwood <david.sherwood@arm.com>
12568 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12569 is_a <scalar_int_mode> instead of != BLKmode.
12571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12572 Alan Hayward <alan.hayward@arm.com>
12573 David Sherwood <david.sherwood@arm.com>
12575 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12576 instead of != VOIDmode.
12577 * combine.c (if_then_else_cond): Likewise.
12578 (change_zero_ext): Likewise.
12579 * dwarf2out.c (mem_loc_descriptor): Likewise.
12580 (loc_descriptor): Likewise.
12581 * rtlanal.c (canonicalize_condition): Likewise.
12582 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12585 Alan Hayward <alan.hayward@arm.com>
12586 David Sherwood <david.sherwood@arm.com>
12588 * simplify-rtx.c (simplify_binary_operation_1): Use
12589 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12592 Alan Hayward <alan.hayward@arm.com>
12593 David Sherwood <david.sherwood@arm.com>
12595 * wide-int.h (int_traits<unsigned char>) New class.
12596 (int_traits<unsigned short>) Likewise.
12597 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12598 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12600 * combine.c (set_nonzero_bits_and_sign_copies): Use
12601 is_a <scalar_int_mode>.
12602 (find_split_point): Likewise.
12603 (combine_simplify_rtx): Likewise.
12604 (simplify_logical): Likewise.
12605 (expand_compound_operation): Likewise.
12606 (expand_field_assignment): Likewise.
12607 (make_compound_operation): Likewise.
12608 (extended_count): Likewise.
12609 (change_zero_ext): Likewise.
12610 (simplify_comparison): Likewise.
12611 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12612 (ucompare_loc_descriptor): Likewise.
12613 (minmax_loc_descriptor): Likewise.
12614 (mem_loc_descriptor): Likewise.
12615 (loc_descriptor): Likewise.
12616 * expmed.c (init_expmed_one_mode): Likewise.
12617 * lra-constraints.c (lra_constraint_offset): Likewise.
12618 * optabs.c (prepare_libcall_arg): Likewise.
12619 * postreload.c (move2add_note_store): Likewise.
12620 * reload.c (operands_match_p): Likewise.
12621 * rtl.h (load_extend_op): Likewise.
12622 * rtlhooks.c (gen_lowpart_general): Likewise.
12623 * simplify-rtx.c (simplify_truncation): Likewise.
12624 (simplify_unary_operation_1): Likewise.
12625 (simplify_binary_operation_1): Likewise.
12626 (simplify_const_binary_operation): Likewise.
12627 (simplify_const_relational_operation): Likewise.
12628 (simplify_subreg): Likewise.
12629 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12630 * var-tracking.c (adjust_mems): Likewise.
12631 (prepare_call_arguments): Likewise.
12633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12634 Alan Hayward <alan.hayward@arm.com>
12635 David Sherwood <david.sherwood@arm.com>
12637 * machmode.h (is_int_mode): New fuction.
12638 * combine.c (find_split_point): Use it.
12639 (combine_simplify_rtx): Likewise.
12640 (simplify_if_then_else): Likewise.
12641 (simplify_set): Likewise.
12642 (simplify_shift_const_1): Likewise.
12643 (simplify_comparison): Likewise.
12644 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12645 * cse.c (notreg_cost): Likewise.
12646 (cse_insn): Likewise.
12647 * cselib.c (cselib_lookup_1): Likewise.
12648 * dojump.c (do_jump_1): Likewise.
12649 (do_compare_rtx_and_jump): Likewise.
12650 * dse.c (get_call_args): Likewise.
12651 * dwarf2out.c (rtl_for_decl_init): Likewise.
12652 (native_encode_initializer): Likewise.
12653 * expmed.c (emit_store_flag_1): Likewise.
12654 (emit_store_flag): Likewise.
12655 * expr.c (convert_modes): Likewise.
12656 (store_field): Likewise.
12657 (expand_expr_real_1): Likewise.
12658 * fold-const.c (fold_read_from_constant_string): Likewise.
12659 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12660 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12661 * optabs.c (expand_binop): Likewise.
12662 (expand_unop): Likewise.
12663 (expand_abs_nojump): Likewise.
12664 (expand_one_cmpl_abs_nojump): Likewise.
12665 * simplify-rtx.c (mode_signbit_p): Likewise.
12666 (val_signbit_p): Likewise.
12667 (val_signbit_known_set_p): Likewise.
12668 (val_signbit_known_clear_p): Likewise.
12669 (simplify_relational_operation_1): Likewise.
12670 * tree.c (vector_type_mode): Likewise.
12672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12673 Alan Hayward <alan.hayward@arm.com>
12674 David Sherwood <david.sherwood@arm.com>
12676 * machmode.h (smallest_mode_for_size): Fix formatting.
12677 (smallest_int_mode_for_size): New function.
12678 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12679 instead of smallest_mode_for_size.
12680 * combine.c (make_extraction): Likewise.
12681 * config/arc/arc.c (arc_expand_movmem): Likewise.
12682 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12683 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12684 * config/s390/s390.c (s390_expand_insv): Likewise.
12685 * config/sparc/sparc.c (assign_int_registers): Likewise.
12686 * config/spu/spu.c (spu_function_value): Likewise.
12687 (spu_function_arg): Likewise.
12688 * coverage.c (get_gcov_type): Likewise.
12689 (get_gcov_unsigned_t): Likewise.
12690 * dse.c (find_shift_sequence): Likewise.
12691 * expmed.c (store_bit_field_1): Likewise.
12692 * expr.c (convert_move): Likewise.
12693 (store_field): Likewise.
12694 * internal-fn.c (expand_arith_overflow): Likewise.
12695 * optabs-query.c (get_best_extraction_insn): Likewise.
12696 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12697 * stor-layout.c (layout_type): Likewise.
12698 (initialize_sizetypes): Likewise.
12699 * targhooks.c (default_get_mask_mode): Likewise.
12700 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12703 Alan Hayward <alan.hayward@arm.com>
12704 David Sherwood <david.sherwood@arm.com>
12706 * machmode.h (opt_mode::else_blk): New function.
12707 (int_mode_for_mode): Declare.
12708 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12709 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12711 * cfgexpand.c (expand_debug_expr): Likewise.
12712 * combine.c (gen_lowpart_or_truncate): Likewise.
12713 (gen_lowpart_for_combine): Likewise.
12714 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12715 * config/avr/avr.c (avr_to_int_mode): Likewise.
12716 (avr_out_plus_1): Likewise.
12717 (avr_out_plus): Likewise.
12718 (avr_out_round): Likewise.
12719 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12720 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12721 (s390_expand_vcond): Likewise.
12722 * config/spu/spu.c (spu_split_immediate): Likewise.
12723 (spu_expand_mov): Likewise.
12724 * dse.c (get_stored_val): Likewise.
12725 * expmed.c (store_bit_field_1): Likewise.
12726 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12728 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12729 (extract_low_bits): Likewise.
12730 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12731 handling rather than repeating the check.
12732 (emit_group_store): Likewise.
12733 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12734 * optabs.c (expand_absneg_bit): Likewise.
12735 (expand_copysign_absneg): Likewise.
12736 (expand_copysign_bit): Likewise.
12737 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12738 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12739 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12740 * var-tracking.c (prepare_call_arguments): Likewise.
12741 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12742 int_mode_for_mode instead of mode_for_size.
12743 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12746 Alan Hayward <alan.hayward@arm.com>
12747 David Sherwood <david.sherwood@arm.com>
12749 * machmode.h (int_mode_for_size): New function.
12750 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12751 instead of mode_for_size.
12752 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12754 * combine.c (expand_field_assignment): Use int_mode_for_size
12755 instead of mode_for_size.
12756 (make_extraction): Likewise.
12757 (simplify_shift_const_1): Likewise.
12758 (simplify_comparison): Likewise.
12759 * dojump.c (do_jump): Likewise.
12760 * dwarf2out.c (mem_loc_descriptor): Likewise.
12761 * emit-rtl.c (init_derived_machine_modes): Likewise.
12762 * expmed.c (flip_storage_order): Likewise.
12763 (convert_extracted_bit_field): Likewise.
12764 * expr.c (copy_blkmode_from_reg): Likewise.
12765 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12766 * internal-fn.c (expand_mul_overflow): Likewise.
12767 * lower-subreg.c (simple_move): Likewise.
12768 * optabs-libfuncs.c (init_optabs): Likewise.
12769 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12770 * tree.c (vector_type_mode): Likewise.
12771 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12772 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12773 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12774 * tree-vect-stmts.c (vectorizable_load): Likewise.
12775 (vectorizable_store): Likewise.
12777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12778 Alan Hayward <alan.hayward@arm.com>
12779 David Sherwood <david.sherwood@arm.com>
12781 * coretypes.h (pod_mode): New type.
12782 (scalar_int_mode_pod): New typedef.
12783 * machmode.h (pod_mode): New class.
12784 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12785 * genmodes.c (emit_mode_int_n): Update accordingly.
12786 * lower-subreg.h (target_lower_subreg): Change type to
12787 scalar_int_mode_pod.
12788 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12789 scalar_int_mode_pod.
12791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12792 Alan Hayward <alan.hayward@arm.com>
12793 David Sherwood <david.sherwood@arm.com>
12795 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12796 machine_mode to scalar_int_mode.
12797 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12798 (rs6000_option_override_internal): Remove cast to int.
12799 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12800 machine_mode to scalar_int_mode.
12801 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12802 (rs6000_option_override_internal): Remove cast to int.
12803 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12804 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12806 * config/s390/s390.c (s390_expand_builtin): Likewise.
12807 * coretypes.h (scalar_int_mode): New type.
12808 (opt_scalar_int_mode): New typedef.
12809 * machmode.h (scalar_int_mode): New class.
12810 (scalar_int_mode::includes_p): New function.
12811 (byte_mode): Change type to scalar_int_mode.
12812 (word_mode): Likewise.
12813 (ptr_mode): Likewise.
12814 * emit-rtl.c (byte_mode): Likewise.
12815 (word_mode): Likewise.
12816 (ptr_mode): Likewise.
12817 (init_derived_machine_modes): Update accordingly.
12818 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12819 and MODE_PARTIAL_INT.
12820 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12821 opt_scalar_int_mode.
12823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12824 Alan Hayward <alan.hayward@arm.com>
12825 David Sherwood <david.sherwood@arm.com>
12827 * target.def (libgcc_floating_mode_supported_p): Take a
12829 * doc/tm.texi: Regenerate.
12830 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12832 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12833 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12837 Alan Hayward <alan.hayward@arm.com>
12838 David Sherwood <david.sherwood@arm.com>
12840 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12841 * doc/tm.texi: Regenerate.
12842 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12843 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12844 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12845 * targhooks.h (default_floatn_mode): Likewise.
12846 * targhooks.c (default_floatn_mode): Likewise.
12847 * tree.c (build_common_tree_nodes): Update accordingly.
12849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12850 Alan Hayward <alan.hayward@arm.com>
12851 David Sherwood <david.sherwood@arm.com>
12853 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12854 (mode_iterator::iterate_p): Likewise.
12855 (mode_iterator::get_wider): Likewise.
12856 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12859 Alan Hayward <alan.hayward@arm.com>
12860 David Sherwood <david.sherwood@arm.com>
12862 * coretypes.h (opt_scalar_float_mode): New typedef.
12863 * machmode.h (float_mode_for_size): New function.
12864 * emit-rtl.c (double_mode): Delete.
12865 (init_emit_once): Use float_mode_for_size.
12866 * stor-layout.c (layout_type): Likewise.
12867 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12870 Alan Hayward <alan.hayward@arm.com>
12871 David Sherwood <david.sherwood@arm.com>
12873 * output.h (assemble_real): Take a scalar_float_mode.
12874 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12875 * config/arm/arm.md (consttable_4): Likewise.
12876 (consttable_8): Likewise.
12877 (consttable_16): Likewise.
12878 * config/mips/mips.md (consttable_float): Likewise.
12879 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12880 * varasm.c (assemble_real): Take a scalar_float_mode.
12881 (output_constant_pool_2): Update accordingly.
12882 (output_constant): 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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12889 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12890 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12891 (native_encode_real): Likewise.
12892 (native_interpret_real): Likewise.
12893 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12894 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12897 Alan Hayward <alan.hayward@arm.com>
12898 David Sherwood <david.sherwood@arm.com>
12900 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12901 <scalar_float_mode>. Simplify.
12902 (gen_extend_conv_libfunc): Likewise.
12904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12905 Alan Hayward <alan.hayward@arm.com>
12906 David Sherwood <david.sherwood@arm.com>
12908 * coretypes.h (scalar_float_mode): New type.
12909 * machmode.h (mode_traits::from_int): Use machine_mode if
12910 USE_ENUM_MODES is defined.
12911 (is_a): New function.
12913 (dyn_cast): Likewise.
12914 (scalar_float_mode): New class.
12915 (scalar_float_mode::includes_p): New function.
12916 (is_float_mode): Likewise.
12917 * gdbhooks.py (MachineModePrinter): New class.
12918 (build_pretty_printer): Use it for scalar_float_mode.
12919 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12920 (format_helper::format_helper): Turn into a template.
12921 * genmodes.c (get_mode_class): New function.
12922 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12923 or machine_mode if none.
12924 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12925 as_a <scalar_float_mode>.
12926 * dwarf2out.c (mem_loc_descriptor): Likewise.
12927 (insert_float): Likewise.
12928 (add_const_value_attribute): Likewise.
12929 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12930 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12931 (expand_unop): Update accordingly.
12932 (expand_abs_nojump): Likewise.
12933 (expand_copysign_absneg): Take a scalar_float_mode.
12934 (expand_copysign_bit): Likewise.
12935 (expand_copysign): Update accordingly.
12937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12938 Alan Hayward <alan.hayward@arm.com>
12939 David Sherwood <david.sherwood@arm.com>
12941 * coretypes.h (opt_mode): New class.
12942 * machmode.h (opt_mode): Likewise.
12943 (opt_mode::else_void): New function.
12944 (opt_mode::require): Likewise.
12945 (opt_mode::exists): Likewise.
12946 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12947 (GET_MODE_2XWIDER_MODE): Likewise.
12948 (mode_iterator::get_wider): Update accordingly.
12949 (mode_iterator::get_2xwider): Likewise.
12950 (mode_iterator::get_known_wider): Likewise, turning into a template.
12951 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12952 forcing a wider mode to exist.
12953 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12954 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12955 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12956 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12957 * lower-subreg.c (init_lower_subreg): Likewise.
12958 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12959 on the final iteration.
12960 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12961 a wider mode exists before asking for a move pattern.
12962 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12963 forcing a wider mode to exist.
12964 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12965 returning false if no such mode exists.
12966 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12967 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12968 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12969 Avoid checking for a MODE_INT if we already know the mode is not a
12971 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12972 forcing a wider mode to exist.
12973 (expmed_mult_highpart_optab): Likewise.
12974 (expmed_mult_highpart): Likewise.
12975 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12977 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12978 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12979 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12980 * internal-fn.c (expand_mul_overflow): Update use of
12981 GET_MODE_2XWIDER_MODE.
12982 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12983 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12984 GET_MODE_WIDER_MODE.
12985 (convert_plusminus_to_widen): Likewise.
12986 * tree-switch-conversion.c (array_value_type): Likewise.
12987 * var-tracking.c (emit_note_insn_var_location): Likewise.
12988 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12989 Return false inside rather than outside the loop if no wider mode
12991 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12992 and GET_MODE_2XWIDER_MODE
12993 (can_compare_p): Use else_void.
12994 * gdbhooks.py (OptMachineModePrinter): New class.
12995 (build_pretty_printer): Use it for opt_mode.
12997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12998 Alan Hayward <alan.hayward@arm.com>
12999 David Sherwood <david.sherwood@arm.com>
13001 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13002 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13005 Alan Hayward <alan.hayward@arm.com>
13006 David Sherwood <david.sherwood@arm.com>
13008 * machmode.h (mode_traits): New structure.
13009 (get_narrowest_mode): New function.
13010 (mode_iterator::start): Likewise.
13011 (mode_iterator::iterate_p): Likewise.
13012 (mode_iterator::get_wider): Likewise.
13013 (mode_iterator::get_known_wider): Likewise.
13014 (mode_iterator::get_2xwider): Likewise.
13015 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13016 (FOR_EACH_MODE): Likewise.
13017 (FOR_EACH_MODE_FROM): Likewise.
13018 (FOR_EACH_MODE_UNTIL): Likewise.
13019 (FOR_EACH_WIDER_MODE): Likewise.
13020 (FOR_EACH_2XWIDER_MODE): Likewise.
13021 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13022 * combine.c (simplify_comparison): Likewise
13023 * config/i386/i386.c (type_natural_mode): Likewise.
13024 * cse.c (cse_insn): Likewise.
13025 * dse.c (find_shift_sequence): Likewise.
13026 * emit-rtl.c (init_derived_machine_modes): Likewise.
13027 (init_emit_once): Likewise.
13028 * explow.c (hard_function_value): Likewise.
13029 * expmed.c (extract_fixed_bit_field_1): Likewise.
13030 (extract_bit_field_1): Likewise.
13031 (expand_divmod): Likewise.
13032 (emit_store_flag_1): Likewise.
13033 * expr.c (init_expr_target): Likewise.
13034 (convert_move): Likewise.
13035 (alignment_for_piecewise_move): Likewise.
13036 (widest_int_mode_for_size): Likewise.
13037 (emit_block_move_via_movmem): Likewise.
13038 (copy_blkmode_to_reg): Likewise.
13039 (set_storage_via_setmem): Likewise.
13040 (compress_float_constant): Likewise.
13041 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13042 * optabs-query.c (get_best_extraction_insn): Likewise.
13043 * optabs.c (expand_binop): Likewise.
13044 (expand_twoval_unop): Likewise.
13045 (expand_twoval_binop): Likewise.
13046 (widen_leading): Likewise.
13047 (widen_bswap): Likewise.
13048 (expand_parity): Likewise.
13049 (expand_unop): Likewise.
13050 (prepare_cmp_insn): Likewise.
13051 (prepare_float_lib_cmp): Likewise.
13052 (expand_float): Likewise.
13053 (expand_fix): Likewise.
13054 (expand_sfix_optab): Likewise.
13055 * postreload.c (move2add_use_add2_insn): Likewise.
13056 * reg-stack.c (reg_to_stack): Likewise.
13057 * reginfo.c (choose_hard_reg_mode): Likewise.
13058 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13059 * stor-layout.c (mode_for_size): Likewise.
13060 (smallest_mode_for_size): Likewise.
13061 (mode_for_vector): Likewise.
13062 (finish_bitfield_representative): Likewise.
13063 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13064 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13065 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13066 * var-tracking.c (prepare_call_arguments): Likewise.
13068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13069 Alan Hayward <alan.hayward@arm.com>
13070 David Sherwood <david.sherwood@arm.com>
13072 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13073 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13074 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13075 * machmode.h (mode_size): Move earlier in file.
13076 (mode_precision): Likewise.
13077 (mode_inner): Likewise.
13078 (mode_nunits): Likewise.
13079 (mode_unit_size): Likewise.
13080 (unit_unit_precision): Likewise.
13081 (mode_wider): Likewise.
13082 (mode_2xwider): Likewise.
13083 (machine_mode): New class.
13084 (mode_to_bytes): New function.
13085 (mode_to_bits): Likewise.
13086 (mode_to_precision): Likewise.
13087 (mode_to_inner): Likewise.
13088 (mode_to_unit_size): Likewise.
13089 (mode_to_unit_precision): Likewise.
13090 (mode_to_nunits): Likewise.
13091 (GET_MODE_SIZE): Use mode_to_bytes.
13092 (GET_MODE_BITSIZE): Use mode_to_bits.
13093 (GET_MODE_PRECISION): Use mode_to_precision.
13094 (GET_MODE_INNER): Use mode_to_inner.
13095 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13096 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13097 (GET_MODE_NUNITS): Use mode_to_nunits.
13098 * system.h (ALWAYS_INLINE): New macro.
13099 * config/powerpcspe/powerpcspe-c.c
13100 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13101 int for arg1_mode and arg2_mode.
13103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13104 Alan Hayward <alan.hayward@arm.com>
13105 David Sherwood <david.sherwood@arm.com>
13107 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13108 Prefix mode names with E_ in case statements.
13109 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13110 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13111 (aarch64_split_simd_move): Likewise.
13112 (aarch64_gen_storewb_pair): Likewise.
13113 (aarch64_gen_loadwb_pair): Likewise.
13114 (aarch64_gen_store_pair): Likewise.
13115 (aarch64_gen_load_pair): Likewise.
13116 (aarch64_get_condition_code_1): Likewise.
13117 (aarch64_constant_pool_reload_icode): Likewise.
13118 (get_rsqrte_type): Likewise.
13119 (get_rsqrts_type): Likewise.
13120 (get_recpe_type): Likewise.
13121 (get_recps_type): Likewise.
13122 (aarch64_gimplify_va_arg_expr): Likewise.
13123 (aarch64_simd_container_mode): Likewise.
13124 (aarch64_emit_load_exclusive): Likewise.
13125 (aarch64_emit_store_exclusive): Likewise.
13126 (aarch64_expand_compare_and_swap): Likewise.
13127 (aarch64_gen_atomic_cas): Likewise.
13128 (aarch64_emit_bic): Likewise.
13129 (aarch64_emit_atomic_swap): Likewise.
13130 (aarch64_emit_atomic_load_op): Likewise.
13131 (aarch64_evpc_trn): Likewise.
13132 (aarch64_evpc_uzp): Likewise.
13133 (aarch64_evpc_zip): Likewise.
13134 (aarch64_evpc_ext): Likewise.
13135 (aarch64_evpc_rev): Likewise.
13136 (aarch64_evpc_dup): Likewise.
13137 (aarch64_gen_ccmp_first): Likewise.
13138 (aarch64_gen_ccmp_next): Likewise.
13139 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13140 (alpha_emit_xfloating_libcall): Likewise.
13141 (emit_insxl): Likewise.
13142 (alpha_arg_type): Likewise.
13143 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13144 (arc_preferred_simd_mode): Likewise.
13145 (arc_secondary_reload): Likewise.
13146 (get_arc_condition_code): Likewise.
13147 (arc_print_operand): Likewise.
13148 (arc_legitimate_constant_p): Likewise.
13149 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13150 * config/arc/arc.md (casesi_load): Likewise.
13151 (casesi_compact_jump): Likewise.
13152 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13153 (cc_use_register): Likewise.
13154 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13155 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13156 (arm_init_iwmmxt_builtins): Likewise.
13157 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13158 (neon_expand_vector_init): Likewise.
13159 (arm_attr_length_move_neon): Likewise.
13160 (maybe_get_arm_condition_code): Likewise.
13161 (arm_emit_vector_const): Likewise.
13162 (arm_preferred_simd_mode): Likewise.
13163 (arm_output_iwmmxt_tinsr): Likewise.
13164 (thumb1_output_casesi): Likewise.
13165 (thumb2_output_casesi): Likewise.
13166 (arm_emit_load_exclusive): Likewise.
13167 (arm_emit_store_exclusive): Likewise.
13168 (arm_expand_compare_and_swap): Likewise.
13169 (arm_evpc_neon_vuzp): Likewise.
13170 (arm_evpc_neon_vzip): Likewise.
13171 (arm_evpc_neon_vrev): Likewise.
13172 (arm_evpc_neon_vtrn): Likewise.
13173 (arm_evpc_neon_vext): Likewise.
13174 (arm_validize_comparison): Likewise.
13175 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13176 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13177 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13178 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13179 (c6x_preferred_simd_mode): Likewise.
13180 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13181 (epiphany_rtx_costs): Likewise.
13182 * config/epiphany/predicates.md (proper_comparison_operator):
13184 * config/frv/frv.c (condexec_memory_operand): Likewise.
13185 (frv_emit_move): Likewise.
13186 (output_move_single): Likewise.
13187 (output_condmove_single): Likewise.
13188 (frv_hard_regno_mode_ok): Likewise.
13189 (frv_matching_accg_mode): Likewise.
13190 * config/h8300/h8300.c (split_adds_subs): Likewise.
13191 (h8300_rtx_costs): Likewise.
13192 (h8300_print_operand): Likewise.
13193 (compute_mov_length): Likewise.
13194 (output_logical_op): Likewise.
13195 (compute_logical_op_length): Likewise.
13196 (compute_logical_op_cc): Likewise.
13197 (h8300_shift_needs_scratch_p): Likewise.
13198 (output_a_shift): Likewise.
13199 (compute_a_shift_length): Likewise.
13200 (compute_a_shift_cc): Likewise.
13201 (expand_a_rotate): Likewise.
13202 (output_a_rotate): Likewise.
13203 * config/i386/i386.c (classify_argument): Likewise.
13204 (function_arg_advance_32): Likewise.
13205 (function_arg_32): Likewise.
13206 (function_arg_64): Likewise.
13207 (function_value_64): Likewise.
13208 (ix86_gimplify_va_arg): Likewise.
13209 (ix86_legitimate_constant_p): Likewise.
13210 (put_condition_code): Likewise.
13211 (split_double_mode): Likewise.
13212 (ix86_avx256_split_vector_move_misalign): Likewise.
13213 (ix86_expand_vector_logical_operator): Likewise.
13214 (ix86_split_idivmod): Likewise.
13215 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13216 (ix86_build_const_vector): Likewise.
13217 (ix86_build_signbit_mask): Likewise.
13218 (ix86_match_ccmode): Likewise.
13219 (ix86_cc_modes_compatible): Likewise.
13220 (ix86_expand_branch): Likewise.
13221 (ix86_expand_sse_cmp): Likewise.
13222 (ix86_expand_sse_movcc): Likewise.
13223 (ix86_expand_int_sse_cmp): Likewise.
13224 (ix86_expand_vec_perm_vpermi2): Likewise.
13225 (ix86_expand_vec_perm): Likewise.
13226 (ix86_expand_sse_unpack): Likewise.
13227 (ix86_expand_int_addcc): Likewise.
13228 (ix86_split_to_parts): Likewise.
13229 (ix86_vectorize_builtin_gather): Likewise.
13230 (ix86_vectorize_builtin_scatter): Likewise.
13231 (avx_vpermilp_parallel): Likewise.
13232 (inline_memory_move_cost): Likewise.
13233 (ix86_tieable_integer_mode_p): Likewise.
13234 (x86_maybe_negate_const_int): Likewise.
13235 (ix86_expand_vector_init_duplicate): Likewise.
13236 (ix86_expand_vector_init_one_nonzero): Likewise.
13237 (ix86_expand_vector_init_one_var): Likewise.
13238 (ix86_expand_vector_init_concat): Likewise.
13239 (ix86_expand_vector_init_interleave): Likewise.
13240 (ix86_expand_vector_init_general): Likewise.
13241 (ix86_expand_vector_set): Likewise.
13242 (ix86_expand_vector_extract): Likewise.
13243 (emit_reduc_half): Likewise.
13244 (ix86_emit_i387_round): Likewise.
13245 (ix86_mangle_type): Likewise.
13246 (ix86_expand_round_sse4): Likewise.
13247 (expand_vec_perm_blend): Likewise.
13248 (canonicalize_vector_int_perm): Likewise.
13249 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13250 (expand_vec_perm_1): Likewise.
13251 (expand_vec_perm_interleave3): Likewise.
13252 (expand_vec_perm_even_odd_pack): Likewise.
13253 (expand_vec_perm_even_odd_1): Likewise.
13254 (expand_vec_perm_broadcast_1): Likewise.
13255 (ix86_vectorize_vec_perm_const_ok): Likewise.
13256 (ix86_expand_vecop_qihi): Likewise.
13257 (ix86_expand_mul_widen_hilo): Likewise.
13258 (ix86_expand_sse2_abs): Likewise.
13259 (ix86_expand_pextr): Likewise.
13260 (ix86_expand_pinsr): Likewise.
13261 (ix86_preferred_simd_mode): Likewise.
13262 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13263 * config/i386/sse.md (*andnot<mode>3): Likewise.
13264 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13265 (*<code><mode>3): Likewise.
13266 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13267 (ia64_expand_atomic_op): Likewise.
13268 (ia64_arg_type): Likewise.
13269 (ia64_mode_to_int): Likewise.
13270 (ia64_scalar_mode_supported_p): Likewise.
13271 (ia64_vector_mode_supported_p): Likewise.
13272 (expand_vec_perm_broadcast): Likewise.
13273 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13274 (iq2000_function_arg_advance): Likewise.
13275 (iq2000_function_arg): Likewise.
13276 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13277 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13278 (m68k_libcall_value): Likewise.
13279 (m68k_function_value): Likewise.
13280 (sched_attr_op_type): Likewise.
13281 * config/mcore/mcore.c (mcore_output_move): Likewise.
13282 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13284 (microblaze_function_arg): Likewise.
13285 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13286 (mips_print_operand): Likewise.
13287 (mips_mode_ok_for_mov_fmt_p): Likewise.
13288 (mips_vector_mode_supported_p): Likewise.
13289 (mips_preferred_simd_mode): Likewise.
13290 (mips_expand_vpc_loongson_even_odd): Likewise.
13291 (mips_expand_vec_unpack): Likewise.
13292 (mips_expand_vi_broadcast): Likewise.
13293 (mips_expand_vector_init): Likewise.
13294 (mips_expand_vec_reduc): Likewise.
13295 (mips_expand_msa_cmp): Likewise.
13296 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13297 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13298 (cc_flags_for_mode): Likewise.
13299 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13300 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13301 (nds32_output_casesi_pc_relative): Likewise.
13302 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13303 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13304 (nvptx_gen_unpack): Likewise.
13305 (nvptx_gen_pack): Likewise.
13306 (nvptx_gen_shuffle): Likewise.
13307 (nvptx_gen_wcast): Likewise.
13308 (nvptx_preferred_simd_mode): Likewise.
13309 * config/pa/pa.c (pa_secondary_reload): Likewise.
13310 * config/pa/predicates.md (base14_operand): Likewise.
13311 * config/powerpcspe/powerpcspe-c.c
13312 (altivec_resolve_overloaded_builtin): Likewise.
13313 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13315 (rs6000_preferred_simd_mode): Likewise.
13316 (output_vec_const_move): Likewise.
13317 (rs6000_expand_vector_extract): Likewise.
13318 (rs6000_split_vec_extract_var): Likewise.
13319 (reg_offset_addressing_ok_p): Likewise.
13320 (rs6000_legitimate_offset_address_p): Likewise.
13321 (rs6000_legitimize_address): Likewise.
13322 (rs6000_emit_set_const): Likewise.
13323 (rs6000_const_vec): Likewise.
13324 (rs6000_emit_move): Likewise.
13325 (spe_build_register_parallel): Likewise.
13326 (rs6000_darwin64_record_arg_recurse): Likewise.
13327 (swap_selector_for_mode): Likewise.
13328 (spe_init_builtins): Likewise.
13329 (paired_init_builtins): Likewise.
13330 (altivec_init_builtins): Likewise.
13331 (do_load_for_compare): Likewise.
13332 (rs6000_generate_compare): Likewise.
13333 (rs6000_expand_float128_convert): Likewise.
13334 (emit_load_locked): Likewise.
13335 (emit_store_conditional): Likewise.
13336 (rs6000_output_function_epilogue): Likewise.
13337 (rs6000_handle_altivec_attribute): Likewise.
13338 (rs6000_function_value): Likewise.
13339 (emit_fusion_gpr_load): Likewise.
13340 (emit_fusion_p9_load): Likewise.
13341 (emit_fusion_p9_store): Likewise.
13342 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13343 (fusion_gpr_mem_load): Likewise.
13344 (fusion_addis_mem_combo_load): Likewise.
13345 (fusion_addis_mem_combo_store): Likewise.
13346 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13347 (fusion_gpr_mem_load): Likewise.
13348 (fusion_addis_mem_combo_load): Likewise.
13349 (fusion_addis_mem_combo_store): Likewise.
13350 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13352 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13353 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13354 (rs6000_preferred_simd_mode): Likewise.
13355 (output_vec_const_move): Likewise.
13356 (rs6000_expand_vector_extract): Likewise.
13357 (rs6000_split_vec_extract_var): Likewise.
13358 (reg_offset_addressing_ok_p): Likewise.
13359 (rs6000_legitimate_offset_address_p): Likewise.
13360 (rs6000_legitimize_address): Likewise.
13361 (rs6000_emit_set_const): Likewise.
13362 (rs6000_const_vec): Likewise.
13363 (rs6000_emit_move): Likewise.
13364 (rs6000_darwin64_record_arg_recurse): Likewise.
13365 (swap_selector_for_mode): Likewise.
13366 (paired_init_builtins): Likewise.
13367 (altivec_init_builtins): Likewise.
13368 (rs6000_expand_float128_convert): Likewise.
13369 (emit_load_locked): Likewise.
13370 (emit_store_conditional): Likewise.
13371 (rs6000_output_function_epilogue): Likewise.
13372 (rs6000_handle_altivec_attribute): Likewise.
13373 (rs6000_function_value): Likewise.
13374 (emit_fusion_gpr_load): Likewise.
13375 (emit_fusion_p9_load): Likewise.
13376 (emit_fusion_p9_store): Likewise.
13377 * config/rx/rx.c (rx_gen_move_template): Likewise.
13378 (flags_from_mode): Likewise.
13379 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13380 (s390_slb_comparison): Likewise.
13381 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13382 (s390_vector_mode_supported_p): Likewise.
13383 (s390_cc_modes_compatible): Likewise.
13384 (s390_match_ccmode_set): Likewise.
13385 (s390_canonicalize_comparison): Likewise.
13386 (s390_emit_compare_and_swap): Likewise.
13387 (s390_branch_condition_mask): Likewise.
13388 (s390_rtx_costs): Likewise.
13389 (s390_secondary_reload): Likewise.
13390 (__SECONDARY_RELOAD_CASE): Likewise.
13391 (s390_expand_cs): Likewise.
13392 (s390_preferred_simd_mode): Likewise.
13393 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13394 * config/sh/sh.c (sh_print_operand): Likewise.
13395 (dump_table): Likewise.
13396 (sh_secondary_reload): Likewise.
13397 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13398 * config/sh/sh.md (casesi_worker_1): Likewise.
13399 (casesi_worker_2): Likewise.
13400 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13401 (fcc_comparison_operator): Likewise.
13402 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13403 (emit_soft_tfmode_cvt): Likewise.
13404 (sparc_preferred_simd_mode): Likewise.
13405 (output_cbranch): Likewise.
13406 (sparc_print_operand): Likewise.
13407 (sparc_expand_vec_perm_bmask): Likewise.
13408 (vector_init_bshuffle): Likewise.
13409 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13410 (spu_vector_mode_supported_p): Likewise.
13411 (spu_expand_insv): Likewise.
13412 (spu_emit_branch_or_set): Likewise.
13413 (spu_handle_vector_attribute): Likewise.
13414 (spu_builtin_splats): Likewise.
13415 (spu_builtin_extract): Likewise.
13416 (spu_builtin_promote): Likewise.
13417 (spu_expand_sign_extend): Likewise.
13418 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13419 (tilegx_simd_int): Likewise.
13420 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13421 (tilepro_simd_int): Likewise.
13422 * config/v850/v850.c (const_double_split): Likewise.
13423 (v850_print_operand): Likewise.
13424 (ep_memory_offset): Likewise.
13425 * config/vax/vax.c (vax_rtx_costs): Likewise.
13426 (vax_output_int_move): Likewise.
13427 (vax_output_int_add): Likewise.
13428 (vax_output_int_subtract): Likewise.
13429 * config/visium/predicates.md (visium_branch_operator): Likewise.
13430 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13431 (visium_print_operand_address): Likewise.
13432 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13433 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13434 (xtensa_expand_conditional_branch): Likewise.
13435 (xtensa_copy_incoming_a7): Likewise.
13436 (xtensa_output_literal): Likewise.
13437 * dfp.c (decimal_real_maxval): Likewise.
13438 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13441 Alan Hayward <alan.hayward@arm.com>
13442 David Sherwood <david.sherwood@arm.com>
13444 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13445 (mode_nunits_inline): Likewise.
13446 (mode_inner_inline): Likewise.
13447 (mode_unit_size_inline): Likewise.
13448 (mode_unit_precision_inline): Likewise.
13449 (emit_insn_modes_h): Likewise. Also emit a #define of the
13451 (emit_mode_wider): Add an E_ prefix to mode names.
13452 (emit_mode_complex): Likewise.
13453 (emit_mode_inner): Likewise.
13454 (emit_mode_adjustments): Likewise.
13455 (emit_mode_int_n): Likewise.
13456 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13457 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13458 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13459 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13460 (CRC32_BUILTIN, ENTRY): Likewise.
13461 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13462 (aarch64_pop_regs): Likewise.
13463 (aarch64_process_components): Likewise.
13464 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13465 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13466 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13467 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13468 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13469 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13471 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13472 (pcmpistr): Likewise.
13473 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13474 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13475 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13477 * config/rl78/rl78.c (mduc_regs): Likewise.
13478 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13479 (htm_expand_builtin): Likewise.
13480 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13481 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13482 * config/xtensa/xtensa.c (print_operand): Likewise.
13483 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13484 (NUM_MODE_VECTOR_INT): Likewise.
13485 * genoutput.c (null_operand): Likewise.
13486 (output_operand_data): Likewise.
13487 * genrecog.c (print_parameter_value): Likewise.
13488 * lra.c (debug_operand_data): Likewise.
13490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13492 * dwarf2out.c (scompare_loc_descriptor_wide)
13493 (scompare_loc_descriptor_narrow): New functions, split out from...
13494 (scompare_loc_descriptor): ...here.
13495 * expmed.c (emit_store_flag_int): New function, split out from...
13496 (emit_store_flag): ...here.
13498 2017-08-30 Richard Biener <rguenther@suse.de>
13500 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13501 (dwarf2out_early_finish): Move setting of AT_pubnames from
13502 early debug output to early finish.
13504 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13506 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13507 and -mdata-region to the assembler.
13509 2017-08-30 Richard Biener <rguenther@suse.de>
13511 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13513 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13514 (dwarf2out_early_global_decl): Only generate a DIE for the
13515 abstract origin if it doesn't already exist or is a declaration DIE.
13516 (resolve_addr): Do not add the linkage name twice when
13517 generating a stub DIE for the DW_TAG_GNU_call_site target.
13519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13521 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13522 Use machine_mode rather than int for arg1_mode.
13524 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13527 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13528 that the second argument of the built-in functions to unpack
13529 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13530 switch statement instead a lot of if statements.
13531 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13532 Allow 64-bit values to be in Altivec registers as well as
13533 traditional floating point registers.
13534 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13536 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13538 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13539 MAX_REGS_PER_ADDRESS == 1.
13541 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13543 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13544 * config/i386/i386.c (ix86_option_override_internal): Simplify
13545 setting of opts->x_flag_entry.
13547 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13548 Jakub Jelinek <jakub@redhat.com>
13549 Richard Biener <rguenther@suse.de>
13551 PR tree-optimization/81503
13552 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13553 folded constant fits in the target type; reorder tests for clarity.
13555 2017-08-29 Martin Liska <mliska@suse.cz>
13557 * passes.def: Include pass_lower_switch.
13558 * stmt.c (dump_case_nodes): Remove and move to
13559 tree-switch-conversion.
13560 (case_values_threshold): Likewise.
13561 (expand_switch_as_decision_tree_p): Likewise.
13562 (emit_case_decision_tree): Likewise.
13563 (expand_case): Likewise.
13564 (balance_case_nodes): Likewise.
13565 (node_has_low_bound): Likewise.
13566 (node_has_high_bound): Likewise.
13567 (node_is_bounded): Likewise.
13568 (emit_case_nodes): Likewise.
13569 (struct simple_case_node): New struct.
13570 (add_case_node): Remove.
13571 (emit_case_dispatch_table): Use vector instead of case_list.
13572 (reset_out_edges_aux): Remove.
13573 (compute_cases_per_edge): Likewise.
13574 (expand_case): Build list of simple_case_node.
13575 (expand_sjlj_dispatch_table): Use it.
13576 * tree-switch-conversion.c (struct case_node): Moved from
13577 stmt.c and adjusted.
13578 (emit_case_nodes): Likewise.
13579 (node_has_low_bound): Likewise.
13580 (node_has_high_bound): Likewise.
13581 (node_is_bounded): Likewise.
13582 (case_values_threshold): Likewise.
13583 (reset_out_edges_aux): Likewise.
13584 (compute_cases_per_edge): Likewise.
13585 (add_case_node): Likewise.
13586 (dump_case_nodes): Likewise.
13587 (balance_case_nodes): Likewise.
13588 (expand_switch_as_decision_tree_p): Likewise.
13589 (emit_jump): Likewise.
13590 (emit_case_decision_tree): Likewise.
13591 (try_switch_expansion): Likewise.
13592 (do_jump_if_equal): Likewise.
13593 (emit_cmp_and_jump_insns): Likewise.
13594 (fix_phi_operands_for_edge): New function.
13595 (record_phi_operand_mapping): Likewise.
13596 (class pass_lower_switch): New pass.
13597 (pass_lower_switch::execute): New function.
13598 (make_pass_lower_switch): Likewise.
13599 (conditional_probability):
13600 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13601 * tree-pass.h: Add make_pass_lower_switch.
13603 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13606 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13609 2017-08-29 Richard Biener <rguenther@suse.de>
13611 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13612 we do not add a DW_AT_inline attribute twice.
13613 (gen_subprogram_die): Remove code setting DW_AT_inline on
13614 DECL_ABSTRACT_P nodes.
13616 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13618 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13619 calls to internal functions.
13620 (gimplify_modify_expr): Likewise.
13621 * tree-call-cdce.c (use_internal_fn): Likewise.
13622 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13623 (convert_to_divmod): Set the nothrow flag.
13624 * tree-if-conv.c (predicate_mem_writes): Likewise.
13625 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13626 (vectorizable_call): Likewise.
13627 (vectorizable_store): Likewise.
13628 (vectorizable_load): Likewise.
13629 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13630 (vect_recog_mask_conversion_pattern): Likewise.
13632 2017-08-29 Martin Liska <mliska@suse.cz>
13635 * gcc.c (driver_handle_option): Add new argument.
13636 * opts-common.c (handle_option): Pass
13637 target_option_override_hook.
13638 * opts-global.c (lang_handle_option): Add new option.
13639 (set_default_handlers): Add new argument.
13640 (decode_options): Likewise.
13641 * opts.c (target_handle_option): Likewise.
13642 (common_handle_option): Call target_option_override_hook.
13643 * opts.h (struct cl_option_handler_func): Add hook for
13644 target option override.
13645 (struct cl_option_handlers): Likewise.
13646 (set_default_handlers): Add new argument.
13647 (decode_options): Likewise.
13648 (common_handle_option): Likewise.
13649 (target_handle_option): Likewise.
13650 * toplev.c (toplev::main): Pass targetm.target_option.override
13653 2017-08-29 Richard Biener <rguenther@suse.de>
13654 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13656 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13657 life to the active subtree.
13659 2017-08-28 Jeff Law <law@redhat.com>
13661 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13662 derive_equivalences.
13663 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13664 Code moved into....
13665 (edge_info::derive_equivalences): New private member function
13667 * tree-ssa-dom.c (class edge_info): Changed from a struct
13668 to a class. Add ctor/dtor, methods and data members.
13669 (edge_info::edge_info): Renamed from allocate_edge_info.
13670 Initialize additional members.
13671 (edge_info::~edge_info): New.
13672 (free_dom_edge_info): Delete the edge info.
13673 (record_edge_info): Use new class & associated member functions.
13674 Tighten forms for testing for edge equivalences.
13675 (record_temporary_equivalences): Iterate over the simple
13676 equivalences rather than assuming there's only one per edge.
13677 (cprop_into_successor_phis): Iterate over the simple
13678 equivalences rather than assuming there's only one per edge.
13679 (optimize_stmt): Use operand_equal_p rather than pointer
13680 equality for mini-DSE code.
13682 2017-08-28 Nathan Sidwell <nathan@acm.org>
13684 * gcc.c (execute): Fold SIGPIPE handling into switch
13685 statement. Adjust internal error message.
13687 2017-08-28 Richard Biener <rguenther@suse.de>
13690 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13691 Do nothing for removed DIEs.
13693 2017-08-28 Richard Biener <rguenther@suse.de>
13695 PR tree-optimization/81977
13696 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13699 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13702 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13703 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13704 using targetm.gen_mem_thread_fence.
13706 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13709 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13710 predicate to register_operand. Reorder operands.
13711 (*btr<mode>): Ditto.
13712 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13713 (*btr<mode>_mask): Ditto.
13715 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13717 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13718 * config/rs6000/xmmintrin.h: New file.
13719 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13721 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13724 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13725 parameter and_insn and return it.
13726 (recombine_lvx_pattern): Insert a copy to ensure availability of
13727 the base register of the copied masking operation at the point of
13728 the instruction replacement.
13729 (recombine_stvx_pattern): Likewise.
13731 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13733 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13734 undocumented switches.
13735 (-mpower9-dform-vector): Likewise.
13736 (-mpower9-dform): Likewise.
13737 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13738 comments to delete references to -mpower9-dform* switches.
13739 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13740 Delete reference to -mpower9-dform* switches, test for
13741 -mpower9-vector instead.
13742 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13743 (OTHER_P9_VECTOR_MASKS): Likewise.
13744 (POWERPC_MASKS): Likewise.
13745 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13746 tests against -mpower9-dform* to -mpower9-vector. Delete code
13747 that checked for -mpower9-dform* consistancy with other options.
13748 Add test for -mpower9-misc to enable other power9 switches.
13749 (rs6000_init_hard_regno_mode_ok): Likewise.
13750 (rs6000_option_override_internal): Likewise.
13751 (rs6000_emit_prologue): Likewise.
13752 (rs6000_emit_epilogue): Likewise.
13753 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13754 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13755 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13757 (emit_fusion_p9_store): Likewise.
13758 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13759 resetting these macros if the assembler does not support ISA 3.0
13761 (TARGET_P9_DFORM_VECTOR): Likewise.
13762 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13763 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13765 2017-08-25 Alan Modra <amodra@gmail.com>
13767 PR rtl-optimization/81747
13768 * cse.c (cse_extended_basic_block): Don't attempt to record
13769 equivalences for degenerate conditional branches that branch
13770 to their fall-through.
13772 2017-08-24 Martin Sebor <msebor@redhat.com>
13774 PR middle-end/81908
13775 * gimple-fold.c (size_must_be_zero_p): New function.
13776 (gimple_fold_builtin_memory_op): Call it.
13778 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13780 * config/rs6000/mm_malloc.h: New file.
13782 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13784 PR tree-optimization/81913
13785 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13786 analysis when either IVs in condition can wrap.
13788 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13790 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13791 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13793 2017-08-24 Richard Biener <rguenther@suse.de>
13796 * targhooks.c (default_target_can_inline_p): Properly
13797 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13798 is present and always compare.
13799 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13800 infer -mfpmath=sse from TARGET_SSE_P.
13801 (ix86_can_inline_p): Properly use target_option_default_node when
13802 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13804 2017-08-24 Richard Biener <rguenther@suse.de>
13807 * dwarf2out.c (output_die): Handle flag_generate_offload like
13809 (output_comp_unit): Likewise.
13810 (gen_array_type_die): Likewise.
13811 (dwarf2out_early_finish): Likewise.
13812 (note_variable_value_in_expr): Likewise.
13813 (dwarf2out_finish): Likewise. Adjust assert.
13814 * cgraphunit.c (symbol_table::compile): Move setting of
13815 flag_generate_offload earlier ...
13816 (symbol_table::finalize_compilation_unit): ... here, before
13817 early debug finalization.
13819 2017-08-24 Richard Biener <rguenther@suse.de>
13821 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13822 and ipa-fnsummary.h.
13823 (ix86_can_inline_p): When ix86_fpmath flags do not match
13824 check whether the callee uses FP math at all.
13826 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13828 PR middle-end/81931
13829 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13830 instead of TYPE_PRECISION.
13832 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13833 Alan Hayward <alan.hayward@arm.com>
13834 David Sherwood <david.sherwood@arm.com>
13836 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13838 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13840 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13842 (ix86_frame::stack_realign_allocate): New field.
13843 (struct machine_frame_state): Modify comments.
13844 (machine_frame_state::sp_realigned_fp_end): New field.
13845 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13846 layout calculation.
13847 (sp_valid_at): Add assertion to assure no attempt to access invalid
13848 offset of a realigned stack.
13849 (fp_valid_at): Likewise.
13850 (choose_baseaddr): Modify comments.
13851 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13852 ix86_expand_prologue.
13853 (ix86_expand_prologue): Modify stack realignment and allocation.
13854 (ix86_expand_epilogue): Modify comments.
13855 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13856 avx2_runtime, avx512f, and avx512f_runtime.
13858 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13860 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13861 (mstackrealign): Do not init to -1.
13862 * config/i386/i386.c (ix86_option_override_internal):
13863 Check opts_set, not opts when setting default value of
13864 opts->x_ix86_force_align_arg_pointer.
13866 2017-08-23 Richard Biener <rguenther@suse.de>
13868 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13869 lang_hooks.decl_printable_name.
13870 * print-rtl-function.c (print_rtx_function): Likewise.
13871 * tree-pretty-print.c (dump_function_header): Likewise.
13873 2017-08-23 Richard Biener <rguenther@suse.de>
13876 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13877 -g0 at compile-time.
13879 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13881 PR middle-end/19706
13882 * doc/sourcebuild.texi (Other hardware attributes):
13885 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13887 PR middle-end/19706
13888 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13889 Fix single-use check.
13891 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13893 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13895 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13897 * doc/install.texi: Modify to add more details on running selected
13900 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13902 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13903 is combined with -mabi=ms.
13904 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13907 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13909 PR tree-optimization/81488
13910 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13911 and cached_basis fields.
13912 (MAX_SPREAD): New constant.
13913 (alloc_cand_and_find_basis): Initialize new fields.
13914 (clear_visited): New function.
13915 (create_phi_basis_1): Rename from create_phi_basis, set visited
13916 and cached_basis fields.
13917 (create_phi_basis): New wrapper function.
13918 (phi_add_costs_1): Rename from phi_add_costs, add spread
13919 parameter, set visited field, short-circuit when limits reached.
13920 (phi_add_costs): New wrapper function.
13921 (record_phi_increments_1): Rename from record_phi_increments, set
13923 (record_phi_increments): New wrapper function.
13924 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13925 (phi_incr_cost): New wrapper function.
13926 (all_phi_incrs_profitable_1): Rename from
13927 all_phi_incrs_profitable, set visited field.
13928 (all_phi_incrs_profitable): New wrapper function.
13930 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13931 Alan Hayward <alan.hayward@arm.com>
13932 David Sherwood <david.sherwood@arm.com>
13934 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13935 that takes the outer and inner modes.
13936 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13937 comparison as the canonical test for a paradoxical subreg.
13938 * combine.c (simplify_set): Use paradoxical_subreg_p.
13939 (make_extraction): Likewise.
13940 (force_to_mode): Likewise.
13941 (rtx_equal_for_field_assignment_p): Likewise.
13942 (gen_lowpart_for_combine): Likewise.
13943 (simplify_comparison): Likewise.
13944 * cse.c (equiv_constant): Likewise.
13945 * expmed.c (store_bit_field_1): Likewise.
13946 * final.c (alter_subreg): Likewise.
13947 * fwprop.c (propagate_rtx): Likewise.
13948 (forward_propagate_subreg): Likewise.
13949 * ira-conflicts.c (ira_build_conflicts): Likewise.
13950 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13951 * lra-constraints.c (curr_insn_transform): Likewise.
13952 (split_reg): Likewise.
13953 * lra-eliminations.c (move_plus_up): Likewise.
13954 (lra_eliminate_regs_1): Likewise.
13955 * recog.c (general_operand): Likewise.
13956 * ree.c (combine_reaching_defs): Likewise.
13957 * reload.c (push_reload): Likewise.
13958 (find_reloads): Likewise.
13959 * reload1.c (elimination_effects): Likewise.
13960 (compute_reload_subreg_offset): Likewise.
13961 (choose_reload_regs): Likewise.
13962 * rtlanal.c (subreg_lsb_1): Likewise.
13963 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13964 (simplify_subreg): Likewise.
13965 * var-tracking.c (track_loc_p): Likewise.
13966 * emit-rtl.c (byte_lowpart_offset): Likewise.
13967 (paradoxical_subreg_p): Delete out-of-line definition.
13969 2017-08-22 Jeff Law <law@redhat.com>
13971 PR tree-optimization/81741
13972 PR tree-optimization/71947
13973 * tree-ssa-dom.c: Include tree-inline.h.
13974 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13975 equivalences if one is more expensive to compute than the other.
13976 * tree-ssa-scopedtables.h (class const_or_copies): Make
13977 record_const_or_copy_raw method private.
13978 (class avail_exprs_stack): New method simplify_binary_operation.
13979 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13980 avail_exprs_stack::simplify_binary_operation as needed.
13981 (avail_exprs_stack::simplify_binary_operation): New function.
13983 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13985 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13986 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13987 (DOT_SYMBOLS): Likewise.
13988 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13989 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13990 (RS6000_ABI_NAME): Likewise.
13991 (TARGET_CMODEL): Likewise.
13992 (TOC_SECTION_ASM_OP): Likewise.
13993 (SET_CMODEL): New macro.
13994 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13996 2017-08-22 Richard Biener <rguenther@suse.de>
13998 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13999 to allow for free-lang-data replacements similar to verify_type_variant.
14001 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14002 Alan Hayward <alan.hayward@arm.com>
14003 David Sherwood <david.sherwood@arm.com>
14005 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14006 VOIDmode for the LABEL_REF.
14008 2017-08-22 Richard Biener <rguenther@suse.de>
14010 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14012 2017-08-22 Richard Biener <rguenther@suse.de>
14014 * common.opt (feliminate-dwarf2-dups): Ignore.
14015 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14016 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14017 same_die_p_wrap, compute_section_prefix,
14018 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14019 (comdat_symbol_id, comdat_symbol_number): Likewise.
14020 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14022 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14023 (output_die): Mark unreachable path unreachable.
14024 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14025 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14026 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14027 (dwarf2out_early_finish): Likewise.
14029 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14031 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14033 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14036 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14037 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14038 (avr_attribute_table) <io, io_low, address>: Initialize
14039 .decl_required with true.
14041 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14043 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14044 undocumented debugging options.
14045 (-mvsx-scalar-double): Likewise.
14046 (-mallow-df-permute): Likewise.
14047 (-mvectorize-builtins): Likewise.
14048 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14049 (rs6000_builtin_vectorized_function): Likewise.
14050 (rs6000_builtin_md_vectorized_function): Likewise.
14051 (rs6000_opt_vars): Likewise.
14053 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14056 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14057 (*btrq_imm): Rename from *btrq.
14058 (*btcq_imm): Rename from *btcq.
14059 (btsc): New code attribute.
14060 (*<btsc><mode>): New insn pattern.
14061 (*btr<mode>): Ditto.
14062 (*<btsc><mode>_mask): New insn_and_split pattern.
14063 (*btr<mode>_mask): Ditto.
14065 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14066 Alan Hayward <alan.hayward@arm.com>
14067 David Sherwood <david.sherwood@arm.com>
14069 * function.c (pad_below): Simplify padding calculation.
14071 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14072 Alan Hayward <alan.hayward@arm.com>
14073 David Sherwood <david.sherwood@arm.com>
14075 * target.def (function_prologue): Remove frame size argument.
14076 (function_epilogue): Likewise.
14077 * doc/tm.texi: Regenerate.
14078 * final.c (final_start_function): Update call to function_prologue.
14079 (final_end_function): Update call to function_epilogue.
14080 (default_function_pro_epilogue): Remove frame size argument.
14081 * output.h (default_function_pro_epilogue): Likewise.
14082 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14083 (arm_output_function_prologue): Likewise.
14084 * config/frv/frv.c (frv_function_prologue): Likewise.
14085 (frv_function_epilogue): Likewise.
14086 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14087 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14088 (ia64_output_function_epilogue): Likewise.
14089 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14090 (m32r_output_function_epilogue): Likewise.
14091 * config/microblaze/microblaze.c (microblaze_function_prologue)
14092 (microblaze_function_epilogue): Likewise.
14093 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14094 (mips_output_function_epilogue): Likewise.
14095 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14096 (mmix_target_asm_function_epilogue): Likewise.
14097 * config/msp430/msp430.c (msp430_start_function): Likewise.
14098 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14099 (nds32_asm_function_epilogue): Likewise.
14100 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14101 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14102 (pa_output_function_epilogue): Likewise.
14103 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14104 (rs6000_output_function_epilogue): Likewise.
14105 * config/rl78/rl78.c (rl78_start_function): Likewise.
14106 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14107 (rs6000_output_function_epilogue): Likewise.
14108 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14109 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14110 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14111 (sparc_asm_function_epilogue): Likewise.
14113 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14115 * tree.h (type_has_mode_precision_p): New function.
14116 * convert.c (convert_to_integer_1): Use it.
14117 * expr.c (expand_expr_real_2): Likewise.
14118 (expand_expr_real_1): Likewise.
14119 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14120 * match.pd: Likewise.
14121 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14122 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14123 * tree-tailcall.c (process_assignment): Likewise.
14124 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14125 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14126 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14127 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14128 (vectorizable_assignment): Likewise.
14129 (vectorizable_shift): Likewise.
14130 (vectorizable_operation): Likewise.
14131 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14133 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14135 * match.pd: Add pow (C, x) simplification.
14137 2017-08-21 Richard Biener <rguenther@suse.de>
14139 PR tree-optimization/81900
14140 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14141 for blocks with abnormal predecessors.
14142 (compute_antic): Do not set visited flag prematurely.
14144 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14147 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14149 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14151 * stor-layout.h (vector_type_mode): Move to...
14152 * tree.h (vector_type_mode): ...here.
14153 * stor-layout.c (vector_type_mode): Move to...
14154 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14156 2017-08-21 Richard Biener <rguenther@suse.de>
14158 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14159 register_external_die hooks.
14160 (debug_false_tree_charstarstar_uhwistar): Declare.
14161 (debug_nothing_tree_charstar_uhwi): Likewise.
14162 * debug.c (do_nothing_debug_hooks): Adjust.
14163 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14164 (debug_nothing_tree_charstar_uhwi): Likewise.
14165 * dbxout.c (dbx_debug_hooks): Adjust.
14166 (xcoff_debug_hooks): Likewise.
14167 * sdbout.c (sdb_debug_hooks): Likewise.
14168 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14169 * dwarf2out.c (macinfo_label_base): New global.
14170 (dwarf2out_register_external_die): New function for the
14171 register_external_die hook.
14172 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14173 (dwarf2_debug_hooks): Use them.
14174 (dwarf2_lineno_debug_hooks): Adjust.
14175 (struct die_struct): Add with_offset flag.
14176 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14177 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14178 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14179 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14180 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14181 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14182 defining section names for the early LTO debug variants.
14183 (reset_indirect_string): New helper.
14184 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14185 (print_dw_val): Add support for offsetted symbol references.
14186 (get_ultimate_context): Split out from is_cxx.
14187 (is_cxx): Use get_ultimate_context.
14188 (is_fortran): Add decl overload.
14189 (compute_comp_unit_symbol): Split out worker from
14190 compute_section_prefix.
14191 (compute_section_prefix): Call compute_comp_unit_symbol and
14192 set comdat_type_p here.
14193 (output_die): Skip DIE symbol output for the LTO added one.
14194 Handle DIE symbol references with offset.
14195 (output_comp_unit): Guard section name mangling properly.
14196 For LTO debug sections emit a symbol at the section beginning
14197 which we use to refer to its DIEs.
14198 (add_abstract_origin_attribute): For DIEs registered via
14199 dwarf2out_register_external_die directly refer to the early
14200 DIE rather than indirectly through the shadow one we created.
14201 Remove obsolete call to dwarf2out_abstract_function for
14202 non-function/block origins.
14203 (gen_array_type_die): When generating early LTO debug do
14204 not emit DW_AT_string_length.
14205 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14206 late when in LTO. As suggested place a gcc_unreachable for
14207 the DECL_ABSTRACT_P case.
14208 (gen_subprogram_die): Avoid another specification DIE
14209 for early built declarations/definitions for the late LTO case.
14210 (gen_variable_die): Add type references for late duplicated VLA dies
14212 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14213 we have the abstract instance already.
14214 (process_scope_var): Adjust decl DIE contexts in LTO which
14215 first puts them in limbo.
14216 (gen_decl_die): Do not generate type DIEs late apart from
14217 types for VLAs or for decls we do not yet have a DIE. Do not
14218 call dwarf2out_abstract_function late.
14219 (dwarf2out_early_global_decl): Make sure to create DIEs
14220 for abstract instances of a decl first.
14221 (dwarf2out_late_global_decl): Adjust comment.
14222 (output_macinfo_op): With multiple macro sections use
14223 macinfo_label_base to distinguish labels.
14224 (output_macinfo): Likewise. Update macinfo_label_base.
14225 Pass in the line info label.
14226 (note_variable_value_in_expr): When generating LTO resolve
14227 all variable values here by generating DIEs as needed.
14228 (init_sections_and_labels): Add early LTO debug flag parameter
14229 and generate different sections and names if set. Add generation
14230 counter for the labels so we can have multiple of them.
14231 (reset_dies): Helper to allow DIEs to be output multiple times.
14232 (dwarf2out_finish): When outputting DIEs to the fat part of an
14233 LTO object first reset DIEs.
14234 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14235 (modified_type_die): Check for decl_ultimate_origin being self
14237 (gen_type_die_with_usage): Likewise.
14238 (gen_typedef_die): Allow decl_ultimate_origin being self.
14239 (set_decl_abstract_flags): Remove.
14240 (set_block_abstract_flags): Likewise.
14241 (dwarf2out_abstract_function): Treat the early generated DIEs
14242 as the abstract copy and only add DW_AT_inline and
14243 DW_AT_artificial here and call set_decl_origin_self.
14244 If the DIE has an abstract origin don't do anything.
14245 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14246 if we have none yet (Go fails to build one, PR78628).
14247 (variably_modified_type_p): Prevent endless recursion for Ada
14248 cyclic pointer types.
14249 * lto-streamer-in.c: Include debug.h.
14250 (dref_queue): New global.
14251 (lto_read_tree_1): Stream in DIE references.
14252 (lto_input_tree): Register DIE references.
14253 (input_function): Stream DECL_DEBUG_ARGS.
14254 * lto-streamer-out.c: Include debug.h.
14255 (lto_write_tree_1): Output DIE references.
14256 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14257 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14258 (output_function): Stream DECL_DEBUG_ARGS.
14259 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14260 Stream DECL_ABSTRACT_ORIGIN.
14261 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14262 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14263 DECL_CONTEXT for file-scope decls.
14264 * lto-streamer.h (struct dref_entry): Declare.
14265 (dref_queue): Likewise.
14266 * cfgexpand.c (pass_expand::execute): Do not call the
14267 outlining_inline_function hook here.
14268 * lto-wrapper.c (debug_obj): New global.
14269 (tool_cleanup): Unlink it if required.
14270 (debug_objcopy): New function.
14271 (run_gcc): Handle early debug sections in the IL files by
14272 extracting them to separate files, partially linkin them and
14273 feeding the result back as result to the linker.
14274 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14275 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14276 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14277 sections into a separate segment.
14278 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14280 (darwin_asm_dwarf_section): Likewise.
14281 (darwin_asm_output_dwarf_offset): Likewise.
14282 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14284 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14285 Alan Hayward <alan.hayward@arm.com>
14286 David Sherwood <david.sherwood@arm.com>
14288 * read-md.h (md_reader::record_potential_iterator_use): Replace
14289 pointer argument with an rtx and an index.
14290 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14291 (apply_mode_iterator): Likewise.
14292 (apply_code_iterator): Likewise.
14293 (apply_int_iterator): Likewise.
14294 (apply_subst_iterator): Likewise.
14295 (record_iterator_use): Likewise.
14296 (record_attribute_use): Likewise.
14297 (md_reader::record_potential_iterator_use): Likewise. Update calls
14298 to record_iterator_use and apply_iterator.
14299 (iterator_use): Replace ptr with x and index.
14300 (attribute_use): Likewise.
14301 (apply_attribute_uses): Update calls to apply_iterator.
14302 (apply_iterators): Likewise. Update initialization of iterator_use.
14303 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14304 and record_potential_iterator_use.
14305 (rtx_reader::read_rtx_operand): Likewise.
14307 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14308 Alan Hayward <alan.hayward@arm.com>
14309 David Sherwood <david.sherwood@arm.com>
14311 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14314 2017-08-21 Richard Biener <rguenther@suse.de>
14316 PR middle-end/81884
14317 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14318 at struct end conservatively when comparing common bases.
14320 2017-08-21 Richard Biener <rguenther@suse.de>
14322 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14323 (mem_ref_in_stmt): Remove.
14324 (determine_max_movement): Use ref index to get at the reference.
14325 (invariantness_dom_walker::before_dom_children): Deal with
14326 lim data already initialized.
14327 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14329 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14331 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14332 (TARGET_ISA_ROUND): Ditto.
14333 (TARGET_ROUND): Ditto.
14334 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14335 * config/i386/i386.md: Ditto.
14336 * config/i386/sse.md: Ditto.
14337 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14338 with OPTION_MASK_ISA_SSE4_1.
14340 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14343 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14344 __builtin_ia32_lzcnt_u16.
14346 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14349 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14350 (rs6000_set_current_function): Rewrite function to use it.
14352 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14355 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14356 and TYPE_WARN_IF_NOT_ALIGN.
14357 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14358 (handle_warn_if_not_align): New.
14359 (place_union_field): Call handle_warn_if_not_align.
14360 (place_field): Call handle_warn_if_not_align.
14361 Copy TYPE_WARN_IF_NOT_ALIGN.
14362 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14363 (layout_type): Likewise.
14364 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14366 (tree_decl_common): Add warn_if_not_align.
14367 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14368 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14369 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14370 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14371 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14372 * doc/extend.texi: Document warn_if_not_aligned attribute.
14373 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14375 2017-08-17 Martin Liska <mliska@suse.cz>
14378 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14379 (get_data_dependence): Use it as pointer type.
14380 (distribute_loop): Likewise.
14382 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14384 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14385 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14386 (p8_vmrgow_<mode>_direct): New define_insn.
14387 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14388 handle endianness for vmrgew and vmrgow permute patterns.
14390 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14392 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14393 * config/rs6000/rs6000-cpus.def: Remove comment.
14394 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14395 (POWERPC_MASKS): Likewise.
14396 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14397 use of TARGET_VSX_TIMODE.
14398 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14399 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14400 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14401 (rs6000_option_override_internal): Remove dead code.
14402 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14403 (rs6000_legitimize_reload_address): Likewise.
14404 (rs6000_legitimate_address_p): Likewise.
14405 (rs6000_opt_masks): Delete "vsx-timode".
14406 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14407 from function comment.
14408 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14409 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14410 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14412 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14413 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14414 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14415 (VSX_TI): Likewise.
14417 (define_peephole2): Likewise.
14419 2017-08-17 Martin Sebor <msebor@redhat.com>
14422 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14423 past the end of an array.
14424 (test_pp_format): Add test cases.
14426 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14428 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14429 missing ECF_NOTHROW flags.
14431 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14434 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14435 operands residing in integer registers.
14436 (*vsx_le_perm_load_<mode>): Likewise.
14437 (*vsx_le_perm_store_<mode>): Likewise.
14438 (define_peephole2): Add peepholes to optimize the above.
14440 2017-08-17 Marek Polacek <polacek@redhat.com>
14442 PR middle-end/81814
14443 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14444 to mimic what shorten_compare did. Change the return type to bool.
14445 (fold_cond_expr_with_comparison): Update call to
14446 operand_equal_for_comparison_p.
14447 (fold_ternary_loc): Likewise.
14449 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14451 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14453 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14455 2017-08-17 Richard Biener <rguenther@suse.de>
14457 * tree-ssa-structalias.c (solve_graph): When propagating
14458 to successors update the graphs succ edges and avoid duplicate work.
14460 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14463 * config/i386/i386.c (ix86_option_override_internal): Save target
14464 specific options after ix86_stack_protector_guard_reg was changed.
14466 2017-08-17 Richard Biener <rguenther@suse.de>
14468 PR tree-optimization/81827
14469 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14470 (new_var_info): Initialize it conservatively.
14471 (get_call_vi): Mark register vars.
14472 (new_scalar_tmp_constraint_exp): Likewise.
14473 (handle_rhs_call): Likewise.
14474 (handle_const_call): Likewise.
14475 (create_function_info_for): Likewise.
14476 (solve_constraints): Sort varinfos to separate register from
14477 non-register vars to pack points-to solution bitmaps during
14480 2017-08-17 Marek Polacek <polacek@redhat.com>
14482 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14484 2017-08-17 Richard Biener <rguenther@suse.de>
14486 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14487 to true when overflow is undefined and we saturated the result.
14489 2017-08-17 Alan Modra <amodra@gmail.com>
14492 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14493 Don't use store multiple if only one reg needs saving.
14494 (interesting_frame_related_regno): New function.
14495 (rs6000_frame_related): Don't emit frame info for regs that
14497 (rs6000_emit_epilogue): Likewise.
14499 2017-08-16 Nathan Sidwell <nathan@acm.org>
14501 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14502 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14503 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14504 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14505 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14506 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14507 (verify_type): Adjust for TYPE_BINFO move.
14508 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14509 process TYPE_BINFO directly.
14510 (hash_tree): Likewise.
14511 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14513 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14516 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14518 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14520 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14523 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14524 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14525 Add ix86_binary_operator_ok to insn constraint.
14526 (*iordi_1_bts): Ditto.
14527 (*xordi_1_btc): Ditto.
14528 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14529 Update corresponding peephole2 pattern.
14533 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14535 PR tree-optimization/81832
14536 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14537 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14539 2017-08-16 Marek Polacek <polacek@redhat.com>
14542 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14543 regarding size_zero_node.
14545 2017-08-16 Martin Liska <mliska@suse.cz>
14548 * config.gcc: Respect previously set extra_objs in case
14551 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14553 PR tree-optimization/81835
14554 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14555 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14556 not depend on the phi.
14558 2017-08-16 Alan Modra <amodra@gmail.com>
14560 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14563 2017-08-16 Alan Modra <amodra@gmail.com>
14565 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14566 (save_reg_p): ..into this. Update all callers.
14567 (first_reg_to_save): Simplify.
14569 2017-08-16 Alan Modra <amodra@gmail.com>
14571 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14574 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14578 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14579 constant count if that count is less than 32.
14581 2017-08-15 Nathan Sidwell <nathan@acm.org>
14583 * gcc.c (execute): Emit friendlier message if inferior is killed
14584 by an external cause.
14586 2017-08-15 Richard Biener <rguenther@suse.de>
14588 PR tree-optimization/81790
14589 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14590 CONSTRUCTORs from simplifying and VN.
14592 2017-08-14 Martin Sebor <msebor@redhat.com>
14594 * builtin-attrs.def: Add comments.
14596 2017-08-14 Martin Sebor <msebor@redhat.com>
14599 * doc/extend.texi (attribute nonstring): Document new attribute.
14601 2017-08-14 Martin Sebor <msebor@redhat.com>
14604 * tree-diagnostic.c (default_tree_printer): Handle %G.
14605 * gimple-pretty-print.h (percent_G_format): Declare new function.
14606 * gimple-pretty-print.c (percent_G_format): Define.
14607 * tree-pretty-print.c (percent_K_format): Add argument.
14609 2017-08-14 Martin Sebor <msebor@redhat.com>
14611 PR translation/79998
14612 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14613 Remove a stray space.
14615 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14618 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14619 (*iordi_1_bts): Ditto.
14620 (*xordi_1_btc): Ditto.
14622 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14625 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14626 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14628 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14629 quoted strings, and make more translator-friendly.
14630 (darwin_rs6000_override_options): Likewise.
14631 (rs6000_option_override_internal): Likewise.
14632 (rs6000_return_in_memory): Fix overlong line.
14633 (init_cmulative_args): Use quoted strings, and make more
14634 translator-friendly.
14635 (rs6000_pass_by_reference): Fix overlong line.
14636 (def_builtin): Use quoted strings.
14637 (altivec_expand_predicate_builtin): Use quoted strings, and make
14638 more translator-friendly.
14639 (htm_expand_builtin): Use quoted strings.
14640 (cpu_expand_builtin): Use quoted strings, and make more
14641 translator-friendly.
14642 (altivec_expand_builtin): Likewise.
14643 (paired_expand_predicate_builtin): Likewise.
14644 (rs6000_invalid_builtin): Likewise.
14645 (builtin_function_type): Use quoted strings.
14646 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14647 more translator-friendly.
14648 (rs6000_trampoline_init): Likewise.
14649 (rs6000_handle_altivec_attribute): Likewise.
14650 (rs6000_inner_target_options): Use quoted strings.
14651 (rs6000_disable_incompatible_switches): Likewise.
14652 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14653 strings, and make more translator-friendly.
14654 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14656 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14658 PR tree-optimization/81799
14659 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14660 cond_expr to simple gimple operand.
14662 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14664 PR middle-end/46932
14665 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14667 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14671 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14672 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14673 TARGET_GASISR_PROLOGUES.
14674 * config/avr/avr.c (avr_option_override): Same.
14675 (avr_pass_pre_proep::execute): Same.
14677 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14680 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14681 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14683 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14685 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14686 define_insn_and_split. Split to a memory load from 0 in
14687 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14688 using PTR mode iterator.
14689 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14690 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14691 (*add_tp_<mode>): Redefine as define_insn_and_split.
14692 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14693 address space. Merge with *add_tp_x32 using PTR mode iterator.
14694 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14695 Split to an add with a memory load from 0 in
14696 DEFAULT_TLS_SEG_REG address space.
14698 2017-08-12 Andrew Pinski <apinski@cavium.com>
14700 * config/aarch64/aarch64-option-extensions.def (rdma):
14701 Fix feature string to what Linux prints out in /proc/cpuinfo.
14703 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14706 * dwarf2out.c (modified_type_die): For C typedef types that have
14707 an ultimate origin, process the ultimate origin instead of the
14709 (gen_typedef_die): Assert that input DECLs have no ultimate
14711 (gen_type_die_with_usage): For typedef variants that have an
14712 ultimate origin, just call gen_decl_die on the original DECL.
14713 (process_scope_var): Avoid creating DIEs for local typedefs and
14714 concrete static variables.
14716 2017-08-12 Alan Modra <amodra@gmail.com>
14720 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14721 match gnu-user.h startfile.
14722 (ENDFILE_LINUX_SPEC): Similarly.
14724 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14727 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14729 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14731 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14732 * config/aarch64/aarch64.md (mov<mode>): Change.
14733 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14734 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14735 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14737 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14739 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14740 for storage order barriers.
14742 2017-08-11 Martin Liska <mliska@suse.cz>
14745 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14746 variables of void type.
14748 2017-08-11 Martin Liska <mliska@suse.cz>
14750 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14751 TARGET_SUPPORTS_ALIASES.
14752 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14753 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14754 (optimize_weakref): Likewise.
14755 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14756 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14757 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14759 2017-08-11 Martin Liska <mliska@suse.cz>
14762 * config/i386/i386.c (make_resolver_func): Do complete
14763 refactoring of the function.
14765 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14768 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14769 * config/i386/i386.c (ix86_stack_protect_guard): Use
14770 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14771 * doc/invoke.texi (x86 Options): Document
14772 -mstack-protector-guard-symbol= option.
14774 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14776 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14777 * config/i386/i386.c (ix86_split_stack_guard): New function.
14778 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14779 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14780 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14781 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14782 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14783 (split_stack_space_check): Call ix86_split_stack_guard.
14785 2017-08-10 Martin Sebor <msebor@redhat.com>
14787 * print-tree.c (print_node): Print location using the established
14789 Replace spaces with colons.
14790 (debug_raw, debug): Ditto.
14792 2017-08-10 Martin Sebor <msebor@redhat.com>
14795 * pretty-print.c (pp_format): Correct the handling of %s precision.
14797 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14800 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14802 (ix86_finalize_stack_frame_flags): This. Also clear
14803 frame_pointer_needed if -fno-omit-frame-pointer is used without
14805 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14806 with ix86_finalize_stack_frame_flags.
14807 (ix86_expand_epilogue): Likewise.
14808 (ix86_expand_split_stack_prologue): Likewise.
14809 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14811 2017-08-10 Martin Liska <mliska@suse.cz>
14814 * c-attribs.c (handle_target_attribute):
14815 Report warning for an empty string argument of target attribute.
14817 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14820 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14822 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14823 or DECL_NONLOCAL labels.
14824 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14825 or DECL_NONLOCAL labels here.
14827 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14829 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14830 to indicate when early gimple folding has been disabled.
14831 (rs6000_gimple_fold_builtin): Add debug content.
14832 (rs6000_invalid_builtin): Fix whitespace.
14833 (rs6000_expand_builtin): Fix whitespace.
14834 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14836 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14839 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14840 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14842 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14844 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14845 (qdf24xx): Likewise.
14846 * config/aarch64/aarch64.md: Include falkor.md.
14847 * config/aarch64/falkor.md: New.
14849 2017-08-09 Marek Polacek <polacek@redhat.com>
14852 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14853 * diagnostic.c (emit_diagnostic): Add a comment.
14854 (emit_diagnostic_valist): New function.
14856 2017-08-09 Marek Polacek <polacek@redhat.com>
14859 * input.c (make_location): New overload.
14860 * input.h (make_location): Declare.
14862 2017-08-08 Alan Modra <amodra@gmail.com>
14863 H.J. Lu <hongjiu.lu@intel.com>
14866 * gcc.c (NO_PIE_SPEC): Delete.
14867 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14869 (LINK_PIE_SPEC): ..to here.
14870 (LINK_COMMAND_SPEC): Support -no-pie.
14871 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14872 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14874 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14875 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14876 (ENDFILE_CRTEND_SPEC): Similarly.
14878 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14881 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14882 (mstack-protector-guard-offset=): Ditto.
14883 * config/i386/i386.c (ix86_option_override): Handle
14884 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14886 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14887 ix86_stack_protect_guard_offset variables.
14888 (TARGET_STACK_PROTECT_GUARD): Always define.
14889 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14890 and -mstack-protector-guard-offset= options.
14892 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14894 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14895 boundary case for the last candidate.
14897 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14899 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14900 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14902 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14904 PR middle-end/19706
14905 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14906 * config/aarch64/aarch64-builtins.c
14907 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14908 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14909 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14911 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14912 Andrew Pinski <pinskia@gmail.com>
14914 PR middle-end/19706
14915 * internal-fn.def (XORSIGN): New.
14916 * optabs.def (xorsign_optab): New.
14917 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14918 (convert_expand_mult_copysign): New.
14919 (pass_optimize_widening_mul::execute): Call
14920 convert_expand_mult_copysign.
14922 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14924 PR tree-optimization/81354
14925 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14926 Insert on edges rather than explicitly creating landing pads.
14927 (analyze_candidates_and_replace): Commit edge inserts.
14929 2017-08-08 Richard Biener <rguenther@suse.de>
14931 PR middle-end/81719
14932 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14933 (expand_simple_operations): Also look through ADDR_EXPRs with
14934 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14936 2017-08-08 Richard Biener <rguenther@suse.de>
14938 PR tree-optimization/81723
14939 * tree-vect-slp.c (struct bst_traits): New hash traits.
14940 (bst_fail): New global.
14941 (vect_build_slp_tree_2): New worker, split out from ...
14942 (vect_build_slp_tree): ... this now wrapping it with using
14943 bst_fail set to cache SLP tree build fails. Properly handle
14945 (vect_analyze_slp_instance): Allocate and free bst_fail.
14947 2017-08-08 Martin Liska <mliska@suse.cz>
14950 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14951 LABEL_DECLs that can be from a different function.
14953 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14955 PR tree-optimization/81744
14956 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14957 loop's number of iterations.
14959 2017-08-08 Martin Liska <mliska@suse.cz>
14961 * asan.c: Include header files.
14962 * attribs.c (build_decl_attribute_variant): New function moved
14964 (build_type_attribute_qual_variant): Likewise.
14965 (cmp_attrib_identifiers): Likewise.
14966 (simple_cst_list_equal): Likewise.
14967 (omp_declare_simd_clauses_equal): Likewise.
14968 (attribute_value_equal): Likewise.
14969 (comp_type_attributes): Likewise.
14970 (build_type_attribute_variant): Likewise.
14971 (lookup_ident_attribute): Likewise.
14972 (remove_attribute): Likewise.
14973 (merge_attributes): Likewise.
14974 (merge_type_attributes): Likewise.
14975 (merge_decl_attributes): Likewise.
14976 (merge_dllimport_decl_attributes): Likewise.
14977 (handle_dll_attribute): Likewise.
14978 (attribute_list_equal): Likewise.
14979 (attribute_list_contained): Likewise.
14980 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14981 (lookup_attribute_by_prefix): Likewise.
14982 * bb-reorder.c: Include header files.
14983 * builtins.c: Likewise.
14984 * calls.c: Likewise.
14985 * cfgexpand.c: Likewise.
14986 * cgraph.c: Likewise.
14987 * cgraphunit.c: Likewise.
14988 * convert.c: Likewise.
14989 * dwarf2out.c: Likewise.
14990 * final.c: Likewise.
14991 * fold-const.c: Likewise.
14992 * function.c: Likewise.
14993 * gimple-expr.c: Likewise.
14994 * gimple-fold.c: Likewise.
14995 * gimple-pretty-print.c: Likewise.
14996 * gimple.c: Likewise.
14997 * gimplify.c: Likewise.
14998 * hsa-common.c: Likewise.
14999 * hsa-gen.c: Likewise.
15000 * internal-fn.c: Likewise.
15001 * ipa-chkp.c: Likewise.
15002 * ipa-cp.c: Likewise.
15003 * ipa-devirt.c: Likewise.
15004 * ipa-fnsummary.c: Likewise.
15005 * ipa-inline.c: Likewise.
15006 * ipa-visibility.c: Likewise.
15008 * lto-cgraph.c: Likewise.
15009 * omp-expand.c: Likewise.
15010 * omp-general.c: Likewise.
15011 * omp-low.c: Likewise.
15012 * omp-offload.c: Likewise.
15013 * omp-simd-clone.c: Likewise.
15014 * opts-global.c: Likewise.
15015 * passes.c: Likewise.
15016 * predict.c: Likewise.
15017 * sancov.c: Likewise.
15018 * sanopt.c: Likewise.
15019 * symtab.c: Likewise.
15020 * toplev.c: Likewise.
15021 * trans-mem.c: Likewise.
15022 * tree-chkp.c: Likewise.
15023 * tree-eh.c: Likewise.
15024 * tree-into-ssa.c: Likewise.
15025 * tree-object-size.c: Likewise.
15026 * tree-parloops.c: Likewise.
15027 * tree-profile.c: Likewise.
15028 * tree-ssa-ccp.c: Likewise.
15029 * tree-ssa-live.c: Likewise.
15030 * tree-ssa-loop.c: Likewise.
15031 * tree-ssa-sccvn.c: Likewise.
15032 * tree-ssa-structalias.c: Likewise.
15033 * tree-ssa.c: Likewise.
15034 * tree-streamer-in.c: Likewise.
15035 * tree-vectorizer.c: Likewise.
15036 * tree-vrp.c: Likewise.
15037 * tsan.c: Likewise.
15038 * ubsan.c: Likewise.
15039 * varasm.c: Likewise.
15040 * varpool.c: Likewise.
15041 * tree.c: Remove functions moved to attribs.[ch].
15042 * tree.h: Likewise.
15043 * config/aarch64/aarch64.c: Add attrs.h header file.
15044 * config/alpha/alpha.c: Likewise.
15045 * config/arc/arc.c: Likewise.
15046 * config/arm/arm.c: Likewise.
15047 * config/avr/avr.c: Likewise.
15048 * config/bfin/bfin.c: Likewise.
15049 * config/c6x/c6x.c: Likewise.
15050 * config/cr16/cr16.c: Likewise.
15051 * config/cris/cris.c: Likewise.
15052 * config/darwin.c: Likewise.
15053 * config/epiphany/epiphany.c: Likewise.
15054 * config/fr30/fr30.c: Likewise.
15055 * config/frv/frv.c: Likewise.
15056 * config/ft32/ft32.c: Likewise.
15057 * config/h8300/h8300.c: Likewise.
15058 * config/i386/winnt.c: Likewise.
15059 * config/ia64/ia64.c: Likewise.
15060 * config/iq2000/iq2000.c: Likewise.
15061 * config/lm32/lm32.c: Likewise.
15062 * config/m32c/m32c.c: Likewise.
15063 * config/m32r/m32r.c: Likewise.
15064 * config/m68k/m68k.c: Likewise.
15065 * config/mcore/mcore.c: Likewise.
15066 * config/microblaze/microblaze.c: Likewise.
15067 * config/mips/mips.c: Likewise.
15068 * config/mmix/mmix.c: Likewise.
15069 * config/mn10300/mn10300.c: Likewise.
15070 * config/moxie/moxie.c: Likewise.
15071 * config/msp430/msp430.c: Likewise.
15072 * config/nds32/nds32-isr.c: Likewise.
15073 * config/nds32/nds32.c: Likewise.
15074 * config/nios2/nios2.c: Likewise.
15075 * config/nvptx/nvptx.c: Likewise.
15076 * config/pa/pa.c: Likewise.
15077 * config/pdp11/pdp11.c: Likewise.
15078 * config/powerpcspe/powerpcspe.c: Likewise.
15079 * config/riscv/riscv.c: Likewise.
15080 * config/rl78/rl78.c: Likewise.
15081 * config/rx/rx.c: Likewise.
15082 * config/s390/s390.c: Likewise.
15083 * config/sh/sh.c: Likewise.
15084 * config/sol2.c: Likewise.
15085 * config/sparc/sparc.c: Likewise.
15086 * config/spu/spu.c: Likewise.
15087 * config/stormy16/stormy16.c: Likewise.
15088 * config/tilegx/tilegx.c: Likewise.
15089 * config/tilepro/tilepro.c: Likewise.
15090 * config/v850/v850.c: Likewise.
15091 * config/vax/vax.c: Likewise.
15092 * config/visium/visium.c: Likewise.
15093 * config/xtensa/xtensa.c: Likewise.
15095 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15098 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15099 constraints since the -mupper-regs-* switches have been
15101 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15102 into a vector from a double word element that was extracted from
15103 another vector, and eliminate extra XXPERMDI instructions.
15104 (vsx_concat_<mode>_2): Likewise.
15105 (vsx_concat_<mode>_3): Likewise.
15106 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15107 concat to allow optimizing inserts from previous extracts.
15109 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15111 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15112 memory reference to a SSP offset in TLS address space.
15113 (ix86_print_operand) <case '@'>: Remove.
15114 (ix86_print_operand_punct_valid_p): Remove '@' code.
15115 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15116 UNSPEC_SP_TLS_TEST.
15117 (stack_tls_protect_set_<mode>): Remove.
15118 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15119 (stack_tls_protect_test_<mode>): Remove.
15120 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15122 2017-08-07 Olivier Hainque <hainque@adacore.com>
15125 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15127 2017-08-07 Douglas Rupp <rupp@adacore.com>
15129 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15130 variable was referenced as multidir in command.
15132 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15135 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15138 2017-08-07 Martin Liska <mliska@suse.cz>
15140 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15141 * config/rl78/rl78.c: Add include of attribs.h.
15142 * config/sh/sh.c: Likewise.
15143 * config/v850/v850.c: Likewise.
15145 2017-08-07 Tom de Vries <tom@codesourcery.com>
15147 PR middle-end/78266
15148 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15150 2017-08-07 Martin Liska <mliska@suse.cz>
15152 * config/mips/mips.c: Include attribs.h.
15154 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15157 * doc/invoke.texi: Document change in behvaior for -Ofast for
15160 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15162 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15164 (aarch64_pop_regs): Likewise.
15165 (aarch64_gen_load_pair): Likewise.
15166 (aarch64_save_callee_saves): Likewise.
15167 (aarch64_restore_callee_saves): Likewise.
15169 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15171 * config/i386/i386.c: Revert the last change.
15173 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15176 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15178 (ix86_finalize_stack_frame_flags): This. Also clear
15179 frame_pointer_needed if -fno-omit-frame-pointer is used without
15181 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15182 with ix86_finalize_stack_frame_flags.
15183 (ix86_expand_epilogue): Likewise.
15184 (ix86_expand_split_stack_prologue): Likewise.
15186 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15189 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15190 to P_AES for Westmere.
15192 2017-08-07 Jonathan Yong <10walls@gmail.com>
15194 * config/i386/mingw.opt (fset-stack-executable): Removed.
15195 * config/i386/cygming.opt (fset-stack-executable): Moved
15197 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15199 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15201 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15203 2017-08-07 Marek Polacek <polacek@redhat.com>
15205 PR middle-end/81737
15206 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15208 2017-08-07 Martin Liska <mliska@suse.cz>
15210 * attribs.h (canonicalize_attr_name): New function.
15211 (cmp_attribs): Move from c-format.c and adjusted.
15212 (is_attribute_p): Moved from tree.h.
15213 * tree-inline.c: Add new includes.
15214 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15215 (private_is_attribute_p): Remove.
15216 (private_lookup_attribute): Likewise.
15217 (private_lookup_attribute_by_prefix): Simplify.
15218 (remove_attribute): Use is_attribute_p.
15219 * tree.h: Remove removed declarations.
15221 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15223 PR middle-end/81698
15224 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15225 instead of computing it in the function. Formatting fix.
15226 (expand_case): Don't rely on default_edge being the first edge,
15227 clear it if removing it, pass default_edge to
15228 emit_case_dispatch_table.
15229 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15232 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15234 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15235 insn in the function, emit NOP after the insn.
15237 2017-08-06 Tom de Vries <tom@codesourcery.com>
15239 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15242 2017-08-06 Tom de Vries <tom@codesourcery.com>
15244 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15247 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15249 PR tree-optimization/57371
15250 * match.pd: New pattern.
15252 2017-08-04 Marek Polacek <polacek@redhat.com>
15254 PR middle-end/81695
15255 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15256 perform the computation in offset_int.
15258 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15260 PR tree-optimization/81136
15261 * tree-vectorizer.h: Include tree-hash-traits.h.
15262 (vec_base_alignments): New typedef.
15263 (vec_info): Add a base_alignments field.
15264 (vect_record_base_alignments): Declare.
15265 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15267 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15268 (create_data_ref): Add an is_conditional_in_stmt argument.
15269 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15270 the is_conditional_in_stmt field.
15271 (data_ref_loc): Add an is_conditional_in_stmt field.
15272 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15273 (find_data_references_in_stmt): Update call to create_data_ref.
15274 (graphite_find_data_references_in_stmt): Likewise.
15275 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15276 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15277 (vect_record_base_alignment): New function.
15278 (vect_record_base_alignments): Likewise.
15279 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15280 for nested statements even if we fail to compute a misalignment.
15281 Use pooled base alignments for unconditional references.
15282 (vect_find_same_alignment_drs): Compare base addresses instead
15284 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15285 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15287 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15289 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15290 Add an explicit name for the enum. Use auto_vec for slp_instances
15291 and grouped_stores.
15292 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15294 (_bb_vec_info): Add a constructor and destructor.
15295 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15296 (destroy_loop_vec_info): Delete.
15297 (vect_destroy_datarefs): Likewise.
15298 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15299 (vec_info::vec_info): New function.
15300 (vec_info::~vec_info): Likewise.
15301 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15302 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15303 destroy_loop_vec_info.
15304 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15305 (_loop_vec_info::_loop_vec_info): ...this.
15306 (destroy_loop_vec_info): Replace with...
15307 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15308 the stmt_vec_infos. Leave handling of vec_info information to its
15309 destructor. Remove explicit vector releases.
15310 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15311 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15312 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15313 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15314 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15315 (destroy_bb_vec_info): Replace with...
15316 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15317 information to its destructor.
15318 (vect_slp_analyze_bb_1): Use new and delete instead of
15319 new_bb_vec_info and destroy_bb_vec_info.
15320 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15323 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15325 * tree-data-ref.h (subscript): Add access_fn field.
15326 (data_dependence_relation): Add could_be_independent_p.
15327 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15328 (same_access_functions): Move to tree-data-ref.c.
15329 * tree-data-ref.c (ref_contains_union_access_p): New function.
15330 (access_fn_component_p): Likewise.
15331 (access_fn_components_comparable_p): Likewise.
15332 (dr_analyze_indices): Add a reference to access_fn_component_p.
15333 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15335 (constant_access_functions): Likewise.
15336 (add_other_self_distances): Likewise.
15337 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15338 (initialize_data_dependence_relation): Use XCNEW and remove
15339 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15340 of access functions that have the same type. Allow the
15341 subsequence to end with different bases in some circumstances.
15342 Record the chosen access functions in SUB_ACCESS_FN.
15343 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15344 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15345 (subscript_dependence_tester_1): Likewise dra and drb.
15346 (build_classic_dist_vector): Update calls accordingly.
15347 (subscript_dependence_tester): Likewise.
15348 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15349 DDR_COULD_BE_INDEPENDENT_P.
15350 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15351 comp_alias_ddrs instead of may_alias_ddrs.
15352 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15354 (vect_analyze_data_ref_dependence): Use it if
15355 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15356 distance vectors if that fails.
15357 (dependence_distance_ge_vf): New function.
15358 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15359 LOOP_VINFO_MAY_ALIAS_DDRS.
15361 2017-08-04 Richard Biener <rguenther@suse.de>
15363 PR middle-end/81705
15364 * fold-const.c (fold_binary_loc): Properly restrict
15365 minus_var0 && minus_var1 case when associating undefined overflow
15368 2017-08-04 Tom de Vries <tom@codesourcery.com>
15370 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15372 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15374 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15375 Don't start diagnostic messages with a capital letter.
15376 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15378 (rs6000_invalid_builtin): Likewise.
15379 (rs6000_trampoline_init): Likewise.
15381 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15384 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15385 after setting changeable df flags.
15387 2017-08-03 Richard Biener <rguenther@suse.de>
15389 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15390 up if the use is in USE - X.
15392 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15394 * toplev.c (dumpfile.h): New include.
15395 (internal_error_reentered): New static function. Use it...
15396 (internal_error_function): ...here to handle reentered internal_error.
15398 2017-08-03 Richard Biener <rguenther@suse.de>
15400 PR middle-end/81148
15401 * fold-const.c (split_tree): Add minus_var and minus_con
15402 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15403 here but always use minus_*.
15404 (associate_trees): Assert we never associate with MINUS_EXPR
15405 and NULL first operand. Do not recurse for PLUS_EXPR operands
15406 when associating as MINUS_EXPR either.
15407 (fold_binary_loc): Track minus_var and minus_con.
15409 2017-08-03 Tom de Vries <tom@codesourcery.com>
15412 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15413 ACCEL_COMPILER, apply finish_options on
15414 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15416 2017-08-03 Tom de Vries <tom@codesourcery.com>
15419 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15420 function_entry_patch_area_size > 0.
15422 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15425 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15426 instead of 10??LU, perform unit multiplication in wide_int,
15427 don't change alloc_object_size_limit if the limit is larger
15430 PR tree-optimization/81655
15431 PR tree-optimization/81588
15432 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15433 the case when ranges[i].low and high are 1 for unsigned type with
15436 PR middle-end/81052
15437 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15438 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15440 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15442 * tree-vrp.h: Add include guard.
15444 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15447 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15448 (ud2): New insn pattern.
15449 * config/i386/i386.c (ix86_expand_epilogue):
15450 For naked functions, generate ud2 instead of trap insn.
15452 2017-08-02 Marek Polacek <polacek@redhat.com>
15455 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15457 2017-08-02 Tom de Vries <tom@codesourcery.com>
15458 Cesar Philippidis <cesar@codesourcery.com>
15460 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15461 Add missing edge probabilities.
15463 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15465 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15466 Correct endianness.
15468 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15470 PR middle-end/79499
15471 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15472 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15473 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15474 prologue_seq sequences - if any.
15476 2017-08-02 Richard Biener <rguenther@suse.de>
15478 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15479 via vectors if supported, integer extracts via punning if supported
15480 or otherwise vector extracts.
15482 2017-08-02 Richard Biener <rguenther@suse.de>
15484 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15486 (bitmap_insert_into_set): ... this.
15488 2017-08-02 Richard Biener <rguenther@suse.de>
15490 PR tree-optimization/81633
15492 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15494 PR tree-optimization/71752
15495 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15497 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15499 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15500 (machine_function::call_ms2sysv_pad_out): Remove field.
15501 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15502 (ix86_compute_frame_layout): Likewise.
15504 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15507 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15508 attribute with interrupt attribute.
15510 2017-08-01 Andrew Pinski <apinski@cavium.com>
15512 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15513 BIT_INSERT_EXPR's operand 1
15514 to see if the types precision matches.
15516 2017-08-01 Martin Liska <mliska@suse.cz>
15518 PR middle-end/70140
15519 * builtins.c (expand_builtin_memcpy_args): Remove.
15520 (expand_builtin_memcpy): Call newly added function
15521 expand_builtin_memory_copy_args.
15522 (expand_builtin_memcpy_with_bounds): Likewise.
15523 (expand_builtin_mempcpy): Remove last argument.
15524 (expand_builtin_mempcpy_with_bounds): Likewise.
15525 (expand_builtin_memory_copy_args): New function created from
15526 expand_builtin_mempcpy_args with small modifications.
15527 (expand_builtin_mempcpy_args): Remove.
15528 (expand_builtin_stpcpy): Remove unused argument.
15529 (expand_builtin): Likewise.
15530 (expand_builtin_with_bounds): Likewise.
15532 2017-08-01 Martin Liska <mliska@suse.cz>
15535 Make mempcpy more optimal (PR middle-end/70140).
15537 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15540 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15541 __builtin_vec_cmpne verify both arguments are compatible vectors
15542 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15543 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15544 move computation of aligned to after checking the argument types.
15548 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15549 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15552 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15553 Doug Rupp <rupp@adacore.com>
15554 Olivier Hainque <hainque@adacore.com>
15556 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15557 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15559 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15560 for TARGET_OS_CPP_BUILTIN.
15561 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15562 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15564 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15565 passing required abi options to the assembler for EABI configurations.
15566 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15567 of .text.hot and .text.unlikely sections for kernel modules when
15568 using ARM style exceptions.
15569 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15570 options. Add EXTRA_CC1_SPEC.
15571 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15573 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15575 (ARM_TARGET2_DWARF_FORMAT): Define.
15576 * config/arm/t-vxworks: Adjust multilib control to removal of the
15577 Diab command line options.
15579 2017-08-01 Martin Liska <mliska@suse.cz>
15581 PR gcov-profile/81561
15582 * gcov.c (unblock): Make unblocking safe as we need to preserve
15583 index correspondence of blocks and block_lists.
15585 2017-08-01 Richard Biener <rguenther@suse.de>
15587 PR tree-optimization/81181
15588 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15589 (compute_antic): ... end of iteration here.
15591 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15593 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15594 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15595 (ftree-slp-vectorize): Likewise.
15596 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15597 can no longer be set independent of flag_tree_loop_vectorize.
15598 * omp-general.c (emp_max_vf): Likewise.
15599 * opts.c (enable_fdo_optimizations): Remove references to
15600 flag_tree_vectorize, these are now implicit.
15601 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15602 and leave it for the options machinery.
15604 2017-08-01 Martin Liska <mliska@suse.cz>
15606 PR middle-end/70140
15607 * builtins.c (expand_builtin_memcpy_args): Remove.
15608 (expand_builtin_memcpy): Call newly added function
15609 expand_builtin_memory_copy_args.
15610 (expand_builtin_memcpy_with_bounds): Likewise.
15611 (expand_builtin_mempcpy): Remove last argument.
15612 (expand_builtin_mempcpy_with_bounds): Likewise.
15613 (expand_builtin_memory_copy_args): New function created from
15614 expand_builtin_mempcpy_args with small modifications.
15615 (expand_builtin_mempcpy_args): Remove.
15616 (expand_builtin_stpcpy): Remove unused argument.
15617 (expand_builtin): Likewise.
15618 (expand_builtin_with_bounds): Likewise.
15620 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15623 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15624 print "ds:" only for immediates in generic address space.
15626 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15629 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15630 (ix86_function_ok_for_sibcall): Return false for naked functions.
15632 2017-08-01 Richard Biener <rguenther@suse.de>
15634 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15635 (compute_antic): Seed worklist with exit block predecessors.
15636 * cfganal.c (dfs_find_deadend): For a cycle return the source
15637 of the edge closing it.
15639 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15641 * config/aarch64/aarch64.c
15642 (aarch64_can_const_movi_rtx_p): Move 0 check.
15644 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15646 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15647 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15649 * match.pd: Ditto in address comparison pattern.
15651 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15653 PR tree-optimization/81627
15654 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15655 closed ssa form for store-store chain.
15657 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15659 PR tree-optimization/81620
15660 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15661 for store-store chain.
15663 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15665 PR tree-optimization/81588
15666 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15667 ranges[i].in_p, invert comparison code ccode. For >/>=,
15668 swap rhs1 and rhs2 and comparison code unconditionally,
15669 for </<= don't do that. Don't swap rhs1/rhs2 again if
15670 ranges[i].in_p, instead invert comparison code ccode if
15671 opcode or oe->rank is BIT_IOR_EXPR.
15674 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15675 a direct optab to conversion optab.
15676 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15677 with GET_MODE_INNER as last argument instead of optab_handler.
15678 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15679 vector extraction if possible and optab is available.
15680 * expr.c (store_constructor): Use convert_optab_handler instead
15681 of optab_handler. Use vector initialization from smaller
15682 vectors if possible and optab is available.
15683 * tree-vect-stmts.c (vectorizable_load): Likewise.
15684 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15685 now have two modes.
15686 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15687 of vec_init from half-sized vectors with the same element mode.
15688 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15689 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15690 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15691 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15692 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15693 after mode in gen_vec_extract* calls.
15694 (vec_extract<mode>): Renamed to ...
15695 (vec_extract<mode><ssescalarmodelower>): ... this.
15696 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15697 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15698 element mode after mode in gen_vec_init* calls.
15699 (VEC_INIT_HALF_MODE): New mode iterator.
15700 (vec_init<mode>): Renamed to ...
15701 (vec_init<mode><ssescalarmodelower>): ... this.
15702 (vec_init<mode><ssehalfvecmodelower>): New expander.
15703 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15704 (vec_extractv2sfsf): ... this.
15705 (vec_initv2sf): Renamed to ...
15706 (vec_initv2sfsf): ... this.
15707 (vec_extractv2si): Renamed to ...
15708 (vec_extractv2sisi): ... this.
15709 (vec_initv2si): Renamed to ...
15710 (vec_initv2sisi): ... this.
15711 (vec_extractv4hi): Renamed to ...
15712 (vec_extractv4hihi): ... this.
15713 (vec_initv4hi): Renamed to ...
15714 (vec_initv4hihi): ... this.
15715 (vec_extractv8qi): Renamed to ...
15716 (vec_extractv8qiqi): ... this.
15717 (vec_initv8qi): Renamed to ...
15718 (vec_initv8qiqi): ... this.
15719 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15720 (vec_init<mode>): Renamed to ...
15721 (vec_init<mode><VEC_base_l>): ... this.
15722 (vec_extract<mode>): Renamed to ...
15723 (vec_extract<mode><VEC_base_l>): ... this.
15724 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15725 (vec_initv2sfsf): ... this.
15726 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15727 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15728 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15729 element mode after mode in gen_vec_init* calls.
15730 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15731 (vec_init<mode><Vel>): ... this.
15732 (vec_extract<mode>): Renamed to ...
15733 (vec_extract<mode><Vel>): ... this.
15734 * config/aarch64/iterators.md (Vel): New mode attribute.
15735 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15736 Add element mode after mode in gen_vec_extract* calls.
15737 * config/s390/vector.md (non_vec_l): New mode attribute.
15738 (vec_extract<mode>): Renamed to ...
15739 (vec_extract<mode><non_vec_l>): ... this.
15740 (vec_init<mode>): Renamed to ...
15741 (vec_init<mode><non_vec_l>): ... this.
15742 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15743 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15745 * config/arm/iterators.md (V_elem_l): New mode attribute.
15746 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15747 (vec_extract<mode><V_elem_l>): ... this.
15748 (vec_extractv2di): Renamed to ...
15749 (vec_extractv2didi): ... this.
15750 (vec_init<mode>): Renamed to ...
15751 (vec_init<mode><V_elem_l>): ... this.
15752 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15753 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15754 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15755 Add element mode after gen_vec_extract* calls.
15756 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15757 (vec_init<mode><unitmode>): ... this.
15758 (vec_extract<mode>): Renamed to ...
15759 (vec_extract<mode><unitmode>): ... this.
15760 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15761 (vec_init<mode><unitmode>): ... this.
15762 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15763 (vec_initv2sfsf): ... this.
15764 (vec_extractv2sf): Renamed to ...
15765 (vec_extractv2sfsf): ... this.
15766 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15767 Add element mode after gen_vec_extract* calls.
15768 * config/mips/mips.md (unitmode): New mode iterator.
15769 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15770 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15771 * config/spu/spu.md (inner_l): New mode attribute.
15772 (vec_init<mode>): Renamed to ...
15773 (vec_init<mode><inner_l>): ... this.
15774 (vec_extract<mode>): Renamed to ...
15775 (vec_extract<mode><inner_l>): ... this.
15776 * config/sparc/sparc.md (veltmode): New mode iterator.
15777 (vec_init<VMALL:mode>): Renamed to ...
15778 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15779 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15780 (vec_initv2sisi): ... this.
15781 (vec_initv2sf): Renamed to ...
15782 (vec_initv2sfsf): ... this.
15783 (vec_extractv2sf): Renamed to ...
15784 (vec_extractv2sfsf): ... this.
15785 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15786 (vec_init<mode>): Renamed to ...
15787 (vec_init<mode><VEC_base_l>): ... this.
15788 (vec_extract<mode>): Renamed to ...
15789 (vec_extract<mode><VEC_base_l>): ... this.
15790 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15791 (vec_initv2sfsf): ... this.
15792 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15793 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15794 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15795 gen_vec_init* calls.
15797 2017-08-01 Richard Biener <rguenther@suse.de>
15799 PR tree-optimization/81297
15800 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15801 TREE_OVERFLOW from INTEGER_CSTs.
15803 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15805 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15807 2017-07-31 Carl Love <cel@us.ibm.com>
15809 * config/rs6000/rs6000-c: Add support for built-in functions
15810 vector signed char vec_xl_be (signed long long, signed char *);
15811 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15812 vector signed int vec_xl_be (signed long long, signed int *);
15813 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15814 vector signed long long vec_xl_be (signed long long, signed long long *);
15815 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15816 vector signed short vec_xl_be (signed long long, signed short *);
15817 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15818 vector double vec_xl_be (signed long long, double *);
15819 vector float vec_xl_be (signed long long, float *);
15820 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15821 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15822 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15824 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15825 (altivec_expand_builtin): Add switch statement to call
15826 altivec_expand_xl_be for each builtin.
15827 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15828 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15829 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15830 __builtin_vsx_le_be_v16qi.
15831 * doc/extend.texi: Update the built-in documentation file for the
15832 new built-in functions.
15834 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15837 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15839 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15841 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15843 * config.gcc: Add z14.
15844 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15845 CPU model numbers for z13s and z14.
15846 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15848 * config/s390/s390-opts.h (enum processor_type): Rename
15849 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15850 * config/s390/s390.c (processor_table): Add field for CPU name to
15851 be passed to Binutils.
15852 (s390_asm_output_machine_for_arch): Use the new field in
15853 processor_table for Binutils.
15854 (s390_expand_builtin): Replace arch12 with z14.
15855 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15856 (s390_get_sched_attrmask): Likewise.
15857 (s390_get_unit_mask): Likewise.
15858 * config/s390/s390.opt: Add z14 to processor_type enum.
15860 2017-07-31 Martin Jambor <mjambor@suse.cz>
15863 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15864 regardless of optimization level.
15866 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15867 Martin Liska <mliska@suse.cz>
15869 * predict.def: Remove old comment and adjust probability.
15870 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15871 PREDICT statements.
15873 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15876 * config/i386/i386.c (ix86_function_naked): New function.
15877 (ix86_can_use_return_insn_p): Return false for naked functions.
15878 (ix86_expand_prologue): Skip prologue for naked functions.
15879 (ix86_expand_epilogue): Skip epilogue for naked functions
15880 and emit trap instruction.
15881 (ix86_warn_func_return): New function.
15882 (ix86_attribute_table): Add "naked" attribute specification.
15883 (TARGET_WARN_FUNC_RETURN): Define.
15884 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15886 2017-07-31 Martin Liska <mliska@suse.cz>
15888 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15889 (dump_gimple_bb_header): Always dump BB info.
15890 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15892 2017-07-31 Martin Liska <mliska@suse.cz>
15895 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15896 also with current_function_decl non-null equality.
15898 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15901 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15902 change type to the element type, instead add eltype variable and
15903 use it where we are interested in the element type.
15905 PR tree-optimization/81603
15906 * ipa-polymorphic-call.c
15907 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15908 offset arithmetic in offset_int, bail out if the resulting bit offset
15909 doesn't fit into shwi.
15911 2017-07-31 Martin Liska <mliska@suse.cz>
15913 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15914 (gimplify_save_expr): Fix comment.
15916 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15919 * config/i386/i386.c (ix86_function_arg): Update arguments for
15921 (ix86_compute_frame_layout): Set the initial stack offset to
15922 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15923 INCOMING_FRAME_SP_OFFSET.
15924 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15925 stack before exception handler returns.
15926 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15927 the 'ERROR_CODE' for exception handler.
15929 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15931 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15932 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15933 (ASM_OUTPUT_REG_POP): Ditto.
15934 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15935 instead of asm_fprintf to output pure string.
15937 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15939 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15940 to imported_module_or_decl hook.
15941 (debug_nothing_tree_tree_tree_bool): Remove.
15942 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15943 * debug.c (do_nothing_debug_hooks): Use
15944 debug_nothing_tree_tree_tree_bool_bool instead of
15945 debug_nothing_tree_tree_tree_bool.
15946 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15947 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15948 * sdbout.c (sdb_debug_hooks): Likewise.
15949 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15950 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15952 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15953 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15954 attribute, don't add anything.
15956 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15958 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15959 (fold_build2_stat_loc): Likewise.
15960 (fold_build3_stat_loc): Likewise.
15961 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15962 (fold_build1_loc): Remove macro.
15963 (fold_build2_loc): Likewise.
15964 (fold_build3_loc): Likewise.
15966 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15968 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15969 (gimple_build_debug_bind_source_stat): Likewise.
15970 * gimple.h (gimple_build_debug_bind): Remove macro.
15971 (gimple_build_debug_bind_source): Likewise.
15973 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15975 * bitmap.c (bitmap_alloc): Adjust.
15976 (bitmap_gc_alloc): Likewise.
15977 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15979 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15981 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15982 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15983 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15984 (bitmap_gc_alloc_stat): Likewise.
15985 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15987 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15989 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15990 * rtl.h (shallow_copy_rtx): Remove macro.
15992 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15994 * emit-rtl.c (gen_raw_REG): Adjust.
15995 * gengenrtl.c (gendef): Likewise.
15996 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15997 * rtl.h (rtx_alloc): Remove macro.
15999 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16001 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16002 (build_tree_list_stat): Likewise.
16003 * tree.h (build_tree_list): Remove macro.
16004 (build_tree_list_vec): Likewise.
16006 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16008 * tree.c (make_vector_stat): Remove _stat from name.
16009 (build_vector_stat): Likewise.
16010 * tree.h (make_vector_stat): Remove macro.
16011 (build_vector_stat): Likewise.
16013 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16015 * tree.h (build_var_debug_value): Remove prototype.
16017 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16019 * tree.c (tree_cons_stat): Remove _stat from name.
16020 * tree.h (tree_cons): Remove macro.
16022 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16024 * tree.c (build_vl_exp_stat): Remove _stat from name.
16025 * tree.h (build_vl_exp): Remove macro.
16027 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16029 * tree.c (build_decl_stat): Remove _stat from name.
16030 * tree.h (build_decl): Remove macro.
16032 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16034 * gimple.c (gimple_build_with_ops_stat): Adjust.
16035 (gimple_alloc_stat): Remove _stat from name.
16036 * gimple.h (gimple_alloc): Remove macro.
16038 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16040 * tree.c (make_tree_vec_stat): Remove _stat from name.
16041 (grow_tree_vec_stat): Likewise.
16042 * tree.h (make_tree_vec_stat): Adjust prototype.
16043 (grow_tree_vec_stat): Likewise.
16044 (make_tree_vec): Remove macro.
16045 (grow_tree_vec): Likewise.
16047 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16049 * fold-const.c (fold_build1_stat_loc): Adjust.
16050 (fold_build2_stat_loc): Likewise.
16051 (fold_build3_stat_loc): Likewise.
16052 * tree.c (build0_stat): Remove _stat from name.
16053 (build1_stat): Likewise.
16054 (build2_stat): Likewise.
16055 (build3_stat): Likewise.
16056 (build4_stat): Likewise.
16057 (build5_stat): Likewise.
16058 * tree.h (build1_loc): Remove macro, and rename _stat function
16060 (build2_loc): Likewise.
16061 (build3_loc): Likewise.
16062 (build4_loc): Likewise.
16063 (build5_loc): Likewise.
16065 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16067 * tree.c (make_int_cst_stat): Remove _stat from name.
16068 * tree.h (make_int_cst_stat): Adjust prototype.
16069 (make_int_cst): Remove macro.
16071 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16073 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16074 * tree.h (make_tree_binfo_stat): Adjust prototype.
16075 (make_tree_binfo): Remove.
16077 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16079 * tree.c (copy_node_stat): Rename to copy_node.
16080 (build_distinct_type_copy): Adjust.
16081 * tree.h (copy_node_stat): Adjust prototype.
16082 (copy_node): Remove macro.
16084 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16086 * tree.c (make_node_stat): rename to make_node.
16087 (build_tree_list_stat): Adjust.
16088 (build0_stat): Likewise.
16089 (build2_stat): Likewise.
16090 (build3_stat): Likewise.
16091 (build4_stat): Likewise.
16092 (build5_stat): Likewise.
16093 (build_decl_stat): Likewise.
16094 * tree.h (make_node_stat): Adjust prototype.
16095 (make_node): remove macro.
16097 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16099 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16100 (PPC_FEATURE2_SCV): Likewise.
16101 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16103 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16105 * config/aarch64/aarch64.c
16106 (aarch64_internal_mov_immediate): Add new special pattern.
16107 * config/aarch64/aarch64.md (*movdi_aarch64):
16108 Add reg/32bit const mov case.
16110 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16111 Richard Sandiford <richard.sandiford@linaro.org>
16113 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16114 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16115 Add integer and movi cases.
16116 (movi-split-hf-df-sf split, fp16): New.
16117 (enabled): Added TARGET_FP_F16INST.
16118 * config/aarch64/iterators.md (GPF_HF): New.
16119 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16121 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16123 * config/aarch64/aarch64.c
16124 (aarch64_simd_container_mode): Add prototype.
16125 (aarch64_expand_mov_immediate): Add HI support.
16126 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16127 (aarch64_can_const_movi_rtx_p): New.
16128 (aarch64_preferred_reload_class):
16129 Remove restrictions of using FP registers for certain SIMD operations.
16130 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16131 (aarch64_valid_floating_const): Add integer move validation.
16132 (aarch64_simd_imm_scalar_p): Remove.
16133 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16134 (aarch64_legitimate_constant_p): Expand list of supported cases.
16135 * config/aarch64/aarch64-protos.h
16136 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16137 (aarch64_reinterpret_float_as_int): New.
16138 (aarch64_simd_imm_scalar_p): Remove.
16139 * config/aarch64/constraints.md (Uvi): New.
16140 (Dd): Split into Ds and new Dd.
16141 * config/aarch64/aarch64.md (*movsi_aarch64):
16143 (*movdi_aarch64): Add SIMD mov case.
16145 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16147 * tree-predcom.c: (struct chain): Handle store-store chain in which
16148 stores for elimination only store loop invariant values.
16149 (execute_pred_commoning_chain): Ditto.
16150 (prepare_initializers_chain_store_elim): Ditto.
16151 (prepare_finalizers): Ditto.
16152 (is_inv_store_elimination_chain): New function.
16153 (initialize_root_vars_store_elim_1): New function.
16155 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16157 * tree-predcom.c: Revise general description of the pass.
16158 (enum chain_type): New enum type for store elimination.
16159 (struct chain): New field supporting store elimination.
16160 (struct component): Ditto.
16161 (dump_chain): Dump store-stores chain.
16162 (release_chain): Release resources.
16163 (split_data_refs_to_components): Compute and create component
16164 contains only stores for elimination.
16165 (get_chain_last_ref_at): New function.
16166 (make_invariant_chain): Initialization.
16167 (make_rooted_chain): Specify chain type in parameter and record it.
16168 (add_looparound_copies): Skip for store-stores chain.
16169 (determine_roots_comp): Compute type of chain and pass it to
16171 (initialize_root_vars_store_elim_2): New function.
16172 (finalize_eliminated_stores): New function.
16173 (remove_stmt): Handle store for elimination.
16174 (execute_pred_commoning_chain): Execute predictive commoning on
16175 store-store chains.
16176 (determine_unroll_factor): Skip unroll for store-stores chain.
16177 (prepare_initializers_chain_store_elim): New function.
16178 (prepare_initializers_chain): Hanlde store-store chain.
16179 (prepare_finalizers_chain, prepare_finalizers): New function.
16180 (tree_predictive_commoning_loop): Return integer value indicating
16181 if loop is unrolled or lcssa form is corrupted.
16182 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16184 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16186 * tree-predcom.c (initialize_root): Delete.
16187 (execute_pred_commoning_chain): Initialize root vars and replace
16188 reference of non-combined chain directly, rather than call above
16191 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16193 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16194 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16196 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16198 * tree-predcom.c (struct chain): New field init_seq.
16199 (release_chain): Release init_seq.
16200 (prepare_initializers_chain): Record intialization stmts in above
16202 (insert_init_seqs): New function.
16203 (tree_predictive_commoning_loop): Call insert_init_seqs.
16205 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16207 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16209 2017-07-28 Richard Biener <rguenther@suse.de>
16211 * match.pd: Remove superfluous :c.
16212 * genmatch.c (simplify::id): Add member.
16213 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16215 (current_id): New global.
16216 (dt_node::parent): Move from ...
16217 (dt_operand::parent): ... here. Add for_id member.
16218 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16219 (decision_tree::find_node): Relax order requirement when
16220 merging DT_TRUE nodes to ones inbetween the current simplify
16221 and the one we try to merge with. Add diagnostic whenever
16222 we need to enforce pattern order by not merging.
16223 (decision_tree::insert): Set current_id.
16224 (decision_tree::print_node): Dump parent node and for_id.
16225 (parser::last_id): Add member.
16226 (parser::push_simplify): Assign unique id.
16227 (parser::parser): Initialize last_id.
16229 2017-07-28 Martin Liska <mliska@suse.cz>
16232 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16233 gimple_build_debug_bind.
16235 2017-07-28 Richard Biener <rguenther@suse.de>
16237 PR tree-optimization/81502
16238 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16240 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16242 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16243 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16244 for BIT_FIELD_REF args.
16245 * fold-const.c (make_bit_field_ref): Likewise.
16246 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16248 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16251 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16252 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16253 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16254 Or it into SANITIZER_UNDEFINED.
16255 * ubsan.c: Include gimple-fold.h and varasm.h.
16256 (ubsan_expand_ptr_ifn): New function.
16257 (instrument_pointer_overflow): New function.
16258 (maybe_instrument_pointer_overflow): New function.
16259 (instrument_object_size): Formatting fix.
16260 (pass_ubsan::execute): Call instrument_pointer_overflow
16261 and maybe_instrument_pointer_overflow.
16262 * internal-fn.c (expand_UBSAN_PTR): New function.
16263 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16264 * sanitizer.def (__ubsan_handle_pointer_overflow,
16265 __ubsan_handle_pointer_overflow_abort): New builtins.
16266 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16267 * internal-fn.def (UBSAN_PTR): New internal function.
16268 * opts.c (sanitizer_opts): Add pointer-overflow.
16269 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16270 * fold-const.c (build_range_check): Compute pointer range check in
16271 integral type if pointer arithmetics would be needed. Formatting
16274 2017-07-28 Martin Liska <mliska@suse.cz>
16277 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16278 parameters that are of a variable-length.
16280 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16282 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16284 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16285 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16286 (CRT_CALL_STATIC_FUNCTION): Likewise.
16287 (ASM_DEFAULT_SPEC): New define.
16288 (ASM_SPEC32): Likewise.
16289 (ASM_SPEC64): Likewise.
16290 (ASM_SPEC_COMMON): Likewise.
16291 (ASM_SPEC): Likewise.
16292 (INVALID_64BIT): Likewise.
16293 (LINK_OS_DEFAULT_SPEC): Likewise.
16294 (LINK_OS_SPEC32): Likewise.
16295 (LINK_OS_SPEC64): Likewise.
16296 (POWERPC_LINUX): Likewise.
16297 (PTRDIFF_TYPE): Likewise.
16298 (RESTORE_FP_PREFIX): Likewise.
16299 (RESTORE_FP_SUFFIX): Likewise.
16300 (SAVE_FP_PREFIX): Likewise.
16301 (SAVE_FP_SUFFIX): Likewise.
16302 (SIZE_TYPE): Likewise.
16303 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16304 (TARGET_64BIT): Likewise.
16305 (TARGET_64BIT): Likewise.
16306 (TARGET_AIX): Likewise.
16307 (WCHAR_TYPE_SIZE): Likewise.
16308 (WCHAR_TYPE): Undefine.
16309 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16310 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16311 (CPP_OS_RTEMS_SPEC): Delete.
16312 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16313 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16315 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16317 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16319 PR tree-optimization/81578
16320 * tree-parloops.c (build_new_reduction): Bail out if
16321 reduction_code isn't one of the standard OpenMP reductions.
16322 Move the details printing after that decision.
16324 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16326 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16327 related to reload_in_progress.
16328 (splat_input_operand): Likewise.
16329 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16331 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16333 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16334 (TARGET_INSTANTIATE_DECLS): Likewise.
16335 (legitimate_indexed_address_p): Delete reload_in_progress code.
16336 (rs6000_debug_legitimate_address_p): Likewise.
16337 (rs6000_eliminate_indexed_memrefs): Likewise.
16338 (rs6000_emit_le_vsx_store): Likewise.
16339 (rs6000_emit_move_si_sf_subreg): Likewise.
16340 (rs6000_emit_move): Likewise.
16341 (register_to_reg_type): Likewise.
16342 (rs6000_pre_atomic_barrier): Likewise.
16343 (rs6000_machopic_legitimize_pic_address): Likewise.
16344 (rs6000_allocate_stack_temp): Likewise.
16345 (rs6000_address_for_fpconvert): Likewise.
16346 (rs6000_address_for_altivec): Likewise.
16347 (rs6000_secondary_memory_needed_rtx): Delete function.
16348 (rs6000_check_sdmode): Likewise.
16349 (rs6000_alloc_sdmode_stack_slot): Likewise.
16350 (rs6000_instantiate_decls): Likewise.
16351 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16352 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16353 Delete reload_in_progress.
16354 (*vec_reload_and_plus_<mptrsize>): Likewise.
16355 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16356 (vsx_div_v2di): Likewise.
16357 (vsx_udiv_v2di): Likewise.
16359 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16361 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16362 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16363 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16364 (rs6000_debug_reg_global): Delete print of LRA status.
16365 (rs6000_option_override_internal): Delete dead LRA related code.
16366 (rs6000_lra_p): Delete function.
16367 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16369 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16371 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16372 * config/riscv/rtems.h: New file.
16374 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16375 Sudakshina Das <sudi.das@arm.com>
16377 * config/aarch64/aarch64.md
16378 (define_split for and<mode>3nr_compare): Move
16379 non aarch64_logical_operand to a register.
16380 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16381 register immediate operand to a register.
16382 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16384 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16386 PR middle-end/81564
16387 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16389 2017-07-27 Richard Biener <rguenther@suse.de>
16391 PR tree-optimization/81573
16392 PR tree-optimization/81494
16393 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16394 multi defuse cycle case.
16396 2017-07-27 Richard Biener <rguenther@suse.de>
16398 PR tree-optimization/81571
16399 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16402 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16404 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16405 earlier and only if MASK_FPU is set. Adjust formatting.
16407 2017-07-27 Martin Liska <mliska@suse.cz>
16409 * opt-functions.awk: Add validation of value of Init.
16410 * optc-gen.awk: Pass new argument.
16412 2017-07-27 Martin Liska <mliska@suse.cz>
16414 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16415 Fix wrong condition.
16417 2017-07-27 Martin Liska <mliska@suse.cz>
16419 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16420 BBs and edges seen by autoFDO.
16422 2017-07-27 Richard Biener <rguenther@suse.de>
16424 PR tree-optimization/81502
16425 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16426 with incompatible but same sized type.
16427 (execute_update_addresses_taken): Likewise.
16429 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16431 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16432 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16434 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16437 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16438 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16439 Change s_operand to memory_operand.
16441 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16443 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16444 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16445 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16446 Emit instructions rather than returning an expression. Handle TFmode
16447 and KFmode by casting to TImode.
16448 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16449 (rs6000_emit_le_vsx_store): Likewise.
16450 * config/rs6000/vsx.md (VSX_TI): New iterator.
16451 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16452 (*vsx_le_undo_permute_<mode>): Likewise.
16453 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16454 emit the split sequence.
16455 (*vsx_le_perm_store_<mode>): Likewise.
16457 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16459 PR tree-optimization/81555
16460 PR tree-optimization/81556
16461 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16462 if true, force CHANGED for the recursive invocation.
16463 (reassociate_bb): Remember original length of ops array, pass
16464 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16466 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16467 attributes for noipa attribute. For naked attribute use
16468 lookup_attribute first before lookup_attribute_spec.
16469 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16471 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16472 for functions with noipa attribute.
16473 (cgraph_externally_visible_p): Return true for functions with noipa
16475 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16476 for functions with noipa attribute.
16477 * doc/extend.texi: Document noipa function attribute.
16478 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16479 also for functions with noipa attribute.
16480 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16482 2017-07-26 Andrew Pinski <apinski@cavium.com>
16484 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16485 vec_unalign_load_cost and vec_unalign_store_cost.
16487 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16489 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16490 -mvsx-small-integer option.
16491 (ISA_3_0_MASKS_IEEE): Likewise.
16492 (OTHER_VSX_VECTOR_MASKS): Likewise.
16493 (POWERPC_MASKS): Likewise.
16494 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16495 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16496 code, only testing for DImode being allowed in non-VSX floating
16498 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16499 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16501 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16502 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16503 TARGET_P8_VECTOR test.
16504 (rs6000_secondary_reload_simple_move): Likewise.
16505 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16506 since TARGET_P9_VECTOR was already tested.
16507 (rs6000_opt_masks): Remove -mvsx-small-integer.
16508 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16509 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16511 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16512 test for TARGET_VEXTRACTUB was used, and that uses
16514 (p9 extract splitter): Likewise.
16515 (vsx_extract_<mode>_di_p9): Likewise.
16516 (vsx_extract_<mode>_store_p9): Likewise.
16517 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16518 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16519 the elimination of TARGET_VSX_SMALL_INTEGER.
16520 (vsx_extract_<mode>_p8): Likewise.
16521 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16522 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16523 (vsx_set_<mode>_p9): Likewise.
16524 (vsx_set_v4sf_p9): Likewise.
16525 (vsx_set_v4sf_p9_zero): Likewise.
16526 (vsx_insert_extract_v4sf_p9): Likewise.
16527 (vsx_insert_extract_v4sf_p9_2): Likewise.
16528 * config/rs6000/rs6000.md (sign extend splitter): Change
16529 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16530 (floatsi<mode>2_lfiwax_mem): Likewise.
16531 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16532 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16533 since a test for TARGET_P9_VECTOR was used.
16534 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16535 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16536 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16537 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16538 TARGET_P8_VECTOR test.
16539 (fix_trunc<mode>si2_stfiwx): Likewise.
16540 (fix_trunc<mode>si2_internal): Likewise.
16541 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16542 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16544 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16545 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16546 TARGET_P8_VECTOR test.
16547 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16548 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16549 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16551 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16552 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16553 since a test for TARGET_P9_VECTOR was used.
16554 (splitter for loading small constants): Likewise.
16556 2017-07-26 Andrew Pinski <apinski@cavium.com>
16558 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16561 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16564 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16565 (fp_valid_at): Likewise.
16567 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16569 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16570 (qdf24xx_addrcost_table): Likewise.
16571 (cortexa57_tunings): Update to use generic_branch_cost.
16572 (cortexa72_tunings): Likewise.
16573 (cortexa73_tunings): Likewise.
16574 (qdf24xx_tunings): Likewise.
16576 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16578 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16579 (thunderx2t99_branch_cost): Likewise.
16580 (cortexa35_tunings): Update to use generic_branch_cost.
16581 (cortexa53_tunings): Likewise.
16582 (cortexa57_tunings): Likewise.
16583 (cortexa72_tunings): Likewise.
16584 (cortexa73_tunings): Likewise.
16585 (thunderx2t99_tunings): Likewise.
16587 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16589 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16590 (sparc_option_override): Honour MASK_FSMULD.
16591 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16592 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16593 * config/sparc/sparc.opt (mfsmuld): New option.
16594 * doc/invoke.texi (mfsmuld): Document option.
16596 2017-07-26 Marek Polacek <polacek@redhat.com>
16598 PR middle-end/70992
16599 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16601 2017-07-26 Richard Biener <rguenther@suse.de>
16603 * gimple-match-head.c (do_valueize): Return OP if valueize
16605 (get_def): New helper to get at the def stmt of a SSA name
16606 if valueize allows.
16607 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16608 do_valueize to get at the def stmt.
16609 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16611 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16613 PR middle-end/46932
16614 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16616 2017-07-26 Martin Liska <mliska@suse.cz>
16619 * function.c (expand_function_start): Make expansion of
16620 nonlocal_goto_save_area after parm_birth_insn.
16622 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16624 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16625 from all CPU target flags enable members.
16627 2017-07-26 Richard Biener <rguenther@suse.de>
16629 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16630 (decision_tree::gen): Make 'type' const.
16631 (write_predicate): Likewise.
16633 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16635 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16636 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16637 (rs6000_option_override_internal): Likewise.
16638 (rs6000_expand_vector_set): Likewise.
16639 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16640 (TARGET_UPPER_REGS_SF): Likewise.
16641 (TARGET_UPPER_REGS_DI): Likewise.
16642 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16643 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16644 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16645 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16646 (Splitters for DI constants in Altivec registers): Likewise.
16647 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16648 (vsx_set_v4sf_p9): Likewise.
16649 (vsx_set_v4sf_p9_zero): Likewise.
16650 (vsx_insert_extract_v4sf_p9): Likewise.
16651 (vsx_insert_extract_v4sf_p9_2): Likewise.
16653 2017-07-25 Carl Love <cel@us.ibm.com>
16655 * doc/extend.texi: Update the built-in documentation file for the
16656 existing built-in functions
16657 vector signed char vec_cnttz (vector signed char);
16658 vector unsigned char vec_cnttz (vector unsigned char);
16659 vector signed short vec_cnttz (vector signed short);
16660 vector unsigned short vec_cnttz (vector unsigned short);
16661 vector signed int vec_cnttz (vector signed int);
16662 vector unsigned int vec_cnttz (vector unsigned int);
16663 vector signed long long vec_cnttz (vector signed long long);
16664 vector unsigned long long vec_cnttz (vector unsigned long long);
16666 2017-07-25 Andrew Pinski <apinski@cavium.com>
16668 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16669 accesses where the use is for the first operand of a BIT_INSERT.
16671 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16674 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16675 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16677 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16679 * config/i386/gstabs.h: Delete.
16680 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16682 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16684 * config/i386/i386.c (ix86_decompose_address): Do not check for
16685 register RTX when looking at index_reg or base_reg.
16686 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16688 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16690 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16691 to update EH info here.
16693 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16695 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16697 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16699 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16701 2017-07-25 Torsten Duwe <duwe@suse.de>
16703 * common.opt: Introduce -fpatchable-function-entry
16704 command line option, and its variables function_entry_patch_area_size
16705 and function_entry_patch_area_start.
16706 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16707 including a two-value parser.
16708 * target.def (print_patchable_function_entry): New target hook.
16709 * targhooks.h (default_print_patchable_function_entry): New function.
16710 * targhooks.c (default_print_patchable_function_entry): Likewise.
16711 * toplev.c (process_options): Switch off IPA-RA if
16712 patchable function entries are being generated.
16713 * varasm.c (assemble_start_function): Look at the
16714 patchable-function-entry command line switch and current
16715 function attributes and maybe generate NOP instructions by
16716 calling the print_patchable_function_entry hook.
16717 * doc/extend.texi: Document patchable_function_entry attribute.
16718 * doc/invoke.texi: Document -fpatchable_function_entry
16719 command line option.
16720 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16722 * doc/tm.texi: Re-generate.
16724 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16727 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16728 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16730 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16732 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16734 2017-07-25 Richard Biener <rguenther@suse.de>
16736 PR tree-optimization/81455
16737 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16738 not walk in cycles when looking for guards.
16740 2017-07-25 Richard Biener <rguenther@suse.de>
16742 PR tree-optimization/81529
16743 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16744 when optimizing backedge uses.
16746 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16748 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16750 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16751 to dl_section_ref. On AIX, append an expression to subtract
16752 the size of the section length to dl_section_ref.
16754 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16756 * configure.ac: If any of the config.* scripts fail, exit 1.
16757 * configure: Regenerate.
16759 2017-07-25 Richard Biener <rguenther@suse.de>
16761 PR middle-end/81546
16762 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16763 of immediate uses, be more verbose on errors.
16765 2017-07-25 Richard Biener <rguenther@suse.de>
16767 PR tree-optimization/81510
16768 * tree-vect-loop.c (vect_is_simple_reduction): When the
16769 reduction stmt is not inside the loop bail out.
16771 2017-07-25 Richard Biener <rguenther@suse.de>
16773 PR tree-optimization/81303
16774 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16775 profitability check against LOOP_VINFO_NITERSM1.
16777 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16779 * domwalk.c (cmp_bb_postorder): Simplify.
16780 (sort_bbs_postorder): New function. Use it...
16781 (dom_walker::walk): ...here to optimize common cases.
16783 2017-07-25 Martin Liska <mliska@suse.cz>
16786 * ipa-visibility.c (function_and_variable_visibility): Make the
16787 redirection just on target that supports aliasing.
16788 Fix GNU coding style.
16790 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16793 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16795 * config/arm/rtems.h: Likewise.
16796 * config/bfin/rtems.h: Likewise.
16797 * config/i386/rtemself.h: Likewise.
16798 * config/lm32/rtems.h: Likewise.
16799 * config/m32c/rtems.h: Likewise.
16800 * config/m68k/rtemself.h: Likewise.
16801 * config/microblaze/rtems.h: Likewise.
16802 * config/mips/rtems.h: Likewise.
16803 * config/moxie/rtems.h: Likewise.
16804 * config/nios2/rtems.h: Likewise.
16805 * config/powerpcspe/rtems.h: Likewise.
16806 * config/rs6000/rtems.h: Likewise.
16807 * config/rtems.h: Likewise.
16808 * config/sh/rtems.h: Likewise.
16809 * config/sh/rtemself.h: Likewise.
16810 * config/sparc/rtemself.h: Likewise.
16812 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16815 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16816 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16817 * tree-ssa-structalias.c (alias_get_name): Same.
16819 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16822 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16823 instructions if no du chain is found.
16825 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16827 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16829 2017-07-25 Richard Biener <rguenther@suse.de>
16831 PR middle-end/81505
16832 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16835 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16837 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16838 upper-regs options.
16839 (ISA_2_7_MASKS_SERVER): Likewise.
16840 (ISA_3_0_MASKS_IEEE): Likewise.
16841 (OTHER_P8_VECTOR_MASKS): Likewise.
16842 (OTHER_VSX_VECTOR_MASKS): Likewise.
16843 (POWERPC_MASKS): Likewise.
16844 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16845 duplicate list of options.
16846 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16847 explicit -mupper-regs options.
16848 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16849 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16850 alias for -mupper-regs-df.
16851 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16852 (rs6000_init_hard_regno_mode_ok): Likewise.
16853 (rs6000_option_override_internal): Likewise.
16854 (rs6000_opt_masks): Likewise.
16855 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16856 options in terms of whether -mvsx or -mpower8-vector was used.
16857 (TARGET_UPPER_REGS_DI): Likewise.
16858 (TARGET_UPPER_REGS_SF): Likewise.
16859 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16860 -mupper-regs-* options.
16862 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16864 * passes.c (emergency_dump_function): Print some empty lines and a
16865 header before the RTL dump.
16867 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16869 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16871 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16874 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16875 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16877 2017-07-24 Carl Love <cel@us.ibm.com>
16879 * config/rs6000/rs6000-c.c: Add support for built-in functions
16880 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16881 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16882 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16883 vec_extract_fp_from_shortl): Add defines for the two builtins.
16884 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16885 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16887 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16888 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16889 * doc/extend.texi: Update the built-in documentation file for the
16890 new built-in function.
16892 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16895 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16897 * doc/generic.texi: Likewise.
16898 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16899 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16901 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16903 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16904 (aarch64_mls_elt_merge<mode>): Likewise.
16906 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16908 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16909 having __cxa_atexit.
16911 2017-07-23 Michael Collison <michael.collison@arm.com>
16913 * config/arm/arm.c (arm_option_override): Deprecate
16914 use of -mstructure-size-boundary.
16915 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16916 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16918 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16921 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16922 Reduce cost estimate for direct moves.
16924 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16927 * config/i386/i386.c (ix86_option_override_internal): Disable
16928 BMI, BMI2 and TBM instructions for -m16.
16930 2017-07-21 Carl Love <cel@us.ibm.com>
16932 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16933 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16934 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16935 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16936 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16937 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16938 VMULOSW): New enum "unspec" values.
16939 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16940 altivec_vmulosw): New patterns.
16941 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16942 VMULOSW): Add definitions.
16944 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16946 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16947 (qdf24xx): Likewise.
16948 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16949 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16950 (AARCH64_FL_V8_1): Renumber.
16951 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16952 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16953 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16954 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16955 rdma to feature modifiers list.
16957 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16959 PR middle-end/56727
16960 * ipa-visibility (function_and_variable_visibility): Convert
16961 recursive PLT call to direct call if appropriate.
16963 2017-07-21 Andrew Pinski <apinski@cavium.com>
16965 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16966 operand 1 to see if the types precision matches.
16967 * fold-const.c (operand_equal_p): Likewise.
16969 2017-07-21 Richard Biener <rguenther@suse.de>
16971 PR tree-optimization/81303
16972 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16973 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16974 (vect_peeling_hash_get_lowest_cost): Adjust.
16975 (vect_enhance_data_refs_alignment): Likewise. Use
16976 vect_get_peeling_costs_all_drs to compute the penalty for no
16977 peeling to match up costs.
16979 2017-07-21 Richard Biener <rguenther@suse.de>
16981 PR tree-optimization/81500
16982 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16983 we didn't identify a reduction path.
16985 2017-07-21 Tom de Vries <tom@codesourcery.com>
16986 Cesar Philippidis <cesar@codesourcery.com>
16988 PR gcov-profile/81442
16989 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16992 2017-07-21 Tom de Vries <tom@codesourcery.com>
16995 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16997 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16998 nvptx_override_options_after_change.
17000 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17002 * dwarf2out.c (output_file_names): Avoid double testing for
17003 dwarf_version >= 5.
17005 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17007 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17009 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17011 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17013 (try_crossjump_to_edge): Do not punt on partitioned functions.
17015 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17017 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17018 Put all BBs reachable only via paths crossing cold region to cold
17020 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17022 2016-07-21 Richard Biener <rguenther@suse.de>
17024 PR tree-optimization/81303
17025 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17026 into account prologue and epilogue iterations when raising
17027 min_profitable_iters to sth at least covering one vector iteration.
17029 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17031 * config/arm/arm.c (arm_test_cpu_arch_dat):
17034 2017-07-20 Nathan Sidwell <nathan@acm.org>
17036 Remove TYPE_METHODS.
17037 * tree.h (TYPE_METHODS): Delete.
17038 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17039 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17040 (dbxout_type_methods): Scan TYPE_FIELDS.
17041 (dbxout_type): Don't check TYPE_METHODS here.
17042 * function.c (use_register_for_decl): Always ignore register for
17043 class types when not optimizing.
17044 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17045 * tree.c (free_lang_data_in_type): Stitch out member functions and
17046 templates from TYPE_FIELDS.
17047 (build_distinct_type_copy, verify_type_variant,
17048 verify_type): Member fns are on TYPE_FIELDS.
17049 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17050 * tree-pretty-print.c (dump_generic_node): Likewise.
17052 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17055 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17056 V2TImode and V4TImode.
17057 (ix86_expand_vector_extract): Likewise.
17058 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17059 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17060 (ssescalarmode): Handle V4TImode and V2TImode.
17061 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17062 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17063 (VEXTRACTI128_MODE): New mode iterator.
17064 (splitter for *vec_extractv?ti first element): New.
17065 (VEC_INIT_MODE): New mode iterator.
17066 (vec_init<mode>): Consolidate 3 expanders into one using
17067 VEC_INIT_MODE mode iterator.
17069 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17071 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17072 non_spilled_static_chain_regno_p.
17074 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17076 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17078 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17080 * bb-reorder.c (connect_traces): Allow copying of blocks within
17083 2017-07-20 Richard Biener <rguenther@suse.de>
17085 * gimple.h (gimple_phi_result): Add gphi * overload.
17086 (gimple_phi_result_ptr): Likewise.
17087 (gimple_phi_arg): Likewise. Adjust index assert to only
17088 allow actual argument accesses rather than all slots available
17090 (gimple_phi_arg_def): Add gphi * overload.
17091 * tree-phinodes.c (make_phi_node): Initialize only actual
17093 (resize_phi_node): Clear memory not covered by old node,
17094 do not initialize excess argument slots.
17095 (reserve_phi_args_for_new_edge): Initialize new argument slot
17098 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17100 PR tree-optimization/81388
17102 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17104 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17105 by removing computation of may_be_zero.
17107 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17108 Tom de Vries <tom@codesourcery.com>
17110 PR middle-end/81030
17111 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17112 when gimple level profile disagrees with what RTL expander did.
17114 2017-07-20 Richard Biener <rguenther@suse.de>
17116 PR tree-optimization/61171
17117 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17118 (vect_analyze_stmt): Add slp instance parameter.
17119 (vectorizable_reduction): Likewise.
17120 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17121 (vect_is_simple_reduction): Deal with chains not detected
17122 as SLP reduction chain, specifically not properly associated
17123 chains containing a mix of plus/minus.
17124 (get_reduction_op): Remove.
17125 (get_initial_defs_for_reduction): Simplify, pass in whether
17126 this is a reduction chain, pass in the SLP node for the PHIs.
17127 (vect_create_epilog_for_reduction): Get the SLP instance as
17129 (vectorizable_reduction): Get the SLP instance as arg.
17130 During analysis remember the SLP node with the PHIs in the
17131 instance. Simplify getting at the vectorized reduction PHIs.
17132 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17133 through SLP instance.
17134 (vect_slp_analyze_operations): Likewise.
17135 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17136 (vect_transform_stmt): Likewise.
17138 2017-07-20 Tom de Vries <tom@codesourcery.com>
17140 PR tree-optimization/81489
17141 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17142 read of phi arg location to before loop that modifies phi.
17144 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17146 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17149 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17151 PR middle-end/81331
17152 * except.c (execute): Fix ordering issue.
17154 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17156 PR rtl-optimization/81423
17157 * combine.c (make_compound_operation_int): Don't try to optimize
17158 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17160 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17162 PR rtl-optimization/81423
17163 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17164 with a constant that is -1 in the truncated to mode.
17166 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17168 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17169 (determine_unlikely_bbs): ... here.
17170 * predict.h (propagate_unlikely_bbs_forward): Declare.
17171 * cfgexpand.c (pass_expand::execute): Use it.
17172 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17174 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17175 propagate_unlikely_bbs_forward.
17177 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17179 PR middle-end/81331
17180 * except.c (maybe_add_nop_after_section_switch): New function.
17183 2017-07-19 Tom de Vries <tom@codesourcery.com>
17185 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17187 2017-07-19 Tom de Vries <tom@codesourcery.com>
17189 * gimple.h (gimple_phi_arg): Make assert more strict.
17191 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17193 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17194 * config/rs6000/mmintrin.h: New file.
17195 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17197 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17199 PR tree-optimization/81346
17200 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17202 2017-07-19 Tom de Vries <tom@codesourcery.com>
17204 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17206 2017-07-19 Tom de Vries <tom@codesourcery.com>
17208 * config/nvptx/nvptx-modes.def: Add V2DImode.
17209 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17210 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17211 (nvptx_output_mov_insn): Handle lack of mov.b128.
17212 (nvptx_print_operand): Handle 'H' and 'L' codes.
17213 (nvptx_vector_mode_supported): Allow V2DImode.
17214 (nvptx_preferred_simd_mode): New function.
17215 (nvptx_data_alignment): New function.
17216 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17217 nvptx_preferred_simd_mode.
17218 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17220 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17222 2017-07-19 Tom de Vries <tom@codesourcery.com>
17224 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17225 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17226 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17227 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17228 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17229 (mov<VECIM>_insn): New define_insn.
17230 (define_expand "mov<VECIM>): New define_expand.
17232 2017-07-19 Tom de Vries <tom@codesourcery.com>
17234 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17236 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17238 PR tree-optimization/81346
17239 * fold-const.h (fold_div_compare, range_check_type): Declare.
17240 * fold-const.c (range_check_type): New function.
17241 (build_range_check): Use range_check_type.
17242 (fold_div_compare): No longer static, rewritten into
17243 a match.pd helper function.
17244 (fold_comparison): Don't call fold_div_compare here.
17245 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17246 as helper function.
17248 2017-07-19 Nathan Sidwell <nathan@acm.org>
17250 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17251 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17252 * tree.c (find_decls_types_r, verify_type): Use
17253 TYPE_{MIN,MAX}_VALUE_RAW.
17254 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17255 (hash_tree): Likewise.
17256 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17258 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17261 2017-07-18 Tom de Vries <tom@codesourcery.com>
17263 PR middle-end/81464
17264 * omp-expand.c (expand_omp_for_static_chunk): Handle
17265 equal-argument loop exit phi.
17267 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17270 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17271 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17272 operand 2 predicate.
17273 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17274 operand 2 predicate.
17275 (ror,rol -> rorx splitters): Use const_int_operand as
17276 operand 2 predicate.
17278 2017-06-18 Richard Biener <rguenther@suse.de>
17280 PR tree-optimization/81410
17281 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17282 the gap in the ! slp_perm SLP case after each group.
17284 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17286 PR middle-end/81463
17287 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17290 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17292 PR middle-end/81462
17293 * predict.c (set_even_probabilities): Cleanup; do not affect
17294 probabilities that are already known.
17295 (combine_predictions_for_bb): Call even when count is set.
17297 2017-07-18 Nathan Sidwell <nathan@acm.org>
17299 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17302 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17305 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17306 optimization for loop niter analysis.
17308 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17311 * config/avr/avr.c (avr_optimize_casesi): Don't use
17312 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17314 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17316 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17317 body_cost_vec from _vect_peel_extended_info.
17318 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17319 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17322 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17324 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17326 2017-07-18 Richard Biener <rguenther@suse.de>
17328 PR tree-optimization/80620
17329 PR tree-optimization/81403
17330 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17331 info when re-using a VN table entry.
17333 2017-07-18 Richard Biener <rguenther@suse.de>
17335 PR tree-optimization/81418
17336 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17337 vectype_in. Verify that with lane-reducing reduction operations
17338 we have a single def-use cycle.
17340 2017-07-17 Carl Love <cel@us.ibm.com>
17342 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17344 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17345 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17346 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17347 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17348 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17349 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17350 VMULOSW): New enum "unspec" values.
17351 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17352 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17353 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17354 altivec_vmulosw): New patterns.
17355 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17356 VMULOSW): Add definitions.
17357 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17359 * config/alpha/alpha.c: Include predict.h.
17361 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17363 * tree-vrp.c (compare_assert_loc): Fix comparison function
17364 to return predictable results.
17366 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17368 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17370 (subdi3): Likewise.
17371 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17372 * doc/invoke.texi (mexpand-adddi): Update text.
17374 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17376 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17377 that also clobbers the CC register. The old expand code is moved
17379 (*arc_clzsi2): ... here.
17380 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17381 the CC register. The old expand code is moved to ...
17382 (arc_ctzsi2): ... here.
17384 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17386 * config/arc/arc.opt (mindexed-loads): Use initial value
17387 TARGET_INDEXED_LOADS_DEFAULT.
17388 (mauto-modify-reg): Use initial value
17389 TARGET_AUTO_MODIFY_REG_DEFAULT.
17390 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17391 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17392 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17393 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17395 2017-07-17 Martin Liska <mliska@suse.cz>
17398 * opts.c (finish_options): Do not allow -fgnu-tm
17399 w/ -fsanitize={kernel-,}address. Say sorry.
17401 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17404 * tree-loop-distribution.c (classify_partition): Only assert on
17405 numer of iterations.
17406 (merge_dep_scc_partitions): Delete prameter. Update function call.
17407 (distribute_loop): Remove code handling loop with unknown niters.
17408 (pass_loop_distribution::execute): Skip loop with unknown niters.
17410 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17413 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17414 function sort_partitions_by_post_order.
17416 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17418 PR tree-optimization/81374
17419 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17420 the max index of basic blocks, rather than number of basic blocks.
17422 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17424 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17426 (arc_legitimate_pic_operand_p): Likewise.
17427 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17429 (arc_needs_pcl_p): Likewise.
17430 (arc_legitimate_pc_offset_p): Likewise.
17431 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17432 function is also used in constrains.md.
17433 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17434 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17435 PLUS. Only return true/false in known cases, otherwise assert.
17436 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17437 is already called in arc_legitimate_constant_p.
17438 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17440 (LEGITIMATE_PIC_OPERAND_P): Use
17441 arc_raw_symbolic_reference_mentioned_p function.
17442 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17447 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17448 Andrew Burgess <andrew.burgess@embecosm.com>
17450 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17451 (arc_return_address_register): New function.
17452 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17453 (arc_handle_fndecl_attribute): Add naked attribute.
17454 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17455 (TARGET_WARN_FUNC_RETURN): Likewise.
17456 (arc_allocate_stack_slots_for_args): New function.
17457 (arc_warn_func_return): Likewise.
17458 (machine_function): Change type fn_type.
17459 (arc_compute_function_type): Consider new naked function type,
17460 change function return type.
17461 (arc_must_save_register): Adapt to handle new
17462 arc_compute_function_type's return type.
17463 (arc_expand_prologue): Likewise.
17464 (arc_expand_epilogue): Likewise.
17465 (arc_return_address_regs): Delete.
17466 (arc_return_address_register): New function.
17467 (arc_epilogue_uses): Use above function.
17468 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17469 (arc_function_type): Change encoding, add naked type.
17470 (ARC_INTERRUPT_P): Change to handle the new encoding.
17471 (ARC_FAST_INTERRUPT_P): Likewise.
17472 (ARC_NORMAL_P): Define.
17473 (ARC_NAKED_P): Likewise.
17474 (arc_compute_function_type): Delete prototype.
17475 * config/arc/arc.md (in_ret_delay_slot): Use
17476 arc_return_address_register function.
17477 (simple_return): Likewise.
17478 (p_return_i): Likewise.
17480 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17482 PR tree-optimization/81428
17483 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17484 can't be built for those types.
17486 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17488 Remove stuff dead since r239246.
17490 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17491 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17492 (avr_inform_devices): Remove dead stuff.
17494 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17496 * config/arm/arm_neon.h: Fix softp typo.
17498 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17500 PR tree-optimization/81365
17501 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17502 aggregate moves onto bb predecessor edges, make sure there are no
17503 loads that could alias the lhs in between the start of bb and the
17506 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17509 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17510 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17511 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17513 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17515 PR tree-optimization/81396
17516 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17517 (init_symbolic_number): Initialize it to 1.
17518 (perform_symbolic_merge): Add n_ops from both operands into the new
17520 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17521 without base_addr as useless if they need more than one operation.
17522 (bswap_replace): Handle !bswap case for NULL base_addr.
17524 2017-07-17 Tom de Vries <tom@codesourcery.com>
17527 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17530 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17532 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17533 conditional builtin define __FIX_LEON3FT_B2BST.
17535 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17537 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17538 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17541 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17543 PR rtl-optimization/81424
17544 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17545 to remove potential trapping from operands if -fnon-call-exceptions.
17547 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17549 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17550 profile_proability for scalling.
17551 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17553 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17555 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17557 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17559 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17560 fixpoint arithmetics.
17562 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17564 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17565 fixpoint arithmetics.
17567 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17569 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17570 fixpoint arithmetics.
17572 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17574 * profile-count.h (profile_probability::from_reg_br_prob_note,
17575 profile_probability::to_reg_br_prob_note): New functions.
17576 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17577 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17578 * predict.c (probability_reliable_p): Update.
17579 (edge_probability_reliable_p): Update.
17580 (br_prob_note_reliable_p): Update.
17581 (invert_br_probabilities): Update.
17582 (add_reg_br_prob_note): New function.
17583 (combine_predictions_for_insn): Update.
17584 * asan.c (asan_clear_shadow): Update.
17585 * cfgbuild.c (compute_outgoing_frequencies): Update.
17586 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17587 (update_br_prob_note): Update.
17588 (rtl_verify_edges): Update.
17589 (purge_dead_edges): Update.
17590 (fixup_reorder_chain): Update.
17591 * emit-rtl.c (try_split): Update.
17592 * ifcvt.c (cond_exec_process_insns): Update.
17593 (cond_exec_process_if_block): Update.
17594 (dead_or_predicable): Update.
17595 * internal-fn.c (expand_addsub_overflow): Update.
17596 (expand_neg_overflow): Update.
17597 (expand_mul_overflow): Update.
17598 * loop-doloop.c (doloop_modify): Update.
17599 * loop-unroll.c (compare_and_jump_seq): Update.
17600 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17601 * predict.h: Update.
17602 * reorg.c (mostly_true_jump): Update.
17604 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17605 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17606 * config/arc/arc.c: (emit_unlikely_jump): Update.
17607 * config/arm/arm.c: (emit_unlikely_jump): Update.
17608 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17609 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17610 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17611 (ix86_print_operand): Update.
17612 (ix86_split_fp_branch): Update.
17613 (predict_jump): Update.
17614 * config/ia64/ia64.c (ia64_print_operand): Update.
17615 * config/mmix/mmix.c (mmix_print_operand): Update.
17616 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17617 (rs6000_expand_split_stack_prologue): Update.
17618 * config/rs6000/rs6000.c: Update.
17619 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17620 (s390_expand_vec_movstr): Update.
17621 (s390_expand_cs_tdsi): Update.
17622 (s390_expand_split_stack_prologue): Update.
17623 * config/sh/sh.c (sh_print_operand): Update.
17624 (expand_cbranchsi4): Update.
17625 (expand_cbranchdi4): Update.
17626 * config/sparc/sparc.c (output_v9branch): Update.
17627 * config/spu/spu.c (get_branch_target): Update.
17628 (ea_load_store_inline): Update.
17629 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17630 * config/tilepro/tilepro.c: Update.
17632 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17634 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17635 (gimplify_save_expr): Likewise.
17637 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17639 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17641 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17643 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17645 * timevar.def (TV_IPA_FNSUMMARY): Define.
17647 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17649 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17650 to back store errata sensitive sequence from being generated.
17651 (sqrtdf2_fix): Likewise.
17653 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17655 * tree-ssa-threadupdate.c (compute_path_counts,
17656 update_joiner_offpath_counts): Use profile_probability.
17658 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17661 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17663 * config/arm/arm-c.c (arm_cpu_builtins): Define
17664 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17666 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17668 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17669 array entries to represent __ieee128 versions of the
17670 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17671 scalar_extract_sig, and scalar_insert_exp built-in functions.
17672 (altivec_resolve_overloaded_builtin): Add special case handling
17673 for the __builtin_scalar_insert_exp function, as represented by
17674 the P9V_BUILTIN_VEC_VSIEDP constant.
17675 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17676 exponent support for __ieee128 argument.
17677 (VSESQP): Add scalar extract signature support for __ieee128
17679 (VSTDCNQP): Add scalar test negative support for __ieee128
17681 (VSIEQP): Add scalar insert exponent support for __int128 argument
17682 with __ieee128 result.
17683 (VSIEQPF): Add scalar insert exponent support for __ieee128
17684 argument with __ieee128 result.
17685 (VSTDCQP): Add scalar test data class support for __ieee128
17687 (VSTDCNQP): Add overload support for scalar test negative with
17688 __ieee128 argument.
17689 (VSTDCQP): Add overload support for scalar test data class
17690 __ieee128 argument.
17691 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17692 UNSPEC_VSX_SXSIGDP.
17693 (UNSPEC_VSX_SIEXPQP): New constant.
17694 (xsxexpqp): New insn for VSX scalar extract exponent quad
17696 (xsxsigqp): New insn for VSX scalar extract significand quad
17698 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17699 precision with floating point argument.
17700 (xststdcqp): New expand for VSX scalar test data class quad
17702 (xststdcnegqp): New expand for VSX scalar test negative quad
17704 (xststdcqp): New insn to match expansions for VSX scalar test data
17705 class quad precision and VSX scalar test negative quad precision.
17706 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17707 special case operand checking to enforce that second operand of
17708 VSX scalar test data class with quad precision argument is a 7-bit
17710 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17711 prototypes and descriptions of __ieee128 versions of
17712 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17713 scalar_test_data_class, and scalar_test_neg built-in functions.
17715 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17717 PR tree-optimization/81162
17718 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17719 replace a negate with an add.
17721 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17723 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17725 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17727 * config/arm/arm-c.c (arm_cpu_builtins): Define
17728 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17730 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17732 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17733 (armv8-r): Set ARM Cortex-R52 as default CPU.
17734 * config/arm/arm-tables.opt: Regenerate.
17735 * config/arm/arm-tune.md: Regenerate.
17736 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17738 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17739 extension for -mcpu=cortex-r52.
17741 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17743 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17744 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17745 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17746 (fp-armv8): Define it as FP_ARMv8 only.
17747 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17748 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17750 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17751 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17752 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17753 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17754 than TARGET_FPU_ARMV8.
17755 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17756 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17757 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17759 * config/arm/neon.md (neon_vrint): Likewise.
17760 (neon_vcvt): Likewise.
17761 (neon_<fmaxmin_op><mode>): Likewise.
17762 (<fmaxmin><mode>3): Likewise.
17763 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17764 * config/arm/predicates.md (arm_cond_move_operator): Check against
17765 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17767 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17769 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17770 to top of function.
17772 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17774 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17775 loop in comment with memset.
17777 2017-07-14 Martin Liska <mliska@suse.cz>
17779 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17780 * dwarf2out.c (is_java): Remove the function.
17781 (output_pubname): Remove usage of the function.
17782 (lower_bound_default): Remove usage of DW_LANG_Java.
17783 (gen_compile_unit_die): Likewise.
17784 * gcc.c: Remove compiler defaults for .java and .zip files.
17785 * gimple-expr.c (remove_suffix): Change as there's no longer
17786 extension than 4-letter one.
17787 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17788 (gimplify_save_expr): Likewise.
17789 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17790 as it's possible even for other languages than Java.
17791 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17792 * lto-opts.c (lto_write_options): Remove reference to Java.
17793 * opts.c (strip_off_ending): Update file extension handling.
17794 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17795 * tree-eh.c (lower_resx): Likewise.
17796 * tree.c (free_lang_data_in_type): Remove dead code.
17797 (find_decls_types_r): Likewise.
17798 (build_common_builtin_nodes): Remove Java from a comment.
17799 (verify_type): Remove dead code.
17800 * varasm.c (assemble_external): Remove Java from a comment.
17802 2017-07-14 Martin Liska <mliska@suse.cz>
17804 * opts.c (finish_options): Add quotes.
17805 (common_handle_option): Likewise.
17807 2017-07-14 Martin Liska <mliska@suse.cz>
17809 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17810 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17811 Remove N_SO_PASCAL.
17812 * dwarf2out.c (lower_bound_default): Do not handle
17814 (gen_compile_unit_die): Likewise.
17815 * gcc.c: Remove default extension binding for GNU Pascal.
17816 * stmt.c: Remove Pascal language from a comment.
17817 * xcoffout.c: Likewise.
17819 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17822 * diagnostic-show-locus.c (fixit_cmp): New function.
17823 (layout::layout): Sort m_fixit_hints.
17824 (column_range::column_range): Assert that the values are valid.
17825 (struct char_span): New struct.
17826 (correction::overwrite): New method.
17827 (struct source_line): New struct.
17828 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17829 calls in terms of classes source_line and char_span, and
17830 correction::overwrite.
17831 (selftest::test_overlapped_fixit_printing_2): New function.
17832 (selftest::diagnostic_show_locus_c_tests): Call it.
17834 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17836 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17837 early if there is no lhs.
17839 2017-07-13 Martin Liska <mliska@suse.cz>
17841 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17842 (gen_reference_type_die): Likewise.
17843 * stor-layout.c: Remove Pascal-related comment.
17845 2017-07-13 Martin Liska <mliska@suse.cz>
17847 * opts.c (finish_options): Add quotes to error messages.
17848 (parse_sanitizer_options): Likewise.
17850 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17852 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17854 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17856 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17858 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17860 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17862 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17864 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17867 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17868 provides the hardware capability bits, define the macro
17869 __BUILTIN_CPU_SUPPORTS__.
17870 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17871 if GLIBC does not provide the hardware capability bits. Add a
17872 gcc_unreachable call if the built-in cpu function is neither
17873 __builtin_cpu_is nor __builtin_cpu_supports.
17874 (rs6000_get_function_versions_dispatcher): Change the warning
17875 that an old GLIBC is used which does not export the capability
17876 bits to be an error.
17877 * doc/extend.texi (target_clones attribute): Document the
17878 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17879 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17880 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17881 the macros defined by GCC if the newer GLIBC is available.
17883 2017-07-12 Jeff Law <law@redhat.com>
17885 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17886 remaining includes slightly.
17887 * config/riscv/riscv-builtins.c: Include profile-count.h.
17889 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17892 * config/avr/avr.c (avr_set_current_function): In diagnostic
17893 messages: Quote keywords and (parts of) identifiers.
17894 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17897 2017-07-12 Carl Love <cel@us.ibm.com>
17899 * config/rs6000/rs6000-c.c: Add support for built-in functions
17900 vector bool char vec_revb (vector bool char);
17901 vector bool short vec_revb (vector short char);
17902 vector bool int vec_revb (vector bool int);
17903 vector bool long long vec_revb (vector bool long long);
17904 * doc/extend.texi: Update the built-in documentation file for the
17905 new built-in functions.
17907 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17909 * config/s390/s390.md: Remove movcc splitter.
17911 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17913 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17914 load/store on condition.
17916 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17919 * config/avr/avr.c (avr_encode_section_info)
17920 [progmem && !TREE_READONLY]: Error if progmem object needs
17923 2017-07-11 Michael Collison <michael.collison@arm.com>
17925 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17928 2017-07-11 Carl Love <cel@us.ibm.com>
17930 * config/rs6000/rs6000-c.c: Add support for builtins
17931 vector unsigned int vec_parity_lsbb (vector signed int);
17932 vector unsigned int vec_parity_lsbb (vector unsigned int);
17933 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17934 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17935 vector unsigned long long vec_parity_lsbb (vector signed long long);
17936 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17937 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17938 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17939 * doc/extend.texi: Update the built-in documentation file for the
17940 new built-in functions.
17942 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17944 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17945 (layout::m_primary_loc): New field.
17946 (layout::layout): Initialize new field. Move location filtering
17947 logic from here to...
17948 (layout::maybe_add_location_range): ...this new method. Add
17949 support for filtering to just the lines already specified by other
17951 (layout::will_show_line_p): New method.
17952 (gcc_rich_location::add_location_if_nearby): New method.
17953 (selftest::test_add_location_if_nearby): New test function.
17954 (selftest::diagnostic_show_locus_c_tests): Call it.
17955 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17958 2017-07-11 Tom de Vries <tom@codesourcery.com>
17960 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17961 (bb_first_real_insn): New function.
17962 (nvptx_single): Add extra initialization of broadcasted condition
17965 2017-07-11 Nathan Sidwell <nathan@acm.org>
17967 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17969 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17971 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17972 Binutils doc as TEXI will mess them up.
17973 * doc/invoke.texi (AVR Options): Same here.
17975 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17977 * config/sparc/sparc.opt (mfix-ut700): New option.
17978 (mfix-gr712rc): Likewise.
17979 (sparc_fix_b2bst): New variable.
17980 * doc/invoke.texi (SPARC options): Document them.
17981 (ARM options): Fix warnings.
17982 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17983 instructions to prevent sequences that can trigger the store-store
17984 errata for certain LEON3FT processors.
17985 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17986 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17987 * config/sparc/sparc.md (fix_b2bst): New attribute.
17988 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17990 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17993 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17995 (*rsqrtsf2_sse): Ditto.
17997 (div<mode>3): Macroize insn from divdf3 and divsf3
17998 using MODEF mode iterator.
18000 2017-07-10 Martin Sebor <msebor@redhat.com>
18002 PR tree-optimization/80397
18003 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18004 instead of testing for equality to INTEGER_TYPE.
18006 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18008 * config.gcc: Remove uclibc from arc target spec.
18010 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18012 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18014 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18017 * lto-wrapper.c (remove_option): New function.
18018 (merge_and_complain): Merge PIC/PIE options more realistically.
18020 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18022 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18026 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18027 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18028 * config.in: Regenerate.
18029 * configure: Regenerate.
18030 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18031 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18032 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18033 (TARGET_GASISR_PROLOGUES): ...target mask.
18034 * common/config/avr/avr-common.c
18035 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18036 Set -mgas-isr-prologues.
18037 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18038 INSERT_PASS_BEFORE for it.
18039 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18040 * config/avr/avr.c (avr_option_override)
18041 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18042 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18043 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18044 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18045 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18047 (make_avr_pass_pre_proep): New function.
18048 (emit_push_sfr) <treg>: Add argument to function and use it
18049 instead of TMP_REG.
18050 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18051 and set machine->gasisr.yes.
18052 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18053 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18054 __gcc_isr.n_pushed to .L__stack_usage.
18055 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18056 (avr_asm_final_postscan_insn): ...this new static function.
18057 * config/avr/avr.h (machine_function)
18058 <is_no_gccisr, use_L__stack_usage>: New fields.
18059 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18060 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18061 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18062 (gasisr, *gasisr): New expander and insn.
18063 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18064 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18065 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18067 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18069 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18072 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18074 Move jump-tables out of .text again.
18077 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18078 (ASM_OUTPUT_ADDR_VEC): New function.
18079 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18080 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18081 INSN_ADDRESSes as asm comment.
18082 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18083 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18084 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18085 * config/avr/avr.md (*tablejump): Adjust comment.
18086 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18087 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18089 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18090 (avr_output_addr_vec): New proto.
18091 (avr_log_t) <insn_addresses>: New field.
18093 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18096 * config/i386/i386.c (ix86_function_arg_advance): Set
18097 outgoing_args_on_stack to true if there are outgoing arguments
18099 (ix86_function_arg): Likewise.
18100 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18101 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18102 * config/i386/i386.h (machine_function): Add
18103 outgoing_args_on_stack.
18105 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18107 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18108 supporting pthreds.
18109 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18111 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18113 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18114 (REAL_H): Remove $(MACHMODE_H).
18115 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18117 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18118 $(MACHMODE_H) and double-int.h.
18119 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18121 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18124 2017-07-07 Andrew Pinski <apinski@cavium.com>
18126 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18127 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18129 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18131 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18132 Add warning if GCC was not configured to link against a GLIBC that
18133 exports the hardware capability bits.
18134 (make_resolver_func): Make resolver function private and not a
18135 COMDAT function. Create the name with clone_function_name instead
18136 of make_unique_name.
18139 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18140 correct operand in doing the split.
18142 2017-07-07 Carl Love <cel@us.ibm.com>
18144 * config/rs6000/rs6000-c: Add support for built-in function
18145 vector unsigned short vec_pack_to_short_fp32 (vector float,
18147 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18148 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18149 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18150 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18151 (convert_4f32_8i16): Add define_expand.
18152 * doc/extend.texi: Update the built-in documentation file for the
18153 new built-in function.
18155 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18157 * config/sparc/m8.md: New file.
18158 * config/sparc/sparc.md: Include m8.md.
18160 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18162 * config/sparc/sparc.opt: New option -mvis4b.
18163 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18164 (sparc_option_override): Handle VIS4B.
18165 (enum sparc_builtins): Define
18166 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18167 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18168 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18169 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18170 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18171 (check_constant_argument): New function.
18172 (sparc_vis_init_builtins): Define builtins
18173 __builtin_vis_dictunpack{8,16,32},
18174 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18175 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18176 __builtin_vis_fpcmpde{8,16,32}shl and
18177 __builtin_vis_fpcmpur{8,16,32}shl.
18178 (sparc_expand_builtin): Check that the constant operands to
18179 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18180 constant and in range.
18181 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18183 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18184 (SPARC_IMM5_P): Likewise.
18185 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18186 (enabled): Handle vis4b.
18187 (UNSPEC_DICTUNPACK): New unspec.
18188 (UNSPEC_FPCMPSHL): Likewise.
18189 (UNSPEC_FPUCMPSHL): Likewise.
18190 (UNSPEC_FPCMPDESHL): Likewise.
18191 (UNSPEC_FPCMPURSHL): Likewise.
18192 (cpu_feature): New CPU feature `vis4b'.
18193 (dictunpack{8,16,32}): New insns.
18194 (FPCSMODE): New mode iterator.
18195 (fpcscond): New code iterator.
18196 (fpcsucond): Likewise.
18197 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18198 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18199 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18200 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18201 * config/sparc/constraints.md: Define constraints `q' for unsigned
18202 2-bit integer constants and `t' for unsigned 5-bit integer
18204 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18206 (imm5_operand_dictunpack16): Likewise.
18207 (imm5_operand_dictunpack32): Likewise.
18208 (imm2_operand): Likewise.
18209 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18210 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18211 ditunpack* and fpcmp*shl builtins.
18213 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18215 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18216 * config.in: Add HAVE_AS_SPARC6 define.
18217 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18219 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18221 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18222 (CPP_CPU_SPEC): Handle m8.
18223 (ASM_CPU_SPEC): Likewise.
18224 * config/sparc/sparc-opts.h (enum processor_type): Add
18226 * config/sparc/sparc.c (m8_costs): New struct.
18227 (sparc_option_override): Handle TARGET_CPU_m8.
18228 (sparc32_initialize_trampoline): Likewise.
18229 (sparc64_initialize_trampoline): Likewise.
18230 (sparc_issue_rate): Likewise.
18231 (sparc_register_move_cost): Likewise.
18232 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18233 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18234 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18235 (CPP_CPU_SPEC): Handle M8.
18236 (ASM_CPU_SPEC): Likewise.
18237 (AS_M8_FLAG): Define.
18238 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18239 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18240 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18242 * configure: Regenerate.
18243 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18246 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18248 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18250 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18251 ("*movdi_insn_sp32"): Do not set v3pipe.
18252 ("*movsi_insn"): Likewise.
18253 ("*movdi_insn_sp64"): Likewise.
18254 ("*movsf_insn"): Likewise.
18255 ("*movdf_insn_sp32"): Likewise.
18256 ("*movdf_insn_sp64"): Likewise.
18257 ("*zero_extendsidi2_insn_sp64"): Likewise.
18258 ("*sign_extendsidi2_insn"): Likewise.
18259 ("*mov<VM32:mode>_insn"): Likewise.
18260 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18261 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18262 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18263 ("<vlop:code><VL:mode>3"): Likewise.
18264 ("*not_<vlop:code><VL:mode>3"): Likewise.
18265 ("*nand<VL:mode>_vis"): Likewise.
18266 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18267 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18268 ("one_cmpl<VL:mode>2"): Likewise.
18269 ("faligndata<VM64:mode>_vis"): Likewise.
18270 ("alignaddrsi_vis"): Likewise.
18271 ("alignaddrdi_vis"): Likweise.
18272 ("alignaddrlsi_vis"): Likewise.
18273 ("alignaddrldi_vis"): Likewise.
18274 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18275 ("bmaskdi_vis"): Likewise.
18276 ("bmasksi_vis"): Likewise.
18277 ("bshuffle<VM64:mode>_vis"): Likewise.
18278 ("cmask8<P:mode>_vis"): Likewise.
18279 ("cmask16<P:mode>_vis"): Likewise.
18280 ("cmask32<P:mode>_vis"): Likewise.
18281 ("pdistn<P:mode>_vis"): Likewise.
18282 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18284 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18286 * config/sparc/sparc.md ("subtype"): New insn attribute.
18287 ("*wrgsr_sp64"): Set insn subtype.
18288 ("*rdgsr_sp64"): Likewise.
18289 ("alignaddrsi_vis"): Likewise.
18290 ("alignaddrdi_vis"): Likewise.
18291 ("alignaddrlsi_vis"): Likewise.
18292 ("alignaddrldi_vis"): Likewise.
18293 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18294 ("fexpand_vis"): Likewise.
18295 ("fpmerge_vis"): Likewise.
18296 ("faligndata<VM64:mode>_vis"): Likewise.
18297 ("bshuffle<VM64:mode>_vis"): Likewise.
18298 ("cmask8<P:mode>_vis"): Likewise.
18299 ("cmask16<P:mode>_vis"): Likewise.
18300 ("cmask32<P:mode>_vis"): Likewise.
18301 ("fchksm16_vis"): Likewise.
18302 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18303 ("fmean16_vis"): Likewise.
18304 ("fp<plusminus_insn>64_vis"): Likewise.
18305 ("<plusminus_insn>v8qi3"): Likewise.
18306 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18307 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18308 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18309 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18310 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18311 ("*movqi_insn"): Likewise.
18312 ("*movhi_insn"): Likewise.
18313 ("*movsi_insn"): Likewise.
18314 ("movsi_pic_gotdata_op"): Likewise.
18315 ("*movdi_insn_sp32"): Likewise.
18316 ("*movdi_insn_sp64"): Likewise.
18317 ("movdi_pic_gotdata_op"): Likewise.
18318 ("*movsf_insn"): Likewise.
18319 ("*movdf_insn_sp32"): Likewise.
18320 ("*movdf_insn_sp64"): Likewise.
18321 ("*zero_extendhisi2_insn"): Likewise.
18322 ("*zero_extendqihi2_insn"): Likewise.
18323 ("*zero_extendqisi2_insn"): Likewise.
18324 ("*zero_extendqidi2_insn"): Likewise.
18325 ("*zero_extendhidi2_insn"): Likewise.
18326 ("*zero_extendsidi2_insn_sp64"): Likewise.
18327 ("ldfsr"): Likewise.
18328 ("prefetch_64"): Likewise.
18329 ("prefetch_32"): Likewise.
18330 ("tie_ld32"): Likewise.
18331 ("tie_ld64"): Likewise.
18332 ("*tldo_ldub_sp32"): Likewise.
18333 ("*tldo_ldub1_sp32"): Likewise.
18334 ("*tldo_ldub2_sp32"): Likewise.
18335 ("*tldo_ldub_sp64"): Likewise.
18336 ("*tldo_ldub1_sp64"): Likewise.
18337 ("*tldo_ldub2_sp64"): Likewise.
18338 ("*tldo_ldub3_sp64"): Likewise.
18339 ("*tldo_lduh_sp32"): Likewise.
18340 ("*tldo_lduh1_sp32"): Likewise.
18341 ("*tldo_lduh_sp64"): Likewise.
18342 ("*tldo_lduh1_sp64"): Likewise.
18343 ("*tldo_lduh2_sp64"): Likewise.
18344 ("*tldo_lduw_sp32"): Likewise.
18345 ("*tldo_lduw_sp64"): Likewise.
18346 ("*tldo_lduw1_sp64"): Likewise.
18347 ("*tldo_ldx_sp64"): Likewise.
18348 ("*mov<VM32:mode>_insn"): Likewise.
18349 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18350 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18352 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18354 * config/sparc/sparc.md ("type"): New insn type viscmp.
18355 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18357 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18358 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18359 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18360 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18362 ("n7_vis_logical_11cycle"): Likewise.
18363 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18364 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18365 * config/sparc/niagara.md ("niag_vis"): Likewise.
18366 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18367 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18369 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18371 * config/sparc/sparc.md: New instruction type `bmask'.
18372 (bmaskdi_vis): Use the `bmask' type.
18373 (bmasksi_vis): Likewise.
18374 * config/sparc/ultra3.md (us3_array): Likewise.
18375 * config/sparc/niagara7.md (n7_array): Likewise.
18376 * config/sparc/niagara4.md (n4_array): Likewise.
18377 * config/sparc/niagara2.md (niag2_vis): Likewise.
18378 (niag3_vis): Likewise.
18379 * config/sparc/niagara.md (niag_vis): Likewise.
18381 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18383 * ipa-comdats.c: Remove optimize check from gate.
18384 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18385 for functions not optimized.
18386 (ipa_fn_summary_read): Skip optimize check.
18387 (ipa_fn_summary_write): Likewise.
18388 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18390 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18392 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18393 (check_callers): Give up on caller which is not optimized.
18394 (inline_small_functions): Likewise.
18395 (ipa_inline): Do not give up when not optimizing.
18396 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18397 away unoptimizes cdtors.
18398 (whole_program_function_and_variable_visibility): Do
18399 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18400 * ipa.c (process_references): Do not check optimize.
18401 (symbol_table::remove_unreachable_nodes): Update optimize check.
18402 (set_writeonly_bit): Update optimize check.
18403 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18404 (pass_ipa_single_use::gate): Remove.
18406 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18408 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18409 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18410 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18411 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18412 permute_load, permute_store, adjust_extract, adjust_splat,
18413 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18414 replace_swap_with_copy, dump_swap_insn_table,
18415 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18416 recombine_lvx_pattern, recombine_stvx_pattern,
18417 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18418 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18419 to file rs6000-p8swap.c.
18420 * config/rs6000/rs6000-p8swap.c: New file.
18421 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18422 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18423 and rs6000*-*-* targets.
18425 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18427 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18429 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18431 * lto-wrapper.c (merge_and_complain): Do not merge
18432 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18433 fsigned_zeros, ftrapping_math, fwrapv.
18434 (append_compiler_options): Do not track these options.
18435 (append_linker_options): Likewie
18437 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18439 * cgraphunit.c (cgraph_node::finalize_function): When
18440 !flag_toplevel_reorde set no_reorder flag.
18441 (varpool_node::finalize_decl): Likewise.
18442 (symbol_table::compile): Drop no toplevel reorder path.
18444 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18446 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18447 edges; zero probability is not better than uninitialized.
18449 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18451 * asan.h (asan_sanitize_allocas_p): Declare.
18452 * asan.c (asan_sanitize_allocas_p): New function.
18453 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18454 (handle_builtin_alloca): Likewise.
18455 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18456 if !asan_sanitize_allocas_p.
18457 * params.def (asan-instrument-allocas): Add new option.
18458 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18459 * opts.c (common_handle_option): Disable allocas sanitization for
18462 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18464 * asan.c: Include gimple-fold.h.
18465 (get_last_alloca_addr): New function.
18466 (handle_builtin_stackrestore): Likewise.
18467 (handle_builtin_alloca): Likewise.
18468 (asan_emit_allocas_unpoison): Likewise.
18469 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18470 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18471 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18472 (instrument_builtin_call): Pass gimple iterator to
18473 get_mem_refs_of_builtin_call.
18474 (last_alloca_addr): New global.
18475 * asan.h (asan_emit_allocas_unpoison): Declare.
18476 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18477 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18478 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18479 if function calls alloca.
18480 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18481 * gimple-fold.h (replace_call_with_value): Declare.
18482 * internal-fn.c: Include asan.h.
18483 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18484 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18486 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18488 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18489 (C_SELFTEST_FLAGS): New.
18490 (CPP_SELFTEST_FLAGS): New.
18491 (SELFTEST_DEPS): New, from deps of s-selftest.
18492 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18493 (CPP_SELFTEST_DEPS): New.
18494 (selftest): Add dependency on s-selftest-c++.
18495 (s-selftest): Rename to...
18496 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18497 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18498 than SELFTEST_FLAGS.
18499 (selftest-gdb): Rename to...
18500 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18502 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18503 (selftest-valgrind): Rename to...
18504 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18506 (selftest-valgrind): Reintroduce as an alias for
18507 selftest-c-valgrind.
18508 (s-selftest-c++): New.
18509 (selftest-c++-gdb): New.
18510 (selftest-c++-valgrind): New.
18512 2017-07-06 Olivier Hainque <hainque@adacore.com>
18514 * gcc.c (process_command): When deciding if undefined variables
18515 should be ignored when processing specs, accept "gcc -v" as well.
18517 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18519 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18520 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18522 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18524 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18525 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18526 * config/arm/arm-tables.opt: Regenerate.
18527 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18529 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18531 2017-07-06 Carl Love <cel@us.ibm.com>
18533 * ChangeLog: Clean up from mid air collision
18535 2017-07-06 Carl Love <cel@us.ibm.com>
18537 * config/rs6000/rs6000-c.c: Add support for built-in functions
18538 vector signed int vec_subc (vector signed int, vector signed int);
18539 vector signed __int128 vec_subc (vector signed __int128,
18540 vector signed __int128);
18541 vector unsigned __int128 vec_subc (vector unsigned __int128,
18542 vector unsigned __int128);
18543 vector signed int vec_sube (vector signed int, vector signed int,
18544 vector signed int);
18545 vector unsigned int vec_sube (vector unsigned int,
18546 vector unsigned int,
18547 vector unsigned int);
18548 vector signed __int128 vec_sube (vector signed __int128,
18549 vector signed __int128,
18550 vector signed__int128);
18551 vector unsigned __int128 vec_sube (vector unsigned __int128,
18552 vector unsigned __int128,
18553 vector unsigned __int128);
18554 vector signed int vec_subec (vector signed int, vector signed int,
18555 vector signed int);
18556 vector unsigned int vec_subec (vector unsigned int,
18557 vector unsigned int,
18558 vector unsigned int);
18559 vector signed __int128 vec_subec (vector signed __int128,
18560 vector signed __int128,
18561 vector signed__int128);
18562 vector unsigned __int128 vec_subec (vector unsigned __int128,
18563 vector unsigned __int128,
18564 vector unsigned __int128);
18565 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18566 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18567 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18568 BU_ALTIVEC_OVERLOAD_X definitions.
18569 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18570 * doc/extend.texi: Update the built-in documentation file for the new
18571 built-in functions.
18573 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18576 * diagnostic-show-locus.c (layout::layout): Use start and finish
18577 spelling location for the start and finish of each range.
18578 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18579 Add unused aspect param.
18580 * input.c (expand_location_1): Add "aspect" param, and use it
18581 to access the correct part of the location.
18582 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18584 (expand_location_to_spelling_point): Likewise.
18585 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18586 param, and pass it to expand_location_1.
18588 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18590 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18591 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18592 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18593 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18594 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18595 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18596 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18597 _mm_maskz_getmant_ss): New intrinsics.
18598 (__builtin_ia32_getexpss128_mask): Changed to ...
18599 __builtin_ia32_getexpss128_round ... this.
18600 (__builtin_ia32_getexpsd128_mask): Changed to ...
18601 __builtin_ia32_getexpsd128_round ... this.
18602 * config/i386/i386-builtin-types.def
18603 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18604 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18605 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18606 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18607 __builtin_ia32_getmantss_mask_round): New builtins.
18608 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18609 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18610 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18611 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18612 * config/i386/sse.md
18613 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18614 avx512f_sgetexp<mode><mask_scalar_name>
18615 <round_saeonly_scalar_name> ... this.
18616 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18617 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18618 vgetexp<ssescalarmodesuffix>
18619 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18620 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18621 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18622 avx512f_vgetmant<mode><mask_scalar_name>
18623 <round_saeonly_scalar_name> ... this.
18624 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18625 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18626 vgetmant<ssescalarmodesuffix>
18627 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18628 %0<mask_scalar_operand4>, %1, %2
18629 <round_saeonly_scalar_mask_op4>, %3} ... this.
18630 * config/i386/subst.md (mask_scalar_operand4,
18631 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18632 round_saeonly_scalar_nimm_predicate): New subst attributes.
18634 2017-07-06 Julia Koval <julia.koval@intel.com>
18636 * config/i386/i386.c (ix86_erase_embedded_rounding):
18637 Remove code for old rounding pattern.
18639 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18641 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18643 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18645 * doc/sourcebuild.texi (Test Directives, Variants of
18646 dg-require-support): Add documentation for dg-require-stack-check.
18648 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18650 * config/i386/subst.md (mask_scalar, round_scalar,
18651 round_saeonly_scalar): New meta-templates.
18652 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18653 round_scalar_mask_operand3, round_scalar_mask_op3,
18654 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18655 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18656 round_saeonly_scalar_constraint,
18657 round_saeonly_scalar_prefix): New subst attribute.
18658 * config/i386/sse.md
18659 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18660 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18661 <round_scalar_name> ... this.
18662 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18663 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18664 <round_scalar_name> ... this.
18665 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18666 <sse>_vm<code><mode>3<mask_scalar_name>
18667 <round_saeonly_scalar_name> ... this.
18668 (v<plusminus_mnemonic><ssescalarmodesuffix>
18669 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18670 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18671 v<plusminus_mnemonic><ssescalarmodesuffix>
18672 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18673 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18674 (v<multdiv_mnemonic><ssescalarmodesuffix>
18675 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18676 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18677 v<multdiv_mnemonic><ssescalarmodesuffix>
18678 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18679 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18680 (v<maxmin_float><ssescalarmodesuffix>
18681 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18682 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18683 v<maxmin_float><ssescalarmodesuffix>
18684 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18685 %0<mask_scalar_operand3>, %1, %<iptr>2
18686 <round_saeonly_scalar_mask_op3>} ... this.
18688 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18690 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18691 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18693 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18694 Alan Hayward <alan.hayward@arm.com>
18695 David Sherwood <david.sherwood@arm.com>
18697 * combine.c (simplify_if_then_else): Remove "enum" before
18699 * compare-elim.c (can_eliminate_compare): Likewise.
18700 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18702 (aarch64_lookup_simd_builtin_type): Likewise.
18703 (aarch64_simd_builtin_type): Likewise.
18704 (aarch64_init_simd_builtin_types): Likewise.
18705 (aarch64_simd_expand_args): Likewise.
18706 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18708 (aarch64_reverse_mask): Likewise.
18709 (aarch64_simd_emit_reg_reg_move): Likewise.
18710 (aarch64_gen_adjusted_ldpstp): Likewise.
18711 (aarch64_ccmp_mode_to_code): Likewise.
18712 (aarch64_operands_ok_for_ldpstp): Likewise.
18713 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18714 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18716 (aarch64_min_divisions_for_recip_mul): Likewise.
18717 (aarch64_reassociation_width): Likewise.
18718 (aarch64_get_condition_code_1): Likewise.
18719 (aarch64_simd_emit_reg_reg_move): Likewise.
18720 (aarch64_simd_attr_length_rglist): Likewise.
18721 (aarch64_reverse_mask): Likewise.
18722 (aarch64_operands_ok_for_ldpstp): Likewise.
18723 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18724 (aarch64_gen_adjusted_ldpstp): Likewise.
18725 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18727 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18728 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18729 (arm_lookup_simd_builtin_type): Likewise.
18730 (arm_simd_builtin_type): Likewise.
18731 (arm_init_simd_builtin_types): Likewise.
18732 (arm_expand_builtin_args): Likewise.
18733 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18734 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18735 (ft32_setup_incoming_varargs): Likewise.
18736 (ft32_function_arg): Likewise.
18737 (ft32_function_arg_advance): Likewise.
18738 (ft32_pass_by_reference): Likewise.
18739 (ft32_arg_partial_bytes): Likewise.
18740 (ft32_valid_pointer_mode): Likewise.
18741 (ft32_addr_space_pointer_mode): Likewise.
18742 (ft32_addr_space_legitimate_address_p): Likewise.
18743 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18745 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18746 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18747 (iamcu_alignment): Likewise.
18748 (canonicalize_vector_int_perm): Likewise.
18749 (ix86_noce_conversion_profitable_p): Likewise.
18750 (ix86_mpx_bound_mode): Likewise.
18751 (ix86_operands_ok_for_move_multiple): Likewise.
18752 * config/microblaze/microblaze-protos.h
18753 (microblaze_expand_conditional_branch_reg): Likewise.
18754 * config/microblaze/microblaze.c
18755 (microblaze_expand_conditional_branch_reg): Likewise.
18756 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18758 (rs6000_reassociation_width): Likewise.
18759 (rs6000_invalid_binary_op): Likewise.
18760 (fusion_p9_p): Likewise.
18761 (emit_fusion_p9_load): Likewise.
18762 (emit_fusion_p9_store): Likewise.
18763 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18765 (riscv_hard_regno_mode_ok_p): Likewise.
18766 (riscv_address_insns): Likewise.
18767 (riscv_split_symbol): Likewise.
18768 (riscv_legitimize_move): Likewise.
18769 (riscv_function_value): Likewise.
18770 (riscv_hard_regno_nregs): Likewise.
18771 (riscv_expand_builtin): Likewise.
18772 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18773 (riscv_build_integer): Likewise.
18774 (riscv_split_integer): Likewise.
18775 (riscv_legitimate_constant_p): Likewise.
18776 (riscv_cannot_force_const_mem): Likewise.
18777 (riscv_regno_mode_ok_for_base_p): Likewise.
18778 (riscv_valid_base_register_p): Likewise.
18779 (riscv_valid_offset_p): Likewise.
18780 (riscv_valid_lo_sum_p): Likewise.
18781 (riscv_classify_address): Likewise.
18782 (riscv_legitimate_address_p): Likewise.
18783 (riscv_address_insns): Likewise.
18784 (riscv_load_store_insns): Likewise.
18785 (riscv_force_binary): Likewise.
18786 (riscv_split_symbol): Likewise.
18787 (riscv_force_address): Likewise.
18788 (riscv_legitimize_address): Likewise.
18789 (riscv_move_integer): Likewise.
18790 (riscv_legitimize_const_move): Likewise.
18791 (riscv_legitimize_move): Likewise.
18792 (riscv_address_cost): Likewise.
18793 (riscv_subword): Likewise.
18794 (riscv_output_move): Likewise.
18795 (riscv_canonicalize_int_order_test): Likewise.
18796 (riscv_emit_int_order_test): Likewise.
18797 (riscv_function_arg_boundary): Likewise.
18798 (riscv_pass_mode_in_fpr_p): Likewise.
18799 (riscv_pass_fpr_single): Likewise.
18800 (riscv_pass_fpr_pair): Likewise.
18801 (riscv_get_arg_info): Likewise.
18802 (riscv_function_arg): Likewise.
18803 (riscv_function_arg_advance): Likewise.
18804 (riscv_arg_partial_bytes): Likewise.
18805 (riscv_function_value): Likewise.
18806 (riscv_pass_by_reference): Likewise.
18807 (riscv_setup_incoming_varargs): Likewise.
18808 (riscv_print_operand): Likewise.
18809 (riscv_elf_select_rtx_section): Likewise.
18810 (riscv_save_restore_reg): Likewise.
18811 (riscv_for_each_saved_reg): Likewise.
18812 (riscv_register_move_cost): Likewise.
18813 (riscv_hard_regno_mode_ok_p): Likewise.
18814 (riscv_hard_regno_nregs): Likewise.
18815 (riscv_class_max_nregs): Likewise.
18816 (riscv_memory_move_cost): Likewise.
18817 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18818 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18819 (rl78_addr_space_address_mode): Likewise.
18820 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18822 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18823 (rs6000_reassociation_width): Likewise.
18824 (rs6000_invalid_binary_op): Likewise.
18825 (fusion_p9_p): Likewise.
18826 (emit_fusion_p9_load): Likewise.
18827 (emit_fusion_p9_store): Likewise.
18828 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18829 (ok_for_simple_move_operands): Likewise.
18830 (ok_for_simple_move_strict_operands): Likewise.
18831 (ok_for_simple_arith_logic_operands): Likewise.
18832 (visium_legitimize_reload_address): Likewise.
18833 (visium_select_cc_mode): Likewise.
18834 (output_cbranch): Likewise.
18835 (visium_split_double_move): Likewise.
18836 (visium_expand_copysign): Likewise.
18837 (visium_expand_int_cstore): Likewise.
18838 (visium_expand_fp_cstore): Likewise.
18839 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18840 (visium_function_arg): Likewise.
18841 (visium_function_arg_advance): Likewise.
18842 (visium_libcall_value): Likewise.
18843 (visium_setup_incoming_varargs): Likewise.
18844 (visium_legitimate_constant_p): Likewise.
18845 (visium_legitimate_address_p): Likewise.
18846 (visium_legitimize_address): Likewise.
18847 (visium_secondary_reload): Likewise.
18848 (visium_register_move_cost): Likewise.
18849 (visium_memory_move_cost): Likewise.
18850 (prepare_move_operands): Likewise.
18851 (ok_for_simple_move_operands): Likewise.
18852 (ok_for_simple_move_strict_operands): Likewise.
18853 (ok_for_simple_arith_logic_operands): Likewise.
18854 (visium_function_value_1): Likewise.
18855 (rtx_ok_for_offset_p): Likewise.
18856 (visium_legitimize_reload_address): Likewise.
18857 (visium_split_double_move): Likewise.
18858 (visium_expand_copysign): Likewise.
18859 (visium_expand_int_cstore): Likewise.
18860 (visium_expand_fp_cstore): Likewise.
18861 (visium_split_cstore): Likewise.
18862 (visium_select_cc_mode): Likewise.
18863 (visium_split_cbranch): Likewise.
18864 (output_cbranch): Likewise.
18865 (visium_print_operand_address): Likewise.
18866 * expmed.c (flip_storage_order): Likewise.
18867 * expmed.h (emit_cstore): Likewise.
18868 (flip_storage_order): Likewise.
18869 * genrecog.c (validate_pattern): Likewise.
18870 * hsa-gen.c (gen_hsa_addr): Likewise.
18871 * internal-fn.c (expand_arith_overflow): Likewise.
18872 * ira-color.c (allocno_copy_cost_saving): Likewise.
18873 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18874 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18875 (process_invariant_for_inheritance): Likewise.
18876 * lra-eliminations.c (move_plus_up): Likewise.
18877 * omp-low.c (lower_oacc_reductions): Likewise.
18878 * simplify-rtx.c (simplify_subreg): Likewise.
18879 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18880 (TARGET_CHKP_BOUND_MODE): Likewise..
18881 * targhooks.c (default_chkp_bound_mode): Likewise.
18882 (default_setup_incoming_vararg_bounds): Likewise.
18883 * targhooks.h (default_chkp_bound_mode): Likewise.
18884 (default_setup_incoming_vararg_bounds): Likewise.
18885 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18886 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18887 (have_whole_vector_shift): Likewise.
18888 * tree-vect-stmts.c (vectorizable_load): Likewise.
18889 * doc/tm.texi: Regenerate.
18891 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18893 Graceful degrade if Binutils PR21472 is not available.
18896 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18897 .rodata in flash test fails.
18898 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18899 * confgure: Regenerate.
18900 * config.in: Regenerate.
18901 * config/avr/avr.c (avr_asm_named_section)
18902 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18903 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18904 (avr_asm_init_sections): Same.
18906 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18908 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18909 (fma<VH:mode>4_intrinsic): Likewise.
18910 (*fmsub<VCVTF:mode>4): Likewise.
18911 (*fmsub<VH:mode>4_intrinsic): Likewise.
18913 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18916 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18917 Don't depend on "optimize > 0".
18918 (out_movhi_r_mr, out_movqi_mr_r): Same.
18919 (out_movhi_mr_r, out_movqi_r_mr): Same.
18920 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18921 io_address_operand on "optimize > 0".
18923 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18925 * tree-loop-distribution.c: Add general explanantion on the pass.
18926 (generate_loops_for_partition): Mark distributed loop.
18927 (pg_add_dependence_edges): New parameter. Handle alias data
18928 dependence specially and record it in the parameter if asked.
18929 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18930 (init_partition_graph_vertices, add_partition_graph_edge): New.
18931 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18932 (free_partition_graph_vdata, build_partition_graph): New.
18933 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18934 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18935 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18936 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18937 (version_for_distribution_p, finalize_partitions): New.
18938 (distribute_loop): Handle alias data dependence specially. Factor
18939 out loop fusion code as functions and call these functions.
18941 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18943 * tree-loop-distribution.c (classify_partition): New parameter and
18944 better handle reduction statement.
18945 (rdg_build_partitions): Revise comment.
18946 (distribute_loop): Compute statements in all partitions and pass it
18947 to classify_partition.
18949 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18951 * tree-loop-distribution.c (enum partition_type): New.
18952 (struct partition): New field type.
18953 (partition_merge_into): Add parameter. Update partition type.
18954 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18955 (build_rdg_partition_for_vertex): Compute partition type.
18956 (rdg_build_partitions): Dump partition type.
18957 (distribute_loop): Update calls to partition_merge_into.
18959 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18961 * tree-loop-distribution.c (struct ddr_hasher): New.
18962 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18964 (classify_partition): Call get_data_dependence.
18965 (pg_add_dependence_edges): Ditto.
18966 (distribute_loop): Release data dependence hash table.
18968 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18970 * tree-loop-distribution.c (ref_base_address): Delete.
18971 (similar_memory_accesses): Rename ...
18972 (share_memory_accesses): ... to this. Check if partitions access
18973 the same memory reference.
18974 (distribute_loop): Call share_memory_accesses.
18976 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18978 * tree-loop-distribution.c (struct partition): New field recording
18979 its data reference.
18980 (partition_alloc, partition_free): Init and release data refs.
18981 (partition_merge_into): Merge data refs.
18982 (build_rdg_partition_for_vertex): Collect data refs for partition.
18983 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18985 (distribute_loop): Remve data refs from vertice data of partition
18988 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18990 * tree-loop-distribution.c (params.h): Include header file.
18991 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18992 (datarefs_vec): New global var.
18993 (create_rdg_vertices): Use datarefs_vec directly.
18994 (free_rdg): Don't free data references.
18995 (build_rdg): Update use. Don't free data references.
18996 (distribute_loop): Compute global variable for data references.
18997 Bail out if there are too many data references.
18999 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19001 * tree-loop-distribution.c (loop_nest): New global var.
19002 (build_rdg): Use loop directly, rather than loop nest.
19003 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19005 (distribute_loop): Compute global variable loop nest. Update use.
19007 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19009 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19010 (partition_merge_into): New parameter. Dump reason for fusion.
19011 (distribute_loop): Update use of partition_merge_into.
19013 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19015 * tree-loop-distribution.c (bb_top_order_index): New.
19016 (bb_top_order_index_size, bb_top_order_cmp): New.
19017 (stmts_from_loop): Use topological order.
19018 (pass_loop_distribution::execute): Compute and release topological
19019 order for basic blocks.
19021 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19023 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19026 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19028 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19029 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19030 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19031 * internal-fn.def (LOOP_DIST_ALIAS): New.
19032 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19033 (fold_loop_internal_call): ... this.
19034 (vect_loop_dist_alias_call): New function.
19035 (set_uid_loop_bbs): Call fold_loop_internal_call.
19036 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19039 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19042 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19043 Require dead FLAGS_REG at the beginning of a peephole.
19045 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19048 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19049 arguments in the call to __builtin_ia32_sbb_u32.
19050 (_subborrow_u64): Swap _X and _Y arguments in the call to
19051 __builtin_ia32_sbb_u64.
19053 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19056 * tree-vrp.c (compare_assert_loc): Turn into a function template
19057 with stable template parameter. Only test if a->e is NULL,
19058 !a->e == !b->e has been verified already. Use e == NULL or
19059 e != NULL instead of e or ! e tests. If stable is true, don't use
19060 iterative_hash_expr, on the other side allow a or b or both NULL
19061 and sort the NULLs last.
19062 (process_assert_insertions): Sort using compare_assert_loc<false>
19063 instead of compare_assert_loc, later sort using
19064 compare_assert_loc<true> before calling process_assert_insertions_for
19065 in a loop. Use break instead of continue once seen NULL pointer.
19067 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19069 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19070 Cortex-R7 and Cortex-R8 processors.
19072 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19074 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19075 uninitialized while src is not.
19077 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19079 * common/config/arm/arm-common.c: Adjust include path for
19081 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19082 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19083 (arm-cpu-data.h): Likewise.
19084 (arm-cpu-cdata.h): Likewise.
19085 * config/arm/arm-cpu.h: Delete.
19086 * config/arm/arm-cpu-cdata.h: Delete.
19087 * config/arm/arm-cpu-data.h: Delete.
19089 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19091 * config/arm/arm-cpus.in (cortex-a55): New.
19092 (cortex-a75): Likewise.
19093 (cortex-a75.cortex-a55): Likewise.
19094 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19096 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19097 * config/arm/arm-cpu-cdata.h: Regenerate.
19098 * config/arm/arm-cpu-data.h: Regenerate.
19099 * config/arm/arm-cpu.h: Regenerate.
19100 * config/arm/arm-tables.opt: Regenerate.
19101 * config/arm/arm-tune.md: Regenerate.
19103 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19105 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19107 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19109 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19112 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19114 PR tree-optimization/81292
19115 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19116 full_string_p, also call adjust_related_strinfos if the adjustment
19117 is simple, otherwise invalidate related strinfos.
19119 2017-07-04 Martin Liska <mliska@suse.cz>
19122 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19123 newly created variable as DECL_IGNORED_P.
19125 2017-07-04 Martin Liska <mliska@suse.cz>
19128 * ipa-inline.c (inline_small_functions):
19129 Use xstrdup_for_dump.
19131 2017-07-04 Tom de Vries <tom@codesourcery.com>
19133 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19135 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19138 * config/darwin.c (darwin_function_switched_text_sections):
19141 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19143 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19145 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19147 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19149 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19151 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19152 min_profitable_iters, and th as inclusive lower bounds.
19153 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19154 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19155 for min_profitable_iters and min_profitable_estimate.
19156 (vect_transform_loop): Treat th as an inclusive lower bound.
19157 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19159 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19162 * config/darwin.c (darwin_function_switched_text_sections):
19163 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19164 in two pieces, and suppress the use of buf.
19166 2017-07-03 Nathan Sidwell <nathan@acm.org>
19168 * hash-table.h (hash_table_mod1): Fix indentation.
19170 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19172 PR middle-end/81290
19173 * predict.c (force_edge_cold): Be more careful about propagation
19175 * profile-count.h (profile_probability::guessed,
19176 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19178 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19180 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19182 * doc/invoke.texi (rcpc architecture extension): Document it.
19184 2017-07-03 Richard Biener <rguenther@suse.de>
19186 PR tree-optimization/60510
19187 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19188 the scalar reduction PHI and use it.
19189 (vectorizable_reduction): Properly guard the single_defuse_cycle
19190 path for non-SLP reduction chains where we cannot use it.
19191 Rework reduc_def/index and vector type deduction. Rework
19192 vector operand gathering during reduction op code-gen.
19193 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19194 chains dissolve the chain and leave it to non-SLP reduction
19197 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19199 * tree-data-ref.h (dr_alignment): Declare.
19200 * tree-data-ref.c (dr_alignment): New function.
19201 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19202 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19204 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19206 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19208 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19209 and base_misalignment fields.
19210 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19211 * tree-data-ref.c: Include builtins.h.
19212 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19213 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19214 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19215 * tree-vect-data-refs.c: Include tree-cfg.h.
19216 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19217 fields instead of calculating an alignment here.
19218 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19219 innermost_loop_behavior fields.
19221 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19223 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19225 (DR_STEP_ALIGNMENT): New macro.
19226 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19227 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19228 (create_data_ref): Print it.
19229 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19230 to tell whether the step preserves vector (mis)alignment.
19231 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19232 Move the check for an integer step and generalise to all INTEGER_CST.
19233 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19234 Print the outer step alignment.
19236 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19238 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19239 with offset_alignment.
19240 (DR_ALIGNED_TO): Delete.
19241 (DR_OFFSET_ALIGNMENT): New macro.
19242 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19243 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19244 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19245 (create_data_ref): Likewise.
19246 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19247 (vect_analyze_data_refs): Likewise.
19248 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19249 creating dummy innermost behavior.
19251 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19253 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19254 with a "innermost_loop_behavior *" and refeence tree.
19255 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19256 (create_data_ref): Update call accordingly.
19257 * tree-predcom.c (find_looparound_phi): Likewise.
19259 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19261 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19262 fields with dr_wrt_vec_loop.
19263 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19264 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19265 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19266 (vect_dr_behavior): New function.
19267 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19268 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19269 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19270 track whether the step preserves the misalignment.
19271 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19272 Use vect_dr_behavior.
19273 (vect_setup_realignment): Update call accordingly.
19274 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19275 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19276 call to vect_create_addr_base_for_vector_ref.
19277 (vect_create_cond_for_align_checks): Likewise.
19278 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19279 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19280 (vect_recog_mask_conversion_pattern): Likewise.
19281 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19282 (new_stmt_vec_info): Remove redundant zeroing.
19284 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19286 * common/config/arm/arm-common.c (arm_be8_option): New function.
19287 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19288 (ISA_ARMv6): Add isa_bit_be8.
19289 * config/arm/arm.h (arm_be8_option): Add prototype.
19290 (BE8_SPEC_FUNCTION): New define.
19291 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19292 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19293 (mlittle-endian): Similarly.
19294 (mbe8, mbe32): New options.
19295 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19296 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19298 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19300 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19302 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19304 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19305 (cleanup_tree_cfg_bb): Use it.
19306 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19308 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19310 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19313 * loop-doloop.c (add_test): Update profile.
19315 2017-07-03 Martin Liska <mliska@suse.cz>
19318 * sanopt.c (rewrite_usage_of_param): New function.
19319 (sanitize_rewrite_addressable_params): Likewise.
19320 (pass_sanopt::execute): Call rewrite_usage_of_param.
19322 2017-07-03 Richard Biener <rguenther@suse.de>
19324 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19325 back to using VIEW_CONVERT_EXPR.
19327 2017-07-03 Martin Liska <mliska@suse.cz>
19330 * doc/extend.texi: Document when a resolver function is
19331 generated for target_clones.
19333 2017-07-03 Martin Liska <mliska@suse.cz>
19335 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19336 and shadow memory of auto variables which are subject of
19337 use-after-scope sanitization.
19338 (asan_expand_mark_ifn): Add do set only when is_poison.
19340 2016-07-03 Richard Biener <rguenther@suse.de>
19342 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19344 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19345 (vectorizable_reduction): Perform reduction PHI creation when
19346 visiting a reduction PHI and adjust and simplify code generation
19347 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19348 (vect_transform_loop): Visit reduction PHIs.
19349 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19350 defs into the SLP tree.
19351 (vect_build_slp_tree): Reduction defs terminate the recursion.
19352 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19354 (vect_get_vec_defs_for_stmt_copy): Export.
19355 (vect_get_vec_defs): Likewise.
19356 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19358 (vect_get_vec_defs_for_stmt_copy): Declare.
19359 (vect_get_vec_defs): Likewise.
19361 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19363 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19364 parameter with a "loop" parameter and use it instead of the
19365 loop containing DR_STMT. Don't check simple_iv when doing
19366 BB analysis. Describe the two analysis modes in the comment.
19368 2017-07-03 Tom de Vries <tom@codesourcery.com>
19370 PR tree-optimization/69468
19371 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19372 (find_same_succ_bb): Handle ignore_edge_flags.
19374 2017-07-03 Tom de Vries <tom@codesourcery.com>
19376 PR tree-optimization/81192
19377 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19379 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19381 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19383 2017-07-03 Tom de Vries <tom@codesourcery.com>
19385 PR tree-optimization/81192
19386 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19387 BB_SAME_SUCC (bb) == NULL.
19389 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19391 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19394 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19396 * dumpfile.c: Include profile-count.h
19397 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19399 (insert_cond_bb): Update profile.
19400 * tree-cfg.h (insert_cond_bb): Update prototype.
19401 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19402 * tree-dump.c: Do not include tree-cfg.
19404 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19406 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19408 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19410 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19413 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19415 * tree-complex.c (expand_complex_div_wide): update profile.
19417 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19418 Alan Hayward <alan.hayward@arm.com>
19419 David Sherwood <david.sherwood@arm.com>
19421 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19422 (CORETYPES_H): New define.
19423 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19424 (insn-modes-inline.h, s-modes-inline-h): New rules.
19425 (generated_files): Add insn-modes-inline.h.
19426 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19427 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19428 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19429 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19430 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19431 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19432 (build/gencodes.o, build/genconditions.o): Likewise.
19433 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19434 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19435 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19436 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19437 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19438 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19439 * coretypes.h: Include everything up to real.h for generators.
19440 Include insn-modes.h first. Include wide-int-print.h after
19441 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19442 * machmode.h: Don't include insn-modes.h here.
19443 * function-tests.c: Remove includes of signop.h, machmode.h,
19444 double-int.h and wide-int.h.
19446 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19448 * optc-save-gen.awk: Likewise.
19449 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19450 * godump.c: Remove include of wide-int-print.h.
19451 * pretty-print.h: Likewise.
19452 * wide-int-print.cc: Likewise.
19453 * wide-int.cc: Likewise.
19454 * hash-map-tests.c: Remove include of signop.h.
19455 * hash-set-tests.c: Likewise.
19456 * rtl-tests.c: Likewise.
19457 * mkconfig.sh: Remove include of machmode.h.
19458 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19460 (emit_insn_modes_inline_h): ...this new function. Emit the code
19461 into an insn-modes-inline.h header file, adding appropriate
19462 include guards and end comments.
19463 (emit_insn_modes_c_header): Remove include of machmode.h.
19464 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19466 (main): Handle -i flag and call emit_insn_modes_inline_h when
19469 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19471 * tree-ssa-strlen.c (strinfo): Rename the length field to
19472 nonzero_chars. Add a full_string_p field.
19473 (compare_nonzero_chars, zero_length_string_p): New functions.
19474 (get_addr_stridx): Add an offset_out parameter.
19475 Use compare_nonzero_chars.
19476 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19477 (new_strinfo): Update after above changes to strinfo.
19478 (set_endptr_and_length): Set full_string_p.
19479 (get_string_length): Update after above changes to strinfo.
19480 (unshare_strinfo): Update call to new_strinfo.
19481 (maybe_invalidate): Likewise.
19482 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19483 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19484 as a uhwi instead of an shwi. Update after above changes to
19485 strinfo and new_strinfo.
19486 (zero_length_string): Assert that chainsi contains full strings.
19487 Use zero_length_string_p. Update call to new_strinfo.
19488 (adjust_related_strinfos): Update after above changes to strinfo.
19489 Copy full_string_p from origsi.
19490 (adjust_last_stmt): Use zero_length_string_p.
19491 (handle_builtin_strlen): Update after above changes to strinfo and
19492 new_strinfo. Install the lhs as the string length if the previous
19493 entry didn't describe a full string.
19494 (handle_builtin_strchr): Update after above changes to strinfo
19496 (handle_builtin_strcpy): Likewise.
19497 (handle_builtin_strcat): Likewise.
19498 (handle_builtin_malloc): Likewise.
19499 (handle_pointer_plus): Likewise.
19500 (handle_builtin_memcpy): Likewise. Track nonzero characters
19501 that aren't necessarily followed by a nul terminator.
19502 (handle_char_store): Likewise.
19504 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19506 PR tree-optimization/80769
19507 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19508 for malloc and calloc. Document the new invariant that all related
19509 strinfos have delayed lengths or none do.
19510 (verify_related_strinfos): Move earlier in file.
19511 (set_endptr_and_length): New function, split out from...
19512 (get_string_length): ...here. Also set the lengths of related
19514 (zero_length_string): Assert that chainsi has known (rather than
19516 (adjust_related_strinfos): Likewise.
19518 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19520 PR tree-optimization/81136
19521 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19522 assert that two references with the same misalignment have the same
19523 compile-time misalignment if those compile-time misalignments
19526 2017-07-01 Andi Kleen <ak@linux.intel.com>
19528 * print-tree.c (print_node): Print all attributes.
19530 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19532 * cfg.c (scale_bbs_frequencies): New function.
19533 * cfg.h (scale_bbs_frequencies): Declare it.
19534 * cfgloopanal.c (single_likely_exit): Cleanup.
19535 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19537 (scale_loop_profile): Likewise.
19538 (loop_version): Likewise.
19539 (create_empty_loop_on_edge): Update.
19540 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19541 scale_loop_frequencies, scale_loop_profile, loopify,
19542 loop_version): Update prototypes.
19543 * modulo-sched.c (sms_schedule): Update.
19544 * predict.c (unlikely_executed_edge_p): Also check probability.
19545 (probably_never_executed_edge_p): Fix typo.
19546 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19547 * tree-parloops.c (gen_parallel_loop): Update.
19548 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19549 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19550 * tree-ssa-loop-split.c (split_loop): Update.
19551 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19552 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19553 (vect_loop_versioning): Update.
19554 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19556 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19558 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19560 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19562 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19563 to keep profile consistent.
19565 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19567 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19568 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19569 * profile-count.h (max_safe_multiplier): Make unsigned.
19570 (profile_count::guessed_zero): New.
19572 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19574 * bb-reorder.c (fix_up_crossing_landing_pad,
19575 fix_crossing_conditional_branches): Use make_single_succ_edge
19576 to keep profile consistent.
19578 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19580 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19583 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19586 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19587 the right scopes, make sure cond_jump isn't preserved between multiple
19588 iterations. Search for fallthru edge whenever there are 3+ edges and
19589 use find_fallthru_edge for it.
19591 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19593 Patch by Alexander Monakov <amonakov@ispras.ru>
19594 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19595 probabilities consistently.
19597 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19599 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19600 emit_cmp_and_jump_insns.
19602 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19605 * tree-inline.c (expand_call_inline): Combine profile statuses.
19607 2017-06-30 Andrew Pinski <apinski@cavium.com>
19609 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19610 fold_stmt returned true.
19612 2017-06-30 Nathan Sidwell <nathan@acm.org>
19614 * ggc.h (empty_string): Delete.
19615 * cfgexpand.c (expand_asm_stmt): Use plain "".
19616 * optabs.c (expand_asm_memory_barrier): Likewise.
19617 * stringpool.c (empty_string): Delete.
19618 (digit_vector, digit_string): Delete.
19619 (ggc_alloc_string): Use plain "", don't optimize single digit
19620 strings. Use ggc_alloc_atomic.
19622 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19624 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19625 comparison set and one other set, use the cost of the non-comparison
19628 2017-06-30 Nathan Sidwell <nathan@acm.org>
19630 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19633 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19635 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19636 loops. Remove now unneeded calls to gimple_switch_set_label() that
19637 just set removed labels to NULL_TREE.
19639 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19641 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19642 (set_range_info): ...here. Only call set_range_info_raw if domain
19644 (set_nonzero_bits): Call set_range_info_raw.
19645 * tree-ssanames.h (set_range_info_raw): New.
19647 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19650 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19651 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19652 of nonimmediate_operand and <store_mask_constraint> instead of m
19653 for the input operand. For V8FI iterator, always split if input
19654 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19655 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19656 <store_mask_predicate> instead of register_operand and
19657 <store_mask_constraint> instead of v for the input operand. Make
19658 sure both operands aren't MEMs for if not <mask_applied>.
19660 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19662 * lto-wrapper.c (copy_file) Close both file descriptors before
19665 2017-06-30 Martin Liska <mliska@suse.cz>
19668 * multiple_target.c (create_dispatcher_calls): Make ifunc
19669 also for function that don't have calls or are not referenced.
19671 2017-06-30 Richard Biener <rguenther@suse.de>
19673 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19674 analyze the first scalar stmt. Move vector type computation
19675 for the BB case here from ...
19676 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19677 live operation processing in the SLP case properly.
19679 2017-06-30 Richard Biener <rguenther@suse.de>
19681 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19683 2017-06-30 Martin Liska <mliska@suse.cz>
19686 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19687 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19689 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19691 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19694 2017-06-30 Richard Biener <rguenther@suse.de>
19696 PR tree-optimization/81249
19697 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19698 condition reduction result to original scalar type.
19700 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19702 * profile-count.h (enum profile_quality): Fix typos and whitespace
19705 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19707 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19708 type for branch probabilities.
19710 2017-06-29 Julian Brown <julian@codesourcery.com>
19711 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19713 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19714 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19715 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19716 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19718 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19720 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19721 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19722 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19723 CC usage from generic code to here.
19724 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19725 CC usage into the target macros.
19727 2017-06-29 Maya Rashish <coypu@sdf.org>
19729 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19732 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19734 * arm/arm-builtins.c: Include profile-count.h
19735 * except.c (sjlj_emit_function_enter): Use
19736 profile_probability::unlikely.
19738 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19740 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19741 and tocrel_offset be pointer args rather than implicitly using
19743 (legitimate_constant_pool_address_p, rs6000_emit_move,
19744 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19745 tocrel_offset and use in toc_relative_expr_p call.
19746 (print_operand, print_operand_address): Use static tocrel_base_oac
19747 and tocrel_offset_oac.
19748 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19751 2017-06-29 Maya Rashish <coypu@sdf.org>
19753 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19755 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19757 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19758 objects, take into account only the alignment of 'op0' and 'mode1' if
19761 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19763 * ccmp.c (ccmp_tree_comparison_p): New function.
19764 (ccmp_candidate_p): Update to use above function.
19765 (get_compare_parts): New function.
19766 (expand_ccmp_next): Update to use new functions.
19767 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19769 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19770 take mode as argument.
19771 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19772 * expr.c (expand_expr_real_1): Pass mode as argument.
19774 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19776 * combine.c (combine_instructions): Print insns to dump_file, together
19779 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19781 * asan.c (asan_emit_stack_protection): Update.
19782 (create_cond_insert_point): Update.
19783 * auto-profile.c (afdo_propagate_circuit): Update.
19784 * basic-block.h (struct edge_def): Turn probability to
19785 profile_probability.
19786 (EDGE_FREQUENCY): Update.
19787 * bb-reorder.c (find_traces_1_round): Update.
19788 (better_edge_p): Update.
19789 (sanitize_hot_paths): Update.
19790 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19791 (make_single_succ_edge): Update.
19792 (check_bb_profile): Update.
19793 (dump_edge_info): Update.
19794 (update_bb_profile_for_threading): Update.
19795 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19796 probabilitycount to 0.
19797 * cfgbuild.c (compute_outgoing_frequencies): Update.
19798 * cfgcleanup.c (try_forward_edges): Update.
19799 (outgoing_edges_match): Update.
19800 (try_crossjump_to_edge): Update.
19801 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19802 (expand_gimple_tailcall): Update.
19803 (construct_init_block): Use make_single_succ_edge.
19804 (construct_exit_block): Use make_single_succ_edge.
19805 * cfghooks.c (verify_flow_info): Update.
19806 (redirect_edge_succ_nodup): Update.
19807 (split_edge): Update.
19808 (account_profile_record): Update.
19809 * cfgloopanal.c (single_likely_exit): Update.
19810 * cfgloopmanip.c (scale_loop_profile): Update.
19811 (set_zero_probability): Remove.
19812 (duplicate_loop_to_header_edge): Update.
19813 * cfgloopmanip.h (loop_version): Update prototype.
19814 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19815 (force_nonfallthru_and_redirect): Update.
19816 (update_br_prob_note): Update.
19817 (rtl_verify_edges): Update.
19818 (purge_dead_edges): Update.
19819 (rtl_lv_add_condition_to_bb): Update.
19820 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19821 * cgraphunit.c (init_lowered_empty_function): Update.
19822 (cgraph_node::expand_thunk): Update.
19823 * cilk-common.c: Include profile-count.h
19824 * dojump.c (inv): Remove.
19825 (jumpifnot): Update.
19826 (jumpifnot_1): Update.
19827 (do_jump_1): Update.
19829 (do_jump_by_parts_greater_rtx): Update.
19830 (do_compare_rtx_and_jump): Update.
19831 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19832 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19833 * dwarf2cfi.c: Include profile-count.h
19834 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19835 (sjlj_emit_dispatch_table): Likewise.
19836 * explow.c: Include profile-count.h
19837 * expmed.c (emit_store_flag_force): Update.
19838 (do_cmp_and_jump): Update.
19839 * expr.c (compare_by_pieces_d::generate): Update.
19840 (compare_by_pieces_d::finish_mode): Update.
19841 (emit_block_move_via_loop): Update.
19842 (store_expr_with_bounds): Update.
19843 (store_constructor): Update.
19844 (expand_expr_real_2): Update.
19845 (expand_expr_real_1): Update.
19846 * expr.h (try_casesi, try_tablejump): Update prototypes.
19847 * gimple-pretty-print.c (dump_probability): Update.
19848 (dump_profile): New.
19849 (dump_gimple_label): Update.
19850 (dump_gimple_bb_header): Update.
19851 * graph.c (draw_cfg_node_succ_edges): Update.
19852 * hsa-gen.c (convert_switch_statements): Update.
19853 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19854 (find_if_case_1): Update.
19855 (find_if_case_2): Update.
19856 * internal-fn.c (expand_arith_overflow_result_store): Update.
19857 (expand_addsub_overflow): Update.
19858 (expand_neg_overflow): Update.
19859 (expand_mul_overflow): Update.
19860 (expand_vector_ubsan_overflow): Update.
19861 * ipa-cp.c (good_cloning_opportunity_p): Update.
19862 * ipa-split.c (split_function): Use make_single_succ_edge.
19863 * ipa-utils.c (ipa_merge_profiles): Update.
19864 * loop-doloop.c (add_test): Update.
19865 (doloop_modify): Update.
19866 * loop-unroll.c (compare_and_jump_seq): Update.
19867 (unroll_loop_runtime_iterations): Update.
19868 * lra-constraints.c (lra_inheritance): Update.
19869 * lto-streamer-in.c (input_cfg): Update.
19870 * lto-streamer-out.c (output_cfg): Update.
19871 * mcf.c (adjust_cfg_counts): Update.
19872 * modulo-sched.c (sms_schedule): Update.
19873 * omp-expand.c (expand_omp_for_init_counts): Update.
19874 (extract_omp_for_update_vars): Update.
19875 (expand_omp_ordered_sink): Update.
19876 (expand_omp_for_ordered_loops): Update.
19877 (expand_omp_for_generic): Update.
19878 (expand_omp_for_static_nochunk): Update.
19879 (expand_omp_for_static_chunk): Update.
19880 (expand_cilk_for): Update.
19881 (expand_omp_simd): Update.
19882 (expand_omp_taskloop_for_outer): Update.
19883 (expand_omp_taskloop_for_inner): Update.
19884 * omp-simd-clone.c (simd_clone_adjust): Update.
19885 * optabs.c (expand_doubleword_shift): Update.
19886 (expand_abs): Update.
19887 (emit_cmp_and_jump_insn_1): Update.
19888 (expand_compare_and_swap_loop): Update.
19889 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19890 * predict.c (predictable_edge_p): Update.
19891 (edge_probability_reliable_p): Update.
19892 (set_even_probabilities): Update.
19893 (combine_predictions_for_insn): Update.
19894 (combine_predictions_for_bb): Update.
19895 (propagate_freq): Update.
19896 (estimate_bb_frequencies): Update.
19897 (force_edge_cold): Update.
19898 * profile-count.c (profile_count::dump): Add missing space into dump.
19899 (profile_count::debug): Add newline.
19900 (profile_count::differs_from_p): Explicitly convert to unsigned.
19901 (profile_count::stream_in): Update.
19902 (profile_probability::dump): New member function.
19903 (profile_probability::debug): New member function.
19904 (profile_probability::differs_from_p): New member function.
19905 (profile_probability::differs_lot_from_p): New member function.
19906 (profile_probability::stream_in): New member function.
19907 (profile_probability::stream_out): New member function.
19908 * profile-count.h (profile_count_quality): Rename to ...
19909 (profile_quality): ... this one.
19910 (profile_probability): New.
19911 (profile_count): Update.
19912 * profile.c (compute_branch_probabilities): Update.
19913 * recog.c (peep2_attempt): Update.
19914 * sched-ebb.c (schedule_ebbs): Update.
19915 * sched-rgn.c (find_single_block_region): Update.
19916 (compute_dom_prob_ps): Update.
19917 (schedule_region): Update.
19918 * sel-sched-ir.c (compute_succs_info): Update.
19919 * stmt.c (struct case_node): Update.
19920 (do_jump_if_equal): Update.
19921 (get_outgoing_edge_probs): Update.
19922 (conditional_probability): Update.
19923 (emit_case_dispatch_table): Update.
19924 (expand_case): Update.
19925 (expand_sjlj_dispatch_table): Update.
19926 (emit_case_nodes): Update.
19927 * targhooks.c: Update.
19928 * tracer.c (better_p): Update.
19929 (find_best_successor): Update.
19930 * trans-mem.c (expand_transaction): Update.
19931 * tree-call-cdce.c: Update.
19932 * tree-cfg.c (gimple_split_edge): Upate.
19933 (move_sese_region_to_fn): Upate.
19934 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19935 * tree-eh.c (lower_resx): Upate.
19936 (cleanup_empty_eh_move_lp): Upate.
19937 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19938 * tree-inline.c (copy_edges_for_bb): Update.
19939 (copy_cfg_body): Update.
19940 * tree-parloops.c (gen_parallel_loop): Update.
19941 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19942 (gimple_gen_time_profiler): Update.
19943 * tree-ssa-dce.c (remove_dead_stmt): Update.
19944 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19945 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19946 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19947 (unloop_loops): Update.
19948 (try_peel_loop): Update.
19949 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19950 * tree-ssa-loop-split.c (connect_loops): Update.
19951 (split_loop): Update.
19952 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19953 (hoist_guard): Update.
19954 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19955 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19956 (value_replacement): Update.
19957 * tree-ssa-reassoc.c (branch_fixup): Update.
19958 * tree-ssa-tail-merge.c (replace_block_by): Update.
19959 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19960 (create_edge_and_update_destination_phis): Update.
19961 (compute_path_counts): Update.
19962 (recompute_probabilities): Update.
19963 (update_joiner_offpath_counts): Update.
19964 (freqs_to_counts_path): Update.
19965 (duplicate_thread_path): Update.
19966 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19967 (struct switch_conv_info): Update.
19968 (gen_inbound_check): Update.
19969 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19970 (vect_do_peeling): Update.
19971 (vect_loop_versioning): Update.
19972 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19973 (optimize_mask_stores): Update.
19974 * ubsan.c (ubsan_expand_null_ifn): Update.
19975 * value-prof.c (gimple_divmod_fixed_value): Update.
19976 (gimple_divmod_fixed_value_transform): Update.
19977 (gimple_mod_pow2): Update.
19978 (gimple_mod_pow2_value_transform): Update.
19979 (gimple_mod_subtract): Update.
19980 (gimple_mod_subtract_transform): Update.
19981 (gimple_ic): Update.
19982 (gimple_stringop_fixed_value): Update.
19983 (gimple_stringops_transform): Update.
19984 * value-prof.h: Update.
19986 2017-06-29 Carl Love <cel@us.ibm.com>
19988 * config/rs6000/rs6000-c.c: Add support for built-in functions
19989 vector signed int vec_signed (vector float);
19990 vector signed long long vec_signed (vector double);
19991 vector signed int vec_signed2 (vector double, vector double);
19992 vector signed int vec_signede (vector double);
19993 vector signed int vec_signedo (vector double);
19994 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19995 instruction generator.
19996 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19997 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19998 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20000 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20001 vunsignede_v2df): Add define_expands.
20002 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20003 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20004 VEC_UNSIGNEDO): Add definitions.
20005 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20006 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20007 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20008 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20009 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20010 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20011 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20012 vec_unsignede, vec_unsignedo): Add builtin defines.
20013 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20015 * doc/extend.texi: Update the built-in documentation file for the
20016 new built-in functions.
20018 2017-06-29 Richard Biener <rguenther@suse.de>
20020 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20021 reduction chains to LOOP_VINFO_REDUCTIONS.
20022 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20023 SLP reductions after processing reduction chains.
20025 2017-06-29 Nathan Sidwell <nathan@acm.org>
20027 * builtins.c (fold_builtin_FUNCTION): Use
20028 lang_hooks.decl_printable_name.
20030 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20032 PR middle-end/81194
20033 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20034 with only one label.
20035 * stmt.c (expand_case): Assert NCASES is greater than one.
20037 2017-06-29 Richard Biener <rguenther@suse.de>
20039 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20041 (group_case_labels): Likewise.
20042 (find_taken_edge): Push sanity checking on val to workers...
20043 (find_taken_edge_cond_expr): ... here
20044 (find_taken_edge_switch_expr): ... and here, handle cases
20045 with just a default label.
20046 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20047 (group_case_labels): Likewise.
20048 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20049 group_case_labels does anything cleanup the CFG again.
20051 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20053 PR tree-optimization/81196
20054 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20055 exit condition comparing two IVs.
20057 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20059 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20060 profile to the dummy entry at the end of the list of architectures.
20061 * config/arm/arm-cpu-cdata.h: Regenerated.
20063 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20064 Michael Collison <michael.collison@arm.com>
20067 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20069 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20070 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20071 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20072 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20073 of shift when the shift amount is masked with constant equal to
20074 the size of the mode.
20075 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20078 2017-06-29 Martin Liska <mliska@suse.cz>
20080 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20082 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20085 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20088 (edge_growth_cache_entry::edge_growth_cache_entry): New
20090 (reset_edge_growth_cache): Update to use constructor.
20092 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20094 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20095 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20096 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20098 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20100 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20101 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20103 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20105 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20106 (*-linux-uclibc*): Add t-uclibc tmake_file.
20107 * config/t-musl: New.
20108 * config/t-uclibc: New.
20110 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20112 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20114 (gen_comm_data): Emit architectural setting of arch_prof.
20115 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20117 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20118 (armv8-m.base, armv8-m.main): Likewise.
20119 * arm-protos.h (arm_build_target): Add profile field.
20120 (arch_option): Likewise.
20121 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20123 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20124 arm_active_target.profile.
20126 2017-06-28 Richard Biener <rguenther@suse.de>
20128 PR middle-end/81227
20129 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20130 TYPE_OVERFLOW_WRAPS.
20131 * match.pd (negate_expr_p): Likewise.
20132 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20133 fold_build2, not fold_binary.
20135 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20137 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20138 Convert memory address to Pmode.
20139 (aarch64_print_operand): Assert MEM operands are always Pmode.
20141 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20144 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20145 Remove redundant if.
20146 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20147 * config/arm/aarch-common-protos.h
20148 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20149 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20151 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20154 * multiple_target.c (create_dispatcher_calls): Set the default
20155 clone to be static, not public.
20157 2017-06-28 Richard Biener <rguenther@suse.de>
20159 * tree-vect-loop.c (vectorizable_reduction): Move special
20160 cond reduction IV var creation ...
20161 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20162 parameter. Use STMT_VINFO_VECTYPE.
20163 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20166 2017-06-28 Martin Liska <mliska@suse.cz>
20169 * ipa-visibility.c (non_local_p): Handle visibility.
20171 2017-06-28 Martin Liska <mliska@suse.cz>
20174 * common.opt: Add IntegerRange to various options.
20175 * opt-functions.awk (integer_range_info): New function.
20176 * optc-gen.awk: Add integer_range_info to cl_options struct.
20177 * opts-common.c (decode_cmdline_option): Handle
20178 CL_ERR_INT_RANGE_ARG.
20179 (cmdline_handle_error): Likewise.
20180 * opts.c (print_filtered_help): Show valid interval in
20181 when --help is provided.
20182 * opts.h (struct cl_option): Add range_min and range_max fields.
20183 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20185 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20187 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20188 (x * C EQ/NE y * C): New transformation.
20190 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20192 * genmultilib (combination_space): Accept '+' in option names.
20194 2017-06-28 Martin Liska <mliska@suse.cz>
20197 * asan.c (instrument_derefs): Bail out inner references
20198 that are hard register variables.
20200 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20203 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20204 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20206 2017-06-28 Richard Biener <rguenther@suse.de>
20208 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20209 (vect_get_slp_defs): Adjust.
20210 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20212 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20214 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20215 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20216 (vectorizable_reduction): Adjust.
20217 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20219 (vect_get_slp_defs): Likewise.
20220 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20221 (vectorizable_bswap): Adjust.
20222 (vectorizable_call): Likewise.
20223 (vectorizable_conversion): Likewise.
20224 (vectorizable_assignment): Likewise.
20225 (vectorizable_shift): Likewise.
20226 (vectorizable_operation): Likewise.
20227 (vectorizable_store): Likewise.
20228 (vectorizable_condition): Likewise.
20229 (vectorizable_comparison): Likewise.
20231 2017-06-28 Michael Collison <michael.collison@arm.com>
20234 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20236 (arm_gen_movmemqi): Removed unused variable 'i'.
20237 Convert 'for' loop into 'while' loop.
20238 (arm_expand_prologue): Remove last unnecessary set of insn.
20239 (thumb_pop): Remove unused variable 'pushed_words'.
20240 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20242 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20244 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20245 * config/s390/s390-protos.h: Add prototype of
20246 s390_rel_address_ok_p.
20247 * config/s390/s390.c (s390_got_symbol): New function.
20248 (s390_rel_address_ok_p): New function.
20249 (legitimize_pic_address): Use s390_rel_address_ok_p.
20250 (s390_load_got): Use s390_got_symbol.
20251 (s390_option_override): Issue error if
20252 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20253 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20255 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20257 2017-06-27 Andrew Pinski <apinski@cavium.com>
20259 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20260 (X * copysign (1.0, X)): New pattern.
20261 (X * copysign (1.0, -X)): New pattern.
20262 (copysign (-1.0, CST)): New pattern.
20264 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20266 * genmultilib (combination_space): Remove variable.
20267 Validate reuse rules against regular expression for any sequence
20268 of multilib options in any order.
20270 2017-06-27 Michael Collison <michael.collison@arm.com>
20272 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20273 call aarch64_split_simd_combine.
20274 * (aarch64_combine_internal<mode>): Delete pattern.
20275 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20276 Allow register and subreg operands.
20278 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20280 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20281 specific need, just fallback on defaults.
20282 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20284 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20285 Olivier Hainque <hainque@adacore.com>
20287 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20289 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20290 targets. Pick a default if no particular attempt applied.
20291 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20294 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20296 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20297 (x86_64-wrs-vxworks7): Likewise.
20299 2017-06-27 Marek Polacek <polacek@redhat.com>
20302 * ubsan.c (instrument_null): Check get_base_address's result for null.
20304 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20306 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20308 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20310 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20311 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20312 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20313 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20314 New function types.
20315 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20316 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20317 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20318 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20319 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20320 BUILT_IN_FEUPDATEENV): New builtins.
20321 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20322 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20323 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20324 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20326 (builtin_structptr_types): Adjust size.
20327 * tree.c (builtin_structptr_types): Add four entries.
20329 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20330 Olivier Hainque <hainque@adacore.com>
20332 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20333 (TLS_SYM): New local macro, forcing reference to __tls__ on
20334 link command lines for VxWorks 7 RTPs, triggering initialization
20336 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20337 OS features TLS support, true for RTPs on VxWorks 7.
20338 * config/vxworks.c (vxworks_override_options): Setup emutls
20341 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20343 * predict.c (test_prediction_value_range): Use -1U instead of -1
20344 to avoid narrowing conversion warning.
20345 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20346 to avoid narrowing conversion warning.
20347 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20349 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20351 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20353 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20354 64bit configurations.
20355 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20356 (SIZE_TYPE): Likewise.
20357 * config/vxworks.c (vxworks_emutls_var_fields): Use
20358 long_unsigned_type_node instead of unsigned_type_node as the offset
20359 field type, which is "pointer" mode in emutls.c.
20361 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20364 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20366 PR middle-end/81207
20367 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20368 gimple_vuse copying separately from gimple_vdef copying.
20370 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20372 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20375 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20376 Olivier Hainque <hainque@adacore.com>
20378 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20379 for all vxworks7 targets.
20380 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20381 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20382 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20383 variations for VX6/VX7 and 32/64bits later on in ...
20384 (VXWORKS_LIB_SPEC): Leverage new macros.
20385 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20386 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20388 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20390 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20391 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20393 2017-06-26 Carl Love <cel@us.ibm.com>
20395 * config/rs6000/rs6000-c.c: Add support for built-in functions
20396 vector bool char vec_reve (vector bool char);
20397 vector signed char vec_reve (vector signed char);
20398 vector unsigned char vec_reve (vector unsigned char);
20399 vector bool int vec_reve (vector bool int);
20400 vector signed int vec_reve (vector signed int);
20401 vector unsigned int vec_reve (vector unsigned int);
20402 vector bool long long vec_reve (vector bool long long);
20403 vector signed long long vec_reve (vector signed long long);
20404 vector unsigned long long vec_reve (vector unsigned long long);
20405 vector bool short vec_reve (vector bool short);
20406 vector signed short vec_reve (vector signed short);
20407 vector double vec_reve (vector double);
20408 vector float vec_reve (vector float);
20409 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20410 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20411 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20412 (altivec_vreve): New pattern.
20413 * config/rs6000/altivec.h (vec_reve): New define.
20414 * doc/extend.texi (vec_rev): Update the built-in documentation file
20415 for the new built-in functions.
20417 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20419 PR tree-optimization/71815
20420 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20422 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20424 (slsr_process_phi): Likewise.
20425 (replace_uncond_cands_and_profitable_phis): Don't replace a
20426 multiply candidate with a stride of 1 (copy or cast).
20427 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20429 (lowest_cost_path): Likewise.
20430 (total_savings): Likewise.
20432 2017-06-26 Richard Biener <rguenther@suse.de>
20435 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20436 Use def_builtin_pure for all gather builtins.
20438 2017-06-26 Richard Biener <rguenther@suse.de>
20440 PR tree-optimization/81203
20441 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20442 non-dominating BBs.
20444 2017-06-26 Marek Polacek <polacek@redhat.com>
20447 * doc/invoke.texi: Document -Wmultistatement-macros.
20449 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20451 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20452 arm_neon_ok_no_float_abi effective target.
20454 2017-06-26 Richard Biener <rguenther@suse.de>
20456 PR tree-optimization/80928
20457 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20458 (copy_bbs): Set BB_DUPLICATED flag early.
20459 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20461 (execute_on_shrinking_pred): Likewise.
20462 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20463 BB_DUPLICATED blocks.
20464 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20465 iterate over all PHIs considering removal of *gsi.
20467 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20469 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20472 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20474 * config/rs6000/rs6000-string.c: (expand_block_clear,
20475 do_load_for_compare, select_block_compare_mode,
20476 compute_current_alignment, expand_block_compare,
20477 expand_strncmp_align_check, expand_strn_compare,
20478 expand_block_move, rs6000_output_load_multiple)
20479 Move functions related to string/block move/compare
20480 to a separate file.
20481 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20482 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20483 for this function which is now used in two files.
20484 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20485 * config.gcc: Add rs6000-string.o to extra_objs for
20486 targets powerpc*-*-* and rs6000*-*-*.
20488 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20491 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20492 32-bit, since indexed is not valid for DImode.
20493 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20494 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20495 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20496 (define_peephole2 for Altivec d-form store): Likewise.
20499 * multiple_target.c (create_dispatcher_calls): Only create the
20500 dispatcher call if the function is the default clone of a
20501 versioned function.
20503 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20505 PR middle-end/80902
20506 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20507 a call, force the call to not be a tail call.
20509 2017-06-23 Jeff Law <law@redhat.com>
20511 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20514 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20516 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20517 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20518 handling for early expansion of vector shifts (sl,sr,sra,rl).
20519 (builtin_function_type): Add vector shift right instructions
20520 to the unsigned argument list.
20522 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20524 rtl-optimizatoin/79286
20525 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20526 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20527 trap. PIC register plus a const unspec without offset can never trap.
20529 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20531 * tree.h (builtin_structptr_type): New type.
20532 (builtin_structptr_types): Declare new array.
20533 * tree.c (builtin_structptr_types): New array.
20534 (free_lang_data, build_common_tree_nodes): Use it.
20536 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20539 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20542 2017-06-22 Matt Turner <mattst88@gmail.com>
20544 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20545 Lake models to skylake case. Assume skylake for unknown
20546 models with clflushopt.
20548 2017-06-22 Jeff Law <law@redhat.com>
20550 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20551 frame sizes that do not satisfy aarch64_uimm12_shift.
20553 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20555 * profile-count.h (apply_probability,
20556 apply_scale, probability_in): Fix checks for zero.
20558 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20560 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20561 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20563 2016-06-22 Richard Biener <rguenther@suse.de>
20565 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20566 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20567 REDUC_MAX_EXPR support.
20568 (vectorizable_reduction): Likewise.
20569 (vect_create_epilog_for_reduction): Likewise.
20571 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20573 * match.pd (A / (1 << B) -> A >> B): New.
20574 * generic-match-head.c: Include optabs-tree.h.
20575 * gimple-match-head.c: Likewise.
20576 * optabs-tree.h (target_supports_op_p): New.
20577 * optabs-tree.c (target_supports_op_p): New.
20579 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20581 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20582 $gcc_cv_ld --help output.
20583 (gcc_cv_ld_demangle): Likewise.
20584 (gcc_cv_ld_eh_frame_hdr): Likewise.
20585 (gcc_cv_ld_pie): Likewise.
20586 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20587 (gcc_cv_ld_buildid): Likewise.
20588 (gcc_cv_ld_sysroot): Likewise.
20589 (ld_bndplt_support): Likewise.
20590 (ld_pushpopstate_support): Likewise.
20591 * configure: Regenerate.
20592 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20594 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20597 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20598 operands indexes to avoid gap between operands and match_dups.
20600 2017-06-21 Andrew Pinski <apinski@cavium.com>
20602 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20603 Increment Arith_shift and Arith_shift_reg by 1.
20604 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20606 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20607 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20608 (aarch64_strip_extend): Add new argument and test for it.
20609 (aarch64_cheap_mult_shift_p): New function.
20610 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20611 add a cost if it is true.
20612 Update calls to aarch64_strip_extend.
20613 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20615 2017-06-21 Andrew Pinski <apinski@cavium.com>
20617 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20619 (thunderxt88): Likewise.
20620 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20621 (thunderx_prefetch_tune): New variable.
20622 (thunderx2t99_prefetch_tune): Update for the correct values.
20623 (thunderxt88_tunings): New variable.
20624 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20625 generic_prefetch_tune.
20626 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20628 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20630 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20631 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20632 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20633 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20634 (aarch64_atomic_cas<mode>, GPI): Likewise.
20636 2017-06-21 Martin Liska <mliska@suse.cz>
20638 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20639 statements on cold and hot labels.
20640 * predict.c (tree_estimate_probability_bb): Remove the
20641 prediction from this place.
20643 2017-06-21 Martin Liska <mliska@suse.cz>
20645 PR tree-optimization/79489
20646 * gimplify.c (maybe_add_early_return_predict_stmt): New
20648 (gimplify_return_expr): Call the function.
20649 * predict.c (tree_estimate_probability_bb): Remove handling
20651 * predict.def: Update comment about early return predictor.
20652 * gimple-predict.h (is_gimple_predict): New function.
20653 * predict.def: Change default value of early return to 66.
20654 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20656 * passes.def: Put pass_strip_predict_hints to the beginning of
20659 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20661 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20662 FUNCTION_DECL declarations.
20663 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20665 (dwaf2out_decl): Likewise.
20666 * godump.c (go_early_global_decl): Skip call to the real debug hook
20667 for FUNCTION_DECL declarations.
20668 * passes.c (rest_of_decl_compilation): Skip call to the
20669 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20670 -fdump-go-spec is passed.
20672 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20674 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20675 Reorder for compactness.
20676 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20677 (def_builtin_pure, def_builtin_pure2): New functions.
20678 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20680 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20682 * match.pd (nop_convert): New predicate.
20683 ((A +- CST1) +- CST2): Allow some NOP conversions.
20685 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20688 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20689 with ctors/dtors if GOVD_SHARED is set.
20691 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20693 * config/aarch64/aarch64.md (movti_aarch64):
20694 Emit mov rather than orr.
20695 (movtf_aarch64): Likewise.
20696 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20697 Emit mov rather than orr.
20699 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20701 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20702 Swap alternatives, make integer dup more expensive.
20704 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20706 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20707 Return true for non-tls symbols.
20709 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20711 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20712 (cortex-a75): Likewise.
20713 (cortex-a75.cortex-a55): Likewise.
20714 * config/aarch64/aarch64-tune.md: Regenerate.
20715 * doc/invoke.texi (-mtune): Document new values for -mtune.
20717 2017-06-21 Tom de Vries <tom@codesourcery.com>
20719 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20720 stack_size feature.
20721 (Effective-Target Keywords, Other attributes): Suggest using
20722 dg-add-options stack_size feature to get stack limit in stack_size
20723 effective target documentation.
20725 2017-06-21 Julian Brown <julian@codesourcery.com>
20726 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20728 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20729 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20730 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20732 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20733 attribute type list for neon_multiply.
20734 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20735 attribute type list for neon_multiply.
20736 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20737 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20738 attribute type list for neon_multiply.
20739 * config/arm/types.md (crypto_pmull): Add.
20740 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20741 attribute type list.
20743 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20745 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20748 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20750 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20751 to make sure not to dereference a NULL cost_classes_ptr pointer.
20753 2017-06-20 Carl Love <cel@us.ibm.com>
20755 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20756 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20757 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20758 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20759 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20760 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20761 VMULOSW): New enum "unspec" values.
20762 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20763 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20764 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20765 altivec_vmulosw): New patterns.
20766 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20767 VMULOSW): Add definitions.
20769 2017-06-20 Julia Koval <julia.koval@intel.com>
20771 * config/i386/i386.c: Fix rounding expand for new pattern.
20772 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20774 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20776 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20777 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20779 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20781 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20784 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20786 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20787 architecture, then by implementer ID.
20788 * config/aarch64/aarch64-tune.md: Regenerate.
20790 2017-06-20 Richard Biener <rguenther@suse.de>
20792 PR middle-end/81097
20793 * fold-const.c (split_tree): Fold to type before negating.
20795 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20797 * diagnostic-show-locus.c
20798 (selftest::test_fixit_deletion_affecting_newline): New function.
20799 (selftest::diagnostic_show_locus_c_tests): Call it.
20801 2017-06-20 Andreas Schwab <schwab@suse.de>
20804 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20807 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20809 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20810 __ARM_FEATURE_COPROC according to support.
20812 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20814 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20815 Rewritten to avoid overflow for > 32-bit pointers.
20818 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20819 by removing enum keyword.
20820 (ubsan_type_descriptor): Likewise. Formatting fix.
20823 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20824 splitter): Require TARGET_SSE2 in the condition.
20826 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20829 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20830 for doing vector set of SFmode on ISA 3.0.
20831 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20832 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20834 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20835 SFmode value into a V4SF variable that was extracted from another
20836 V4SF variable without converting the element to double precision
20837 and back to single precision vector format.
20838 (vsx_insert_extract_v4sf_p9_2): Likewise.
20840 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20842 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20843 in UWHI to avoid undefined overflow.
20846 * ubsan.h (enum ubsan_encode_value_phase): New.
20847 (ubsan_encode_value): Change second argument to
20848 enum ubsan_encode_value_phase with default value of
20849 UBSAN_ENCODE_VALUE_GENERIC.
20850 * ubsan.c (ubsan_encode_value): Change second argument to
20851 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20852 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20853 create_tmp_var_raw instead of create_tmp_var and use a
20855 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20856 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20857 ubsan_encode_value callers.
20860 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20861 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20862 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20864 2017-06-19 Richard Biener <rguenther@suse.de>
20866 PR middle-end/81118
20867 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20868 estimates if we changed anything.
20870 2017-06-19 Richard Biener <rguenther@suse.de>
20872 PR tree-optimization/80887
20873 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20874 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20875 simplified lookups, then reset mprts_hook.
20876 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20878 (try_to_simplify): Likewise.
20880 2017-06-19 Martin Liska <mliska@suse.cz>
20883 * gimplify.c (gimplify_switch_expr):
20884 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20886 2017-06-19 Martin Liska <mliska@suse.cz>
20888 * doc/install.texi: Document that PGO runs in 4 stages.
20890 2017-06-19 Martin Liska <mliska@suse.cz>
20893 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20894 to dispatcher function name.
20895 * multiple_target.c (replace_function_decl): New function.
20896 (create_dispatcher_calls): Redirect both edges and references.
20898 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20900 * profile-count.c (profile_count::dump): Dump quality.
20901 (profile_count::differs_from_p): Update for unsigned val.
20902 * profile-count.h (profile_count_quality): New enum.
20903 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20905 2017-06-19 Richard Biener <rguenther@suse.de>
20907 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20908 struct function as arg.
20909 (estimate_numbers_of_iterations): Export overload with loop arg.
20910 (free_numbers_of_iterations_estimates_loop): Use an overload of
20911 free_numbers_of_iterations_estimates instead.
20912 * tree-cfg.c (remove_bb): Adjust.
20913 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20914 * tree-parloops.c (gen_parallel_loop): Likewise.
20915 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20917 (tree_unroll_loops_completely): Likewise.
20918 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20919 Use an overload instead and export.
20920 (estimated_loop_iterations): Adjust.
20921 (max_loop_iterations): Likewise.
20922 (likely_max_loop_iterations): Likewise.
20923 (estimate_numbers_of_iterations): Take struct function as arg
20925 (loop_exits_before_overflow): Adjust.
20926 (free_numbers_of_iterations_estimates_loop): Use an overload.
20927 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20928 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20930 2017-06-19 Richard Biener <rguenther@suse.de>
20933 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20934 RANGE_EXPR conservatively.
20936 2017-06-16 Carl Love <cel@us.ibm.com>
20938 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20939 definitions for vec_float, vec_float2, vec_floato,
20940 vec_floate built-ins.
20941 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20942 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20944 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20945 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20946 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20947 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20948 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20949 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20950 vec_floato): Add builtin defines.
20951 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20952 Update the built-in documentation file for the new built-in
20955 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20957 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20958 (mthumb): Mark as the negative of -marm.
20960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20962 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20964 (ARM Options, -mtune): Document that this accepts the same
20965 extension options as -mcpu.
20966 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20970 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20971 permitted extensions.
20973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20975 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20976 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20977 (armv8-m.main): Add option +nodsp.
20978 * config/arm/arm-cpu-cdata.h: Regenerated.
20980 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20982 * config/arm/t-fuchsia: New file.
20983 * config.gcc (arm*-*-fuchsia*): Use it.
20985 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20987 * config/arm/t-symbian: Rewrite for new option infrastructure.
20989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20991 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20992 (MULTILIB_REQUIRED): Likewise.
20994 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20996 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20997 (MULTILIB_RESUE): Likewise.
20998 (MULTILIB_MATCHES): Likewise.
20999 (MULTLIB_REQUIRED): Likewise.
21001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21003 * config/arm/t-rtems: Rewrite for new option framework.
21005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21007 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21008 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21009 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21010 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21011 * config/arm/t-multilib: ... here.
21012 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21013 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21014 armv7-a and armv8*-a when A-profile libraries have not been built.
21015 * config/arm/t-rmprofile: Rewrite.
21017 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21019 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21020 with a backslash. Remove the backslash after substituting unescaped
21022 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21026 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21027 the driver to pass through the default setting of -mfloat-abi.
21028 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21030 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21031 (all_feat_combs): New rule.
21032 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21034 * config/arm/t-aprofile: Rewrite.
21036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21038 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21039 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21040 fpu is not specified by the user/command-line.
21041 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21042 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21043 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21044 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21045 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21046 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21048 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21050 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21051 * config/arm/t-arm-elf: Rewritten.
21053 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21055 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21056 have some floating-point instructions.
21057 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21058 (TARGET_MAYBE_HARD_FLOAT): New macro.
21059 * config/arm/arm-builtins.c (arm_init_builtins): Use
21060 TARGET_MAYBE_HARD_FLOAT.
21061 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21063 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21065 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21066 (configargs.h): Include it.
21067 (arm_print_hint_for_fpu_option): New function.
21068 (arm_parse_fpu_option): New function.
21069 (candidate_extension): New class.
21070 (arm_canon_for_multilib): New function.
21071 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21072 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21073 (ARCH_CANONICAL_SPECS): New macro.
21074 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21076 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21078 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21079 are set after handling multilib fragments. Set target_cpu_default2
21082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21084 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21086 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21087 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21089 (check_arch): Likewise.
21090 * config/arm/arm.c (arm_configure_build_target): Handle
21091 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21092 options in the default.
21094 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21096 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21097 when an option is an alias of another.
21098 * config/arm/parsecpu.awk (optalias): New parser token.
21099 (gen_comm_data): Mark non-alias options as such. Emit entries
21100 for extension aliases.
21101 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21102 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21103 (armv6kz, armv6zk, armv6t2): Likewise.
21104 (armv7): Make vfpv3-d16 an alias.
21105 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21107 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21108 Sort in canonical order.
21109 (armv8-a): Sort in canonical order.
21110 (armv8.1-a, armv8.2-a): Likewise.
21111 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21113 (cortex-a9): Sort in canonical order.
21114 * config/arm/arm.c (selftests.h): Include it.
21115 (arm_test_cpu_arch_data): New function.
21116 (arm_run_self_tests): New function.
21117 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21118 (targetm): Move declaration to the end of the file.
21119 * arm-cpu-cdata.h: Regenerated.
21121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21123 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21124 call to target_mode_check describing the type of option passed.
21125 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21126 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21127 arm_parse_cpu_option_name to match parameters against list of
21129 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21130 arm_arch_core_flags data structure.
21131 * config/arm/arm-cpu_cdata.h: Regenerated.
21133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21135 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21137 (arm_print_hint_for_cpu_option): Likewise.
21138 (arm_print_hint_for_arch_option): Likewise.
21139 (arm_parse_cpu_option_name): Likewise.
21140 (arm_parse_arch_option_name): Likewise.
21141 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21142 of entries in the all_fpus list.
21143 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21144 (arm_parse_cpu_option_name): Declare.
21145 (arm_parse_arch_option_name): Declare.
21146 (arm_parse_option_features): Declare.
21147 (arm_intialize_isa): Declare.
21148 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21150 (gen_comm_data): ... here. Make definitions non-static.
21151 * config/arm/arm-cpu-data.h: Regenerated.
21152 * config/arm/arm-cpu-cdata.h: Regenerated.
21154 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21156 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21157 (cpu_arch_extension): New structure.
21158 (cpu_arch_option, arch_option, cpu_option): New structures.
21159 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21160 architecture types.
21161 (gen_data): Generate new format data tables.
21162 * config/arm/arm.c (cpu_tune): New structure.
21163 (cpu_option, processors): Delete.
21164 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21165 (arm_print_hint_for_cpu_option): ... this and ...
21166 (arm_print_hint_for_arch_option): ... this.
21167 (arm_parse_arch_cpu_name): Delete. Replace with ...
21168 (arm_parse_cpu_option_name): ... this and ...
21169 (arm_parse_arch_option_name): ... this.
21170 (arm_unrecognized_feature): Change type of target parameter to
21172 (arm_parse_arch_cpu_features): Delete. Replace with ...
21173 (arm_parse_option_features): ... this.
21174 (arm_configure_build_target): Rework to use new configuration data
21176 (arm_print_tune_info): Rework for new configuration data tables.
21177 * config/arm/arm-cpu-data.h: Regenerated.
21178 * config/arm/arm-cpu.h: Regenerated.
21180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21182 * Makefile.in (OBJS): Move sbitmap.o from here ...
21183 (OBJS-libcommon): ... to here.
21185 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21187 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21188 (ISA_ALL_CRYPTO): New macro.
21189 (ISA_ALL_SIMD): New macro
21190 (ISA_ALL_FP): New macro.
21191 * config/arm/arm.c (fpu_bitlist): Update initializer.
21192 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21194 (arm9e): Add fpu. Add option for nofp
21195 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21196 (arm926ej-s, arm1026ej-s): Likewise.
21197 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21198 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21199 neon-fp16, neon-vfpv4, nofp and nosimd.
21200 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21201 (cortex-a8): Add fpu. Add option for nofp.
21202 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21203 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21204 (cortex-r4f): Add fpu.
21205 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21206 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21208 (cortex-r8): Likewise.
21209 (cortex-m4): Add fpu. Add option for nofp.
21210 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21211 (cortex-a17.cortex-a7): Likewise.
21212 (cortex-a32): Add fpu. Add options for crypto and nofp.
21213 (cortex-a35, cortex-a53): Likewise.
21214 (cortex-a57): Add fpu. Add option for crypto.
21215 (cortex-a72, cortex-a73): Likewise.
21216 (exynos-m1): Likewise.
21217 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21218 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21219 (cortex-m33): Add fpu. Add option for nofp.
21220 * config/arm/arm-cpu-cdata.h: Regenerated
21221 * config/arm/arm-cpu-data.h: Regenerated.
21223 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21225 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21226 (armv5te, armv5tej): Likewise.
21227 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21228 (armv7): Add options fp and vfpv3-d16.
21229 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21230 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21232 (armv7ve): Likewise.
21233 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21234 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21235 (armv8-a): Add nocrypto option.
21236 (armv8.1-a, armv8.2-a): Likewise.
21237 (armv8-m.main): add options fp, fp.dp and nofp.
21239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21241 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21243 (armv8-a+crc): Delete.
21244 (armv8.1-a): Add options simd, crypto and nofp.
21245 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21246 (armv8.2-a+fp16): Delete.
21247 (armv8-m.main): Add option dsp.
21248 (armv8-m.main+dsp): Delete.
21249 (cortex-a8): Add fpu. Add nofp option.
21250 (cortex-a9): Add fpu. Add nofp and nosimd options.
21251 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21252 link to main cpu and architecture data structures.
21253 (gen_comm_data): Only put isa attributes from the main architecture
21255 (option): New statement for architecture and CPU entries.
21256 * arm.c (struct cpu_option): New structure.
21257 (struct processors): Add entry for options.
21258 (arm_unrecognized_feature): New function.
21259 (arm_parse_arch_cpu_name): Ignore any characters after the first
21261 (arm_parse_arch_cpu_feature): New function.
21262 (arm_configure_build_target): Separate out any CPU and architecture
21263 features and parse separately. Don't error out if -mfpu=auto is
21264 used with only an architecture string.
21265 (arm_print_asm_arch_directives): New function.
21266 (arm_file_start): Call it.
21267 * config/arm/arm-cpu-cdata.h: Regenerated.
21268 * config/arm/arm-cpu-data.h: Likewise.
21269 * config/arm/arm-tables.opt: Likewise.
21271 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21273 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21274 assembler when it is not -mfpu=auto.
21276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21278 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21279 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21280 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21281 (ASM_CPU_SPEC): Rewrite.
21282 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21283 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21284 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21285 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21286 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21287 copied string is NUL-terminated. Also strip any characters prefixed
21289 (arm_rewrite_selected_arch): New function.
21290 (arm_rewrite_march): New function.
21292 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21294 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21295 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21296 (march, mcpu, mtune): Convert to string-based options.
21297 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21298 (arm_parse_arch_cpu_name): New function.
21299 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21300 identify selected architecture or CPU.
21301 (arm_option_save): New function.
21302 (TARGET_OPTION_SAVE): Redefine.
21303 (arm_option_restore): Restore string options.
21304 (arm_option_print): Print string options.
21306 2017-06-16 Martin Sebor <msebor@redhat.com>
21308 PR tree-optimization/80933
21309 PR tree-optimization/80934
21310 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21311 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21312 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21313 (gimple_fold_builtin): Call them.
21315 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21317 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21318 as unlikely; update profile.
21320 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21322 * predict.c (force_edge_cold): Handle declaring edges impossible
21325 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21327 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21329 (try_unroll_loop_completely): Fix reporting.
21331 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21333 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21335 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21338 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21339 if given a non-constant argument for an intrinsic which requires a
21342 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21344 * profile.c (compare_freqs): New function.
21345 (branch_prob): Sort edge list.
21346 (find_spanning_tree): Assume that the list is priority sorted.
21348 2017-06-16 Richard Biener <rguenther@suse.de>
21350 PR tree-optimization/81090
21351 * passes.def (pass_record_bounds): Remove.
21352 * tree-pass.h (make_pass_record_bounds): Likewise.
21353 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21354 make_pass_record_bounds): Likewise.
21355 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21356 not free niter estimates at the beginning but at the end.
21357 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21359 2017-06-16 Richard Biener <rguenther@suse.de>
21361 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21362 initializer to workaround ICE in host GCC 4.8.
21364 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21366 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21368 (clone_inlined_nodes): Update.
21370 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21372 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21373 prefetch settings, and enable prefetching by default at -O3.
21375 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21377 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21378 Set flag_prefetch_loop_arrays according to tuning data.
21380 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21382 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21383 New tune structure.
21384 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21385 [Unrelated to main purpose of the patch] Place the pointer field last
21386 to enable type checking errors when tune structure are wrongly merged.
21387 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21388 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21389 (thunderx2t99_prefetch_tune): New tune constants.
21390 (tune_params *_tunings): Update all tunings (no functional change).
21391 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21392 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21393 from tunings structures.
21395 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21398 * ubsan.c (instrument_null): Add T argument, use it instead
21399 of computing it based on IS_LHS.
21400 (instrument_object_size): Likewise.
21401 (pass_ubsan::execute): Adjust instrument_null and
21402 instrument_object_size callers to pass gimple_get_lhs or
21403 gimple_assign_rhs1 result to it. Use instrument_null instead of
21404 calling get_base_address and instrument_mem_ref. Handle
21405 aggregate call arguments for object-size sanitization.
21407 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21409 PR tree-optimization/81089
21410 * tree-vrp.c (is_masked_range_test): Validate operands of
21413 2017-06-15 Martin Sebor <msebor@redhat.com>
21416 * dumpfile.c (dump_register): Avoid calling memset to initialize
21417 a class with a default ctor.
21418 * gcc.c (struct compiler): Remove const qualification.
21419 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21420 * hash-table.h: Ditto.
21421 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21423 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21424 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21426 * params.h (struct param_info): Make struct members non-const.
21427 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21428 with default initialization.
21429 * vec.h (vec_copy_construct, vec_default_construct): New helper
21431 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21432 with vec_copy_construct.
21433 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21434 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21435 * doc/invoke.texi (-Wclass-memaccess): Document.
21437 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21439 * emit-rtl.h (is_leaf): Update comment about local
21440 register allocator.
21442 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21445 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21446 for a variable to have a section before checking if the section has a
21448 Set section to.persistent if persistent attribute is set.
21449 Warn if .persistent attribute is used on an automatic variable.
21451 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21453 PR rtl-optimization/80474
21454 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21456 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21458 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21461 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21463 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21464 (casesi): Emit bounds checking as RTL.
21465 (casesi_internal_mips16_<mode>): Remove bounds checking.
21467 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21469 * config/xtensa/xtensa.c (xtensa_option_override): Append
21470 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21471 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21472 xtensa_doloop_hooks): Define unconditionally.
21473 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21475 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21476 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21477 for it in xtensa_option_override.
21478 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21479 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21481 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21483 * doc/cppopts.texi: Document '-' special value to -MF.
21485 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21487 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21488 (cortex_a53_fconst): Likewise.
21489 (cortex_a53_fpmul): Likewise.
21490 (cortex_a53_f_load_64): Likewise.
21491 (cortex_a53_f_load_many): Likewise.
21492 (cortex_a53_advsimd_alu): Likewise.
21493 (cortex_a53_advsimd_alu_q): Likewise.
21494 (cortex_a53_advsimd_mul): Likewise.
21495 (cortex_a53_advsimd_mul_q): Likewise.
21496 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21497 Add missing fmul, r2f_cvt and fconst cases.
21499 2017-06-14 Richard Biener <rguenther@suse.de>
21501 PR middle-end/81088
21502 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21504 (fold_binary_loc): When associating do not treat pre-existing
21505 TREE_OVERFLOW on literal constants as a reason to allow
21506 TREE_OVERFLOW on associated literal constants.
21508 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21510 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21511 (MASK_FEATURES): New macro.
21512 * config/sparc/sparc.c (sparc_option_override): Remove the special
21513 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21515 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21517 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21518 a division of 0 if non-call exceptions are enabled.
21520 2017-06-14 Andrew Pinski <apinski@cavium.com>
21521 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21524 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21525 Improve vector initialization code gen for only variable case.
21527 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21529 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21531 2017-06-14 Richard Biener <rguenther@suse.de>
21533 PR tree-optimization/81083
21534 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21537 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21539 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21540 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21541 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21542 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21543 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21544 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21546 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21548 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21549 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21551 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21553 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21555 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21557 * config/rs6000/t-rtems: Don't handle SPE.
21559 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21561 * config/rs6000/t-linux: Don't handle SPE.
21563 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21565 * config/rs6000/eabispe.h: Delete file.
21567 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21569 * config/rs6000/t-spe: Delete file.
21571 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21573 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21574 (rs6000_legitimate_offset_address_p): Return false for anything in
21575 V2SImode or V2SFmode.
21577 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21579 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21580 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21581 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21583 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21584 (rs6000_legitimate_offset_address_p): Ditto.
21585 (rs6000_emit_move): Ditto.
21586 (rs6000_init_builtins): Remove V4HI_type_node.
21588 2017-06-13 Martin Liska <mliska@suse.cz>
21591 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21592 (gate_asan): Likewise.
21593 * asan.h (asan_no_sanitize_address_p): Remove the function.
21594 (sanitize_flags_p): New function.
21595 * builtins.def: Fix coding style.
21596 * common.opt: Use renamed enum value.
21597 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21598 * doc/extend.texi: Document no_sanitize attribute.
21599 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21600 to SANITIZE_UNDEFINED_NONDEFAULT.
21601 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21602 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21603 Use sanitize_flags_p.
21604 * gimplify.c (gimplify_function_tree): Likewise.
21605 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21606 * opts.c (parse_no_sanitize_attribute): New function.
21607 (common_handle_option): Use renamed enum value.
21608 * opts.h (parse_no_sanitize_attribute): Declare.
21609 * tree.c (sanitize_flags_p): New function.
21610 * tree.h: Declared here.
21611 * tsan.c: Use sanitize_flags_p.
21612 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21613 (instrument_mem_ref): Likewise.
21614 (instrument_bool_enum_load): Likewise.
21615 (do_ubsan_in_current_function): Remove the function.
21616 (pass_ubsan::execute): Use sanitize_flags_p.
21617 * ubsan.h: Remove do_ubsan_in_current_function
21618 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21619 (dump_function_to_file): Use it here.
21621 2017-06-13 Martin Jambor <mjambor@suse.cz>
21623 PR tree-optimization/80803
21624 PR tree-optimization/81063
21625 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21626 (propagate_subaccesses_across_link): Enqueue subtree whenever
21627 necessary instead of relying on the caller.
21629 2017-06-13 Martin Jambor <mjambor@suse.cz>
21631 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21632 that have a first_link.
21633 (sort_and_splice_var_accesses): Do not check first_link before
21635 (subtree_mark_written_and_enqueue): Likewise.
21636 (propagate_all_subaccesses): Likewise and do not stop at first
21637 parent with a first_link.
21639 2017-06-13 Martin Jambor <mjambor@suse.cz>
21641 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21644 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21646 * match.pd: New pattern.
21648 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21650 * tree-vrp.c (is_masked_range_test): New function.
21651 (register_edge_assert_for): Determine ranges for
21654 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21656 PR tree-optimization/67328
21657 * fold-const.c (maskable_range_p): New function.
21658 (build_range_check): Generate bittests if possible.
21660 2017-06-13 Martin Liska <mliska@suse.cz>
21662 * gimple-pretty-print.c (dump_probability): Add new argument.
21663 (dump_edge_probability): Dump both probability and count.
21664 (dump_gimple_label): Likewise.
21665 (dump_gimple_bb_header): Likewise.
21667 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21670 * config/avr/avr-devices.c: Fix indentation.
21671 * config/avr/gen-avr-mmcu-specs.c: Dito.
21673 2017-06-13 Richard Biener <rguenther@suse.de>
21675 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21676 instead get vector type from stmt_info.
21677 (vectorizable_reduction): Adjust. Remove dead code.
21679 2017-06-13 Richard Biener <rguenther@suse.de>
21681 PR middle-end/81065
21682 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21683 case of C * (x * C2 + C3).
21684 (fold_addr_of_array_ref_difference): Properly fold index difference.
21686 2017-06-12 David S. Miller <davem@davemloft.net>
21689 * config/sparc/sparc.md (return expander): Emit frame blockage if
21690 function uses alloca.
21692 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21694 * combine.c (make_field_assignment): Check len rather than the mode
21695 precision when calling force_to_mode.
21697 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21699 Support multilibs and devices that see flash in RAM address range.
21702 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21703 (avr_mcu_t) <flash_pm_offset>: New field.
21704 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21705 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21706 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21707 (AVR_TINY_PM_OFFSET): Remove macro.
21708 * config/avr/avr.opt (-mshort-calls): New option.
21709 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21710 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21711 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21712 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21713 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21714 instead of avr_arch->have_jmp_call.
21715 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21716 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21717 avr_arch->flash_pm_offset to define.
21718 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21719 new field flash_pm_offset. Add entry for avrxmega3.
21720 (avr_texinfo): Add entry for avrxmega3.
21721 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21722 attiny212, attiny214,
21723 attiny412, attiny414, attiny416, attiny417,
21724 attiny814, attiny816, attiny817,
21725 attiny1614, attiny1616, attiny1617,
21726 attiny3214, attiny3216, attiny3217.
21727 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21728 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21729 (avr_print_operand_address) [AVR_TINY]: Same.
21730 (avr_asm_init_sections) <readonly_data_section>: Only patch
21731 callback if avr_arch->flash_pm_offset = 0.
21732 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21733 for rodata if avr_arch->flash_pm_offset != 0.
21734 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21735 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21736 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21737 (m_options): Append opt_rcall.
21738 (m_dirnames): Append dir_rcall.
21739 * config/avr/t-multilib: Regenerate.
21741 * configure.ac [target=avr]: Check whether avrxmega3 default
21742 linker description file works as needed.
21743 * configure: Regenerate.
21744 * doc/avr-mmcu.texi: Regenerate.
21745 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21746 <__AVR_ARCH__>: Document avrxmega3 and 103.
21747 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21748 <__AVR_SHORT_CALLS__>: Document it.
21749 <__AVR_PM_BASE_ADDRESS__>: Document it.
21750 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21751 (AVR Variable Attributes) <progmem>: Document this is
21752 not needed for avrxmega3.
21753 (AVR Named Address Spaces) <__flash>: Dito.
21755 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21757 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21759 2017-06-12 Doug Rupp <rupp@adacore.com>
21761 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21762 Append vxworks-stdint.h to the tm_file list.
21763 * config/vxworks-stdint.h: New file.
21765 2017-06-12 Martin Liska <mliska@suse.cz>
21767 PR tree-optimization/81041
21768 * tree-profile.c (gimple_gen_ic_func_profiler):
21769 Create an extra BB in profile-generate
21770 (gimple_gen_time_profiler): Likewise.
21772 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21774 PR tree-optimization/81003
21775 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21776 (update_range_test): Use it instead of force_gimple_operand_gsi.
21778 2017-06-12 Richard Biener <rguenther@suse.de>
21780 PR tree-optimization/81053
21781 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21782 with backedge value not defined in loop. Simplify def stmt
21785 2017-06-11 Tom de Vries <tom@codesourcery.com>
21788 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21790 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21791 nvptx_cannot_force_const_mem.
21793 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21795 * opts.c (finish_options): Move test for flag_split_stack after
21796 it has been initialized.
21798 2017-06-11 Jason Merrill <jason@redhat.com>
21800 * tree.h (id_equal): New.
21801 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21802 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21803 instead of strcmp of IDENTIFIER_POINTER.
21805 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21807 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21808 (mark_all_inlined_calls_cdtor): Fix formating.
21809 (inline_transform): Rescale profile before inlining.
21811 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21813 * cgraph.h (cgraph_edge::clone): Update prototype.
21814 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21815 (cgraph_node::create_clone): Update.
21816 (cgraph_node::create_version_clone): Update.
21817 * tree-inline.c (copy_bb): Update.
21818 (expand_call_inline): Update.
21820 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21822 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21823 factored out from ...
21824 (rs6000_emit_prologue): ... here.
21826 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21828 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21829 factored out from ...
21830 (rs6000_emit_prologue): ... here.
21832 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21834 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21836 (handle_missing_profiles): Fix computation of tp_first_run.
21837 (counts_to_freqs): Do not touch freqs when count is 0.
21839 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21841 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21844 2017-06-10 Tom de Vries <tom@codesourcery.com>
21846 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21847 attributes): Document signal effective target.
21849 2017-06-10 Tom de Vries <tom@codesourcery.com>
21851 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21852 Document effective target stack_size.
21854 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21856 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21857 to the edit_context if they can be auto-applied.
21859 2017-06-9 Ian Lance Taylor <iant@golang.org>
21861 * opts.c (finish_options): If -fsplit-stack, disable implicit
21862 -forder-blocks-and-partition.
21863 * doc/invoke.texi (Optimize Options): Document that when using
21864 -fsplit-stack -forder-blocks-and-partition is not implicitly
21867 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21869 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21870 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21871 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21872 * builtins.def (abort, trap, unreachable): Declare cold.
21873 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21874 * tree-core.h (ECF_COLD): New.
21875 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21876 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21878 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21880 * predict.c (unlikely_executed_stmt_p): Cleanup.
21882 2017-06-09 Richard Biener <rguenther@suse.de>
21884 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21885 model if the ref is always written to.
21887 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21889 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21891 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21893 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21896 2017-06-09 Tom de Vries <tom@codesourcery.com>
21899 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21900 "target cannot support label values" when encountering LABEL_REF.
21902 2017-06-09 Martin Liska <mliska@suse.cz>
21904 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21905 (gimple_gen_ic_func_profiler): Emit direct comparison
21906 of __gcov_indirect_call_callee with NULL.
21907 (gimple_gen_time_profiler): Change probability from
21908 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21910 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21912 * profile.c (edge_gcov_counts): Turn to pointer.
21913 (compute_branch_probabilities, compute_branch_probabilities): Update.
21914 (branch_prob): Do not clear edge_gcov_count.
21915 * profile.h (edge_gcov_counts): Turn to pointer.
21916 (edge_gcov_count): Update.
21918 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21920 * gimple.h (gimple_check_failed): Mark cold.
21922 2017-06-09 Richard Biener <rguenther@suse.de>
21924 PR tree-optimization/66623
21925 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21926 refactor check_reduction into two parts, properly computing
21927 whether we have to check reduction validity for outer loop
21930 2017-06-09 Richard Biener <rguenther@suse.de>
21932 PR tree-optimization/79483
21933 * graphite-scop-detection.c (order): New global.
21934 (get_order): Compute bb to order mapping that satisfies code
21935 generation constraints.
21936 (cmp_pbbs): New helper.
21937 (build_scops): Start domwalk at entry block, sort generated
21940 2017-06-09 Richard Biener <rguenther@suse.de>
21942 PR middle-end/81007
21943 * ipa-polymorphic-call.c
21944 (ipa_polymorphic_call_context::restrict_to_inner_class):
21945 Skip FIELD_DECLs with error_mark_node type.
21946 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21949 2017-06-09 Martin Liska <mliska@suse.cz>
21951 * predict.c (struct branch_predictor): New struct.
21952 (test_prediction_value_range): New test.
21953 (predict_c_tests): New function.
21954 * selftest-run-tests.c (selftest::run_tests): Run the function.
21955 * selftest.h: Declare new tests.
21957 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21960 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21961 gen_add3_insn did not fail.
21962 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21963 r0, construct that number in a temporary reg and add that reg to r0.
21964 If asked to put the result in r0 as well, fail.
21966 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21968 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21969 for early expansion of vec_eqv.
21971 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21973 PR middle-end/81005
21974 * ubsan.c (instrument_null): Avoid pointless code temporary.
21975 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21977 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21981 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21984 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21987 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21989 * predict.c (unlikely_executed_edge_p): Move ahead.
21990 (probably_never_executed_edge_p): Use it.
21992 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21994 PR middle-end/79988
21995 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21996 gimple_call_builtin_p call.
21998 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22000 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22001 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22002 rtl_check_failed_type2, rtl_check_failed_code1,
22003 rtl_check_failed_code2, rtl_check_failed_code_mode,
22004 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22005 rtvec_check_failed_bounds, rtl_check_failed_flag,
22006 _fatal_insn_not_found, _fatal_insn): Likewise.
22007 * tree.h (tree_contains_struct_check_failed,
22008 tree_check_failed, tree_not_check_failed,
22009 tree_class_check_failed, tree_range_check_failed,
22010 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22011 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22012 tree_operand_check_failed, omp_clause_check_failed,
22013 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22016 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22018 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22019 flag_branch_probabilities.
22020 * ipa-inline.c (edge_badness): Likewise.
22021 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22022 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22023 * predict.c (maybe_hot_frequency_p): Likewise.
22024 (probably_never_executed): Likewise.
22025 * sched-ebb.c (schedule_ebbs): Likewise.
22026 * sched-rgn.c (find_single_block_region): Likewise.
22027 * tracer.c (tail_duplicate): Likewise.
22029 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22031 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22032 longer requires x_flag_profile_use.
22034 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22036 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22037 instead of flag_reorder_blocks_and_partition.
22038 * dbxout.c (dbxout_function_end): Likewise.
22039 * dwarf2out.c (gen_subprogram_die): Likewise.
22040 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22041 * hw-doloop.c (reorg_loops): Likewise.
22042 * varasm.c (assemble_start_function,
22043 assemble_end_function): Likewise.
22044 (decide_function_section): Do not check for
22045 flag_reorder_blocks_and_partition.
22047 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22049 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22051 (chkp_get_hard_register_fake_addr_expr): Ditto.
22052 (chkp_build_addr_expr): Add check for hard reg case.
22053 (chkp_parse_array_and_component_ref): Ditto.
22054 (chkp_find_bounds_1): Ditto.
22055 (chkp_process_stmt): Don't generate bounds store for
22058 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22060 * predict.c (maybe_hot_bb_p): Do not check profile status.
22061 (maybe_hot_edge_p): Likewise.
22062 (probably_never_executed): Check for zero counts even if profile
22064 (unlikely_executed_edge_p): New function.
22065 (unlikely_executed_stmt_p): New function.
22066 (unlikely_executed_bb_p): New function.
22067 (set_even_probabilities): Use unlikely predicates.
22068 (combine_predictions_for_bb): Likewise.
22069 (predict_paths_for_bb): Likewise.
22070 (predict_paths_leading_to_edge): Likewise.
22071 (determine_unlikely_bbs): New function.
22072 (estimate_bb_frequencies): Use it.
22073 (compute_function_frequency): Use zero counts even if profile is
22075 * profile-count.h: Fix typo.
22077 2017-08-08 Julia Koval <julia.koval@intel.com>
22079 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22080 _mm512_mask_cvtsepi16_storeu_epi8,
22081 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22082 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22083 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22084 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22085 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22086 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22087 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22088 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22089 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22090 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22091 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22092 __builtin_ia32_pmovuswb256mem_mask,
22093 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22094 __builtin_ia32_pmovwb512mem_mask): New builtins.
22096 2017-08-08 Julia Koval <julia.koval@intel.com>
22098 PR target/73350,80862
22099 * config/i386/subst.md (round): Fix round pattern.
22100 * config/i386/i386.c (ix86_erase_embedded_rounding):
22101 Fix erasing rounding for the fixed pattern.
22103 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22105 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22107 2017-06-08 Martin Liska <mliska@suse.cz>
22109 PR gcov-profile/80911
22110 * gcov.c (block_info::block_info): New constructor.
22112 2017-06-07 Carl Love <cel@us.ibm.com>
22114 * config/rs6000/rs6000-c: The return type of the following
22115 built-in functions was implemented as int not long long. Fix sign
22116 of return value for the unsigned version of vec_mulo and vec_mule.
22117 vector unsigned long long vec_bperm (vector unsigned long long,
22118 vector unsigned char)
22119 vector signed long long vec_mule (vector signed int,
22121 vector unsigned long long vec_mule (vector unsigned int,
22122 vector unsigned int)
22123 vector signed long long vec_mulo (vector signed int,
22125 vector unsigned long long vec_mulo (vector unsigned int,
22126 vector unsigned int)
22127 * doc/extend.texi: Fix the documentation for the built-in
22130 2017-06-07 Carl Love <cel@us.ibm.com>
22133 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22136 2017-06-07 Carl Love <cel@us.ibm.com>
22138 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22139 support, Generate doublehv for signed int/float for BE case only.
22141 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22143 * doc/invoke.texi (mcx16): Rewrite.
22145 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22147 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22148 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22149 *mov<mode>_softfloat, and an anonymous splitter): Use
22150 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22152 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22154 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22156 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22157 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22158 (rs6000_debug_reg_global): Adjust.
22159 (rs6000_init_hard_regno_mode_ok): Adjust.
22160 (rs6000_dbx_register_number): Adjust.
22161 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22162 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22163 Remove SPE_ACC and SPEFSCR.
22164 (REG_ALLOC_ORDER): Ditto.
22165 (FRAME_POINTER_REGNUM): Change to 111.
22166 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22167 (REG_CLASS_NAMES): Ditto.
22168 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22169 (REGISTER_NAMES): Ditto.
22170 (ADDITIONAL_REG_NAMES): Ditto.
22171 (rs6000_reg_names): Ditto.
22172 * config/rs6000/rs6000.md: Renumber some register number
22175 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22177 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22179 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22180 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22182 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22183 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22184 Delete the SPE high registers.
22185 (REG_ALLOC_ORDER): Ditto.
22186 (enum reg_class): Remove SPE_HIGH_REGS.
22187 (REG_CLASS_NAMES): Ditto.
22188 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22189 (REGISTER_NAMES): Ditto.
22190 (rs6000_reg_names): Ditto.
22191 * doc/tm.texi.in: Remove SPE as example.
22192 * doc/tm.texi: Regenerate.
22194 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22196 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22197 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22198 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22199 * config/rs6000/rs6000.md (type): Remove "brinc".
22201 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22203 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22204 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22205 * config/rs6000/linuxspe.h: Delete file.
22206 * config/rs6000/rs6000.md: Don't include spe.md.
22207 * config/rs6000/spe.h: Delete file.
22208 * config/rs6000/spe.md: Delete file.
22209 * config/rs6000/t-rs6000: Remove spe.md.
22211 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22213 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22214 (reg_or_none500mem_operand): Delete.
22215 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22216 instead of reg_or_none500mem_operand.
22218 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22220 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22221 handling of SPE flags.
22222 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22224 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22226 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22228 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22229 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22230 paired_divv2sf3): Similar.
22231 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22232 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22233 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22235 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22236 Rename the paired_* instruction patterns.
22237 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22239 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22240 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22241 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22242 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22243 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22244 PAIRED_VECTOR_MODE.
22245 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22246 (spe_func_has_64bit_regs_p): Delete.
22247 (spe_expand_predicate_builtin): Delete.
22248 (spe_expand_evsel_builtin): Delete.
22249 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22250 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22251 (invalid_e500_subreg): Delete.
22252 (rs6000_legitimize_address): Always force_reg op2 as well, for
22253 paired single memory accesses.
22254 (rs6000_member_type_forces_blk): Delete.
22255 (rs6000_spe_function_arg): Delete.
22256 (rs6000_expand_unop_builtin): Delete SPE handling.
22257 (rs6000_expand_binop_builtin): Ditto.
22258 (spe_expand_stv_builtin): Delete.
22259 (bdesc_2arg_spe): Delete.
22260 (spe_expand_builtin): Delete.
22261 (spe_expand_predicate_builtin): Delete.
22262 (spe_expand_evsel_builtin): Delete.
22263 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22264 (spe_init_builtins): Delete.
22265 (spe_func_has_64bit_regs_p): Delete.
22266 (savres_routine_name): Delete "info" parameter. Adjust callers.
22267 (rs6000_emit_stack_reset): Ditto.
22268 (rs6000_dwarf_register_span): Delete.
22269 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22270 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22271 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22273 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22275 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22276 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22277 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22278 mulv2sf3, divv2sf3): Delete expanders.
22280 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22282 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22284 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22286 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22287 * config/rs6000/rs6000.c: Ditto.
22289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22291 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22292 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22293 comparison_operator.
22295 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22297 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22298 * config/rs6000/rs6000.opt: Ditto.
22299 * config/rs6000/t-rtems: Ditto.
22301 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22303 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22304 TARGET_E500_SINGLE by 0, simplify.
22305 * config/rs6000/rs6000.c: Ditto.
22306 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22307 (spe_build_register_parallel): Delete.
22308 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22309 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22310 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22311 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22312 (E500_CONVERT): Delete.
22313 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22315 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22317 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22318 * config/rs6000/dfp.md: Ditto.
22319 (negdd2, *negdd2_fpr): Merge.
22320 (absdd2, *absdd2_fpr): Merge.
22321 (negtd2, *negtd2_fpr): Merge.
22322 (abstd2, *abstd2_fpr): Merge.
22323 * config/rs6000/e500.h: Delete file.
22324 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22325 TARGET_FPRS by 1 and simplify.
22326 * config/rs6000/rs6000-c.c: Ditto.
22327 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22328 TARGET_DF_SPE by 0.
22329 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22331 * config/rs6000/rs6000.md: Ditto.
22332 (floatdidf2, *floatdidf2_fpr): Merge.
22333 (move_from_CR_gt_bit): Delete.
22334 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22335 (E500_CR_IOR_COMPARE): Delete.
22336 (All patterns that require !TARGET_FPRS): Delete.
22337 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22339 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22341 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22343 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22345 * graphds.c (add_edge): Intitialize edge's attached data.
22346 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22347 pointer parameter. Call pointed function on each edge during
22348 graph traversing. Skip traversing the edge when the function
22350 (graphds_dfs, graphds_scc): Ditto.
22351 (for_each_edge): New parameter. Pass the new parameter to callback
22353 * graphds.h (skip_edge_callback): New function pointer type.
22354 (graphds_dfs, graphds_scc): New function pointer parameter.
22355 (graphds_edge_callback, for_each_edge): New parameter.
22357 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22359 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22360 out code checking if runtime alias check is possible to below ...
22361 Call the new function.
22362 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22363 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22365 2017-06-07 Marek Polacek <polacek@redhat.com>
22368 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22369 TYPE_OVERFLOW_WRAPS check.
22371 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22373 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22374 if versioning is required.
22375 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22376 peeling with the check for versioning.
22378 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22380 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22381 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22382 Set true to new parameter if new ssa variable is defined.
22383 (vect_gen_vector_loop_niters): Refactor. Set range information
22384 for the new vector loop bound variable.
22385 (vect_do_peeling): Ditto.
22387 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22389 * tree-affine.c (ssa.h): Include header file.
22390 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22391 has wrapping overflow behavior.
22393 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22395 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22397 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22399 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22400 (tree_to_aff_combination): ... here.
22402 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22404 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22405 reg_pressure model function.
22406 (ivopts_global_cost_for_size): Delete.
22407 (determine_set_costs, iv_ca_recount_cost): Call new model function
22408 ivopts_estimate_reg_pressure.
22410 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22412 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22413 expensive than udiv. Remove floating point cases from mod.
22415 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22417 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22418 Increase idiv cost.
22420 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22422 * config/aarch64/aarch64.md
22423 (copysignsf3): Fix mask generation.
22425 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22427 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22429 (class dump_manager): Add register_dumps method.
22430 * dumpfile.c: Include langhooks.h.
22431 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22432 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22433 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22434 (dump_manager::dump_register): Start with 512 entries instead of 32.
22435 (dump_manager::register_dumps): New method.
22436 * toplev.c (general_init): Instead of invoking register_dumps
22437 langhook, invoke register_dumps method on the dump manager.
22438 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22441 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22443 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22444 Say that # requires an associated define_split to exist, and that
22445 the define_split must be suitable for use after register allocation.
22447 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22449 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22450 (compute_outgoing_frequencies): Also initialize zero counts.
22451 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22452 around loops; preserve more of profile when nothing changes.
22454 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22456 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22458 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22459 * config/arm/arm-cpu-cdata.h: Regenerate.
22460 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22461 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22462 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22463 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22464 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22466 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22468 * config/arm/t-rmprofile: Likewise.
22469 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22471 2017-06-06 David S. Miller <davem@davemloft.net>
22474 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22475 blockage if function uses alloca.
22477 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22479 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22480 New "uid" fields to hold pretty-print IDs of group and ref.
22481 Memory references are now identified as <group_id>:<ref_id>
22482 instead of using [random] addresses.
22483 (dump_mem_details): Simplify, no functional change.
22484 (dump_mem_ref): Simplify and make output more concise.
22485 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22486 (find_or_create_group): Initialize group uid.
22487 (record_ref): Initialize ref uid. Improve debug output.
22488 (prune_group_by_reuse, should_issue_prefetch_p,)
22489 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22490 (mark_nontemporal_store, determine_loop_nest_reuse):
22491 Improve debug output.
22493 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22495 * dbgcnt.def (prefetch): New debug counter.
22496 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22497 (schedule_prefetches): Stop issueing prefetches if debug counter
22500 2017-06-06 Tom de Vries <tom@codesourcery.com>
22502 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22503 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22505 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22507 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22508 Use aarch64_reg_or_zero predicate for operand 4.
22509 (aarch64_compare_and_swap<mode> define_insn_and_split):
22510 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22511 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22513 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22515 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22516 (arm_compute_save_core_reg_mask): This.
22517 (thumb1_compute_save_reg_mask): Rename into ...
22518 (thumb1_compute_save_core_reg_mask): This.
22519 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22520 (arm_compute_frame_layout): Likewise.
22522 2017-06-06 Richard Biener <rguenther@suse.de>
22524 PR tree-optimization/80974
22525 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22526 keep or clear leaders SSA info.
22528 2017-06-06 Tom de Vries <tom@codesourcery.com>
22530 * config/nvptx/nvptx.c (split_mode_p): New function.
22531 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22533 2017-06-06 Tom de Vries <tom@codesourcery.com>
22535 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22537 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22540 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22543 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22545 * shrink-wrap.c (handle_simple_exit): Update profile.
22546 (try_shrink_wrapping): Upate profile.
22548 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22550 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22551 (tree_guess_outgoing_edge_probabilities): New.
22552 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22553 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22555 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22557 * ipa-split.c (split_function): Initialize return bb profile.
22559 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22561 * profile.c (compute_branch_probabilities): Also initialize
22562 EXIT_BLOCK profile.
22564 2017-06-06 Richard Biener <rguenther@suse.de>
22566 PR tree-optimization/80928
22567 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22568 (vect_analyze_loop_operations): Properly guard analysis for
22570 (vect_transform_loop): Likewise.
22571 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22572 (vect_model_induction_cost): Do not cost for pure SLP.
22573 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22574 of induction in inner loop vectorization.
22575 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22576 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22577 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22579 (vect_analyze_slp_cost_1): Cost induction.
22580 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22581 (vect_get_slp_vect_defs): Likewise.
22582 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22583 (vect_transform_stmt): Handle SLP reductions.
22584 * tree-vectorizer.h (vectorizable_induction): Adjust.
22586 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22588 * config/rs6000/rs6000.c (make_resolver_func): Update
22589 init_lowered_empty_function call.
22591 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22593 * doc/invoke.texi: Document the -fprofile-abs-path option.
22594 * common.opt (fprofile-abs-path): New option.
22595 * gcov-io.h (gcov_write_filename): Declare.
22596 * gcov-io.c (gcov_write_filename): New function.
22597 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22598 * profile.c (output_location): Likewise.
22600 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22602 * shring-wrap.c: Revert accidental commit.
22604 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22606 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22608 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22610 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22612 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22613 profile in callgraph edge.
22614 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22615 (apply_scale): Likewise.
22616 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22617 Also scale profile when inlining function with zero profile.
22618 (initialize_cfun): Update exit block profile even when it is zero.
22619 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22620 when profile is read.
22622 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22624 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22625 (CLONE_*): New constants to define the processors we can generate
22626 code for with the target_clone attribute.
22627 (rs6000_clone_map): New array to identify which clone processors
22628 the current program is running on.
22629 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22630 target_clone attribute.
22631 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22632 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22633 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22634 (cpu_expand_builtin): Add support for target_clone attribute.
22635 (rs6000_valid_attribute_p): Allow "default" attribute.
22636 (get_decl_name): New debug function to simplify printing the
22637 current function name in debugging statements.
22638 (rs6000_clone_priority): New functions to support the target_clone
22639 attribute, and be able to generate code to switch between ISA 2.05
22640 through ISA 3.0 (power6 through power9).
22641 (rs6000_compare_version_priority): Likewise.
22642 (rs6000_get_function_versions_dispatcher): Likewise.
22643 (make_resolver_func): Likewise.
22644 (add_condition_to_bb): Likewise.
22645 (dispatch_function_versions): Likewise.
22646 (rs6000_generate_version_dispatcher_body): Likewise.
22647 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22648 (fusion_gpr_load_p): Fix a spacing issue.
22649 * doc/extend.texi (Common Function Attributes): Document that the
22650 PowerPC supports the target_clone attribute.
22652 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22654 * config/arm/arm.h: explain F symbol found in description of ARM
22655 register allocation in its legend.
22657 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22659 * config/mips/frame-header-opt.c: Include profile-count.h.
22660 * config/riscv/riscv.c: Include profile-count.h
22662 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22664 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22666 (sm_set_flag_if_changed): Add bbs field.
22667 (execute_sm_if_changed_flag_set): Pass BBS.
22668 (execute_sm): Update.
22670 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22672 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22675 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22677 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22678 (peephole2): New peephole2 to emit the above.
22679 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22681 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22683 * config/aarch64/aarch64.c (define_peephole2 above
22684 *sub_<shift>_<mode>): New peephole.
22686 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22688 * config/i386/i386.c (make_resolver_func): Update.
22689 * Makefile.in: Add profile-count.h and profile-count.o
22690 * auto-profile.c (afdo_indirect_call): Update to new API.
22691 (afdo_set_bb_count): Update.
22692 (afdo_propagate_edge): Update.
22693 (afdo_propagate_circuit): Update.
22694 (afdo_calculate_branch_prob): Update.
22695 (afdo_annotate_cfg): Update.
22696 * basic-block.h: Include profile-count.h
22697 (struct edge_def): Turn count to profile_count.
22698 (struct basic_block_def): Likewie.
22699 (REG_BR_PROB_BASE): Move to profile-count.h
22700 (RDIV): Move to profile-count.h
22701 * bb-reorder.c (max_entry_count): Turn to profile_count.
22702 (find_traces): Update.
22703 (rotate_loop):Update.
22704 (connect_traces):Update.
22705 (sanitize_hot_paths):Update.
22706 * bt-load.c (migrate_btr_defs): Update.
22707 * cfg.c (RDIV): Remove.
22708 (init_flow): Use alloc_block.
22709 (alloc_block): Uninitialize count.
22710 (unchecked_make_edge): Uninitialize count.
22711 (check_bb_profile): Update.
22712 (dump_edge_info): Update.
22713 (dump_bb_info): Update.
22714 (update_bb_profile_for_threading): Update.
22715 (scale_bbs_frequencies_int): Update.
22716 (scale_bbs_frequencies_gcov_type): Update.
22717 (scale_bbs_frequencies_profile_count): New.
22718 * cfg.h (update_bb_profile_for_threading): Update.
22719 (scale_bbs_frequencies_profile_count): Declare.
22720 * cfgbuild.c (compute_outgoing_frequencies): Update.
22721 (find_many_sub_basic_blocks): Update.
22722 * cfgcleanup.c (try_forward_edges): Update.
22723 (try_crossjump_to_edge): Update.
22724 * cfgexpand.c (expand_gimple_tailcall): Update.
22725 (construct_exit_block): Update.
22726 * cfghooks.c (verify_flow_info): Update.
22727 (dump_bb_for_graph): Update.
22728 (split_edge): Update.
22729 (make_forwarder_block): Update.
22730 (duplicate_block): Update.
22731 (account_profile_record): Update.
22732 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22733 (get_estimated_loop_iterations): Update.
22734 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22735 (single_likely_exit): Update.
22736 * cfgloopmanip.c (scale_loop_profile): Update.
22738 (set_zero_probability): Update.
22739 (lv_adjust_loop_entry_edge): Update.
22740 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22741 (purge_dead_edges): Update.
22742 (rtl_account_profile_record): Update.
22743 * cgraph.c (cgraph_node::create): Uninitialize count.
22744 (symbol_table::create_edge): Uninitialize count.
22745 (cgraph_update_edges_for_call_stmt_node): Update.
22746 (cgraph_edge::dump_edge_flags): Update.
22747 (cgraph_node::dump): Update.
22748 (cgraph_edge::maybe_hot_p): Update.
22749 * cgraph.h: Include profile-count.h
22750 (create_clone), create_edge, create_indirect_edge): Update.
22751 (cgraph_node): Turn count to profile_count.
22752 (cgraph_edge0: Likewise.
22753 (make_speculative, clone): Update.
22754 (create_edge): Update.
22755 (init_lowered_empty_function): Update.
22756 * cgraphclones.c (cgraph_edge::clone): Update.
22757 (duplicate_thunk_for_node): Update.
22758 (cgraph_node::create_clone): Update.
22759 * cgraphunit.c (cgraph_node::analyze): Update.
22760 (cgraph_node::expand_thunk): Update.
22761 * final.c (dump_basic_block_info): Update.
22762 * gimple-streamer-in.c (input_bb): Update.
22763 * gimple-streamer-out.c (output_bb): Update.
22764 * graphite.c (print_global_statistics): Update.
22765 (print_graphite_scop_statistics): Update.
22766 * hsa-brig.c: Include basic-block.h.
22767 * hsa-dump.c: Include basic-block.h.
22768 * hsa-gen.c (T sum_slice): Update.
22769 (convert_switch_statements):Update.
22770 * hsa-regalloc.c: Include basic-block.h.
22771 * ipa-chkp.c (chkp_produce_thunks): Update.
22772 * ipa-cp.c (struct caller_statistics): Update.
22773 (init_caller_stats): Update.
22774 (gather_caller_stats): Update.
22775 (ipcp_cloning_candidate_p): Update.
22776 (good_cloning_opportunity_p): Update.
22777 (get_info_about_necessary_edges): Update.
22778 (dump_profile_updates): Update.
22779 (update_profiling_info): Update.
22780 (update_specialized_profile): Update.
22781 (perhaps_add_new_callers): Update.
22782 (decide_about_value): Update.
22783 (ipa_cp_c_finalize): Update.
22784 * ipa-devirt.c (struct odr_type_warn_count): Update.
22785 (struct decl_warn_count): Update.
22786 (struct final_warning_record): Update.
22787 (possible_polymorphic_call_targets): Update.
22788 (ipa_devirt): Update.
22789 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22790 * ipa-icf.c (sem_function::merge): Update.
22791 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22792 * ipa-inline.c (compute_uninlined_call_time): Update.
22793 (compute_inlined_call_time): Update.
22794 (want_inline_small_function_p): Update.
22795 (want_inline_self_recursive_call_p): Update.
22796 (edge_badness): Update.
22797 (lookup_recursive_calls): Update.
22798 (recursive_inlining): Update.
22799 (inline_small_functions): Update.
22800 (dump_overall_stats): Update.
22801 (dump_inline_stats): Update.
22802 * ipa-profile.c (ipa_profile_generate_summary): Update.
22803 (ipa_propagate_frequency): Update.
22804 (ipa_profile): Update.
22805 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22806 * ipa-utils.c (ipa_merge_profiles): Update.
22807 * loop-doloop.c (doloop_modify): Update.
22808 * loop-unroll.c (report_unroll): Update.
22809 (unroll_loop_runtime_iterations): Update.
22810 * lto-cgraph.c (lto_output_edge): Update.
22811 (lto_output_node): Update.
22812 (input_node): Update.
22813 (input_edge): Update.
22814 (merge_profile_summaries): Update.
22815 * lto-streamer-in.c (input_cfg): Update.
22816 * lto-streamer-out.c (output_cfg): Update.
22817 * mcf.c (create_fixup_graph): Update.
22818 (adjust_cfg_counts): Update.
22819 (sum_edge_counts): Update.
22820 * modulo-sched.c (sms_schedule): Update.
22821 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22822 * predict.c (maybe_hot_count_p): Update.
22823 (probably_never_executed): Update.
22824 (dump_prediction): Update.
22825 (combine_predictions_for_bb): Update.
22826 (propagate_freq): Update.
22827 (handle_missing_profiles): Update.
22828 (counts_to_freqs): Update.
22829 (rebuild_frequencies): Update.
22830 (force_edge_cold): Update.
22831 * predict.h: Include profile-count.h
22832 (maybe_hot_count_p, counts_to_freqs): UPdate.
22833 * print-rtl-function.c: Do not include cfg.h
22834 * print-rtl.c: Include basic-block.h
22835 * profile-count.c: New file.
22836 * profile-count.h: New file.
22837 * profile.c (is_edge_inconsistent): Update.
22838 (correct_negative_edge_counts): Update.
22839 (is_inconsistent): Update.
22840 (set_bb_counts): Update.
22841 (read_profile_edge_counts): Update.
22842 (compute_frequency_overlap): Update.
22843 (compute_branch_probabilities): Update; Initialize and deinitialize
22845 (branch_prob): Update.
22846 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22847 (edge_gcov_count): New.
22848 (bb_gcov_count): New.
22849 * shrink-wrap.c (try_shrink_wrapping): Update.
22850 * tracer.c (better_p): Update.
22851 * trans-mem.c (expand_transaction): Update.
22852 (ipa_tm_insert_irr_call): Update.
22853 (ipa_tm_insert_gettmclone_call): Update.
22854 * tree-call-cdce.c: Update.
22855 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22856 (gimple_duplicate_sese_tail): Update.
22857 (gimple_account_profile_record): Update.
22858 (execute_fixup_cfg): Update.
22859 * tree-inline.c (copy_bb): Update.
22860 (copy_edges_for_bb): Update.
22861 (initialize_cfun): Update.
22862 (freqs_to_counts): Update.
22863 (copy_cfg_body): Update.
22864 (expand_call_inline): Update.
22865 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22866 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22867 (try_unroll_loop_completely): Update.
22868 (try_peel_loop): Update.
22869 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22870 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22871 * tree-ssa-loop-split.c (connect_loops): Update.
22872 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22873 * tree-ssa-reassoc.c (branch_fixup): Update.
22874 * tree-ssa-tail-merge.c (replace_block_by): Update.
22875 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22876 (compute_path_counts): Update.
22877 (update_profile): Update.
22878 (recompute_probabilities): Update.
22879 (update_joiner_offpath_counts): Update.
22880 (estimated_freqs_path): Update.
22881 (freqs_to_counts_path): Update.
22882 (clear_counts_path): Update.
22883 (ssa_fix_duplicate_block_edges): Update.
22884 (duplicate_thread_path): Update.
22885 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22886 (struct switch_conv_info): Update.
22887 * tree-tailcall.c (decrease_profile): Update.
22888 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22889 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22890 * value-prof.c (check_counter): Update.
22891 (gimple_divmod_fixed_value): Update.
22892 (gimple_mod_pow2): Update.
22893 (gimple_mod_subtract): Update.
22894 (gimple_ic_transform): Update.
22895 (gimple_stringop_fixed_value): Update.
22896 * value-prof.h (gimple_ic): Update.
22898 2017-06-02 Carl Love <cel@us.ibm.com>
22900 * config/rs6000/rs6000-c: Add support for built-in functions
22901 vector double vec_doublee (vector signed int);
22902 vector double vec_doublee (vector unsigned int);
22903 vector double vec_doublee (vector float);
22904 vector double vec_doubleh (vector signed int);
22905 vector double vec_doubleh (vector unsigned int);
22906 vector double vec_doubleh (vector float);
22907 vector double vec_doublel (vector signed int);
22908 vector double vec_doublel (vector unsigned int);
22909 vector double vec_doublel (vector float);
22910 vector double vec_doubleo (vector signed int);
22911 vector double vec_doubleo (vector unsigned int);
22912 vector double vec_doubleo (vector float);.
22913 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22914 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22916 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22917 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22918 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22920 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22921 vec_doublel, vec_doubleh.
22922 * doc/extend.texi: Update the built-in documentation file for the
22923 new built-in functions.
22925 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22928 * ipa-inline-analysis.c (free_growth_caches): Set
22929 edge_removal_hook_holder to NULL after removing it.
22931 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22933 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22934 comparision with zero.
22936 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22937 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22938 for early expansion of vec_min and vec_max builtins.
22939 (builtin_function_type): Add min/max unsigned variants to those
22940 identified as having unsigned arguments.
22942 2017-06-02 Olivier Hainque <hainque@adacore.com>
22944 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22946 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22948 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22949 Use VALL_F16 iterator rather than VALL.
22951 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22953 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22954 Emit CBNZ inside loop when doing a strong exchange and comparing
22955 against zero. Generate the CC flags after the loop.
22957 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22959 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22960 (dl_section_ref): New.
22961 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22962 On AIX, append an expression to subtract the size of the
22963 section length to dl_section_ref.
22965 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22967 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22968 for early expansion of vector absolute builtins.
22970 2017-06-02 Richard Biener <rguenther@suse.de>
22972 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22973 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22975 2017-06-02 Richard Biener <rguenther@suse.de>
22977 PR tree-optimization/80948
22978 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22979 stmt order as well.
22981 2017-06-02 Richard Biener <rguenther@suse.de>
22983 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22985 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22986 for inductions as relevant.
22988 2017-06-02 Richard Biener <rguenther@suse.de>
22990 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22991 (vectorizable_induction): ... this. Remove dead code.
22993 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22995 * builtins. (expand_builtin_alloca): Remove second parameter and
22996 infer its value from the first parameter instead.
22997 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22999 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23001 PR rtl-optimization/80903
23002 * loop-doloop.c (add_test): Unshare sequence.
23004 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23006 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23008 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23010 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23012 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23013 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23014 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23015 (xlogue_layout::get_stub_rtx): Make static.
23016 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23017 (xlogue_layout::compute_stub_managed_regs): Rename to...
23018 (xlogue_layout::count_stub_managed_regs): ...this.
23019 (xlogue_layout::is_stub_managed_reg): New function.
23020 (xlogue_layout::m_stub_names): Rename to...
23021 (xlogue_layout::s_stub_names): ...this, make static.
23022 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23023 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23024 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23025 xlogue_layout::s_stub_names): Instantiate statics.
23026 (stub_managed_regs): Remove.
23027 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23028 (disable_call_ms2sysv_xlogues): Rename to...
23029 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23030 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23032 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23033 change after reload_completed.
23034 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23036 (ix86_expand_prologue): Likewise.
23037 (ix86_expand_epilogue): Likewise.
23038 (ix86_expand_split_stack_prologue): Likewise.
23039 (ix86_compute_frame_layout): Remove frame parameter ...
23040 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23041 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23043 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23044 (ix86_frame): Move from here ...
23045 * config/i386/i386.h (ix86_frame): ... to here.
23046 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23047 complete ix86_frame data structure instead. Remove some_ld_name.
23049 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23051 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23052 symbols that hold a DECL_VALUE_EXPR.
23054 2017-06-01 Martin Jambor <mjambor@suse.cz>
23056 PR tree-optimization/80898
23057 * tree-sra.c (process_subtree_disqualification): Removed.
23058 (disqualify_candidate): Do not acll
23059 process_subtree_disqualification.
23060 (subtree_mark_written_and_enqueue): New function.
23061 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23062 RHS has been disqualified and re-queue LHS if necessary. Apart
23063 from that, ignore disqualified RHS.
23065 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23067 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23068 address fetch for z10 or later.
23070 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23072 * config/arc/arc.md (tst_movb): Add guard when splitting.
23074 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23076 * config/arc/arc.c (arc_can_eliminate): Test against
23077 arc_frame_pointer_needed.
23079 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23081 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23082 to prevent store reordering.
23083 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23084 (type): Add block type.
23085 (stack_tie): Define special instruction to be used in
23088 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23090 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23091 constraint. It is not valid for the pattern.
23092 (noncommutative_binary_comparison): Likewise.
23094 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23096 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23099 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23101 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23102 be used by the reg-alloc.
23104 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23106 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23107 reg-alloc when having mul64 or mul32x16 instructions.
23108 (mulsidi3): Likewise.
23109 (umulsidi3): Likewise.
23110 (mulsi32x16): New pattern.
23111 (mulsi64): Likewise.
23112 (mulsidi64): Likewise.
23113 (umulsidi64): Likewise.
23114 (MUL32x16_REG): Define.
23115 (mul64_600): Use MUL32x16_REG.
23116 (mac64_600): Likewise.
23117 (umul64_600): Likewise.
23118 (umac64_600): Likewise.
23120 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23122 * config/arc/arc.md (mulsi3_700): Make it commutative.
23124 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23126 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23128 (*sign_extendsidi2_insn): Likewise for movstosw.
23130 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23132 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23133 the type of the input discriminant value. Convert the
23134 discriminant value of signedness vary.
23136 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23138 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23139 Add to -Wall section.
23141 2017-06-01 Richard Biener <rguenther@suse.de>
23143 PR middle-end/66313
23144 * fold-const.c (fold_plusminus_mult_expr): If the factored
23145 factor may be zero use a wrapping type for the inner operation.
23146 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23147 and handle moved defs.
23148 (process_assignment): Properly guard the unary op case. Return a
23149 tri-state indicating that moving the stmt before the call may allow
23150 to continue. Pass through to_move.
23151 (find_tail_calls): Handle moving unrelated defs before
23154 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23157 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23158 splitter result in the canonical way.
23160 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23162 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23163 also for 32bit target. Update insn attributes.
23164 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23166 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23168 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23169 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23170 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23171 (_mm_maskz_min_ss): New intrinsics.
23173 2017-05-31 Martin Liska <mliska@suse.cz>
23175 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23176 Change comment style to one we normally use.
23177 (vectorizable_reduction): Likewise.
23178 (vectorizable_induction): Likewise.
23179 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23180 (vectorizable_call): Likewise.
23181 (vectorizable_simd_clone_call): Likewise.
23182 (vectorizable_conversion): Likewise.
23183 (vectorizable_assignment): Likewise.
23184 (vectorizable_shift): Likewise.
23185 (vectorizable_operation): Likewise.
23186 (vectorizable_store): Likewise.
23187 (vectorizable_load): Likewise.
23188 * tree-vectorizer.h: Likewise.
23190 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23192 * passes.c (emergency_dump_function): New.
23193 * tree-pass.h (emergency_dump_function): Declare.
23194 * plugin.c (plugins_internal_error_function): Remove.
23195 * plugin.h (plugins_internal_error_function): Remove declaration.
23196 * toplev.c (internal_error_function): New static function. Use it...
23197 (general_init): ...here.
23199 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23201 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23202 (arc_rtx_costs): Add costs for new patterns.
23203 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23204 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23206 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23208 * tree-ssa-strlen.c (get_next_strinfo): New function.
23209 (get_stridx_plus_constant): Use it.
23210 (zero_length_string): Likewise.
23211 (adjust_related_strinfos): Likewise.
23212 (adjust_last_stmt): Likewise.
23214 2017-05-31 Richard Biener <rguenther@suse.de>
23217 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23218 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23220 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23222 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23223 loop_vinfo argument and use of dependence distance vectors.
23224 Check instead whether the two references differ only in their
23225 initial value and assume that they have the same alignment if the
23226 difference is a multiple of the vector alignment.
23227 (vect_analyze_data_refs_alignment): Update call accordingly.
23229 2017-05-31 Martin Liska <mliska@suse.cz>
23232 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23234 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23236 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23237 (create_intersect_range_checks): Move from ...
23238 * tree-data-ref.c (create_intersect_range_checks_index)
23239 (create_intersect_range_checks): ... to here.
23240 (create_runtime_alias_checks): New function factored from ...
23241 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23242 here. Call above function.
23243 * tree-data-ref.h (create_runtime_alias_checks): New function.
23245 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23247 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23248 segment length for dr_b and compute it in wide_int.
23250 2017-05-31 Richard Biener <rguenther@suse.de>
23252 PR tree-optimization/80906
23253 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23254 and pass through iv_map.
23255 (copy_bb_and_scalar_dependences): Adjust.
23256 (translate_pending_phi_nodes): Likewise.
23257 (copy_loop_close_phi_args): Handle code-generating IVs instead
23260 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23262 * diagnostic-color.c (color_dict): Add "type-diff".
23263 (parse_gcc_colors): Update comment.
23264 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23265 -fdiagnostics-show-template-tree and -fno-elide-type.
23266 (GCC_COLORS): Add type-diff to example.
23268 (-fdiagnostics-show-template-tree): New.
23269 (-fno-elide-type): New.
23270 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23271 the pp_format_decoder callback. Call any m_format_postprocessor's
23273 (pretty_printer::pretty_printer): Initialize
23274 m_format_postprocessor.
23275 (pretty_printer::~pretty_printer): Delete any
23276 m_format_postprocessor.
23277 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23278 (class format_postprocessor): New class.
23279 (struct pretty_printer::format_decoder): Document the new parameters.
23280 (struct pretty_printer::m_format_postprocessor): New field.
23281 * tree-diagnostic.c (default_tree_printer): Update for new
23282 bool and const char ** params.
23283 * tree-diagnostic.h (default_tree_printer): Likewise.
23285 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23287 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23288 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23289 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23290 rs6000_gen_cell_microcode code.
23291 (rs6000_final_prescan_insn): Delete.
23292 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23293 "warn-cell-microcode" entries.
23294 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23295 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23296 throughout. Change cc_reg_not_micro_cr0_operand to
23297 cc_reg_not_cr0_operand throughout.
23298 (*extendhi<mode>2_noload): Delete.
23299 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23300 (mwarn-cell-microcode): Delete.
23301 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23302 -mgen-cell-microcode and -mwarn-cell-microcode.
23304 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23307 * config/i386/constraints.md (Yd): New constraint.
23309 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23310 and (?Yd, r) alternatives. Update insn attributes.
23311 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23312 and (?*Yd, r) alternatives. Update insn attributes.
23313 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23315 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23317 * gimplify.c (gimplify_modify_expr): Don't create a
23318 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23321 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23323 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23325 2017-05-30 Richard Biener <rguenther@suse.de>
23327 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23328 and reduc_def fields.
23329 (STMT_VINFO_REDUC_TYPE): New define.
23330 (STMT_VINFO_REDUC_DEF): Likewise.
23331 (vect_force_simple_reduction): Adjust prototype.
23332 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23333 (vect_is_simple_reduction): Remove check_reduction argument.
23334 (vect_force_simple_reduction): Adjust and set
23335 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23336 (vectorizable_reduction): Do not re-do reduction analysis
23337 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23338 * tree-parloops.c (gather_scalar_reductions): Adjust.
23340 2017-05-30 Richard Biener <rguenther@suse.de>
23342 PR middle-end/80901
23343 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23346 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23348 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23349 Introduce unknown_misalignment parameter and remove vf.
23350 (vect_peeling_hash_get_lowest_cost):
23351 Pass unknown_misalignment parameter.
23352 (vect_enhance_data_refs_alignment):
23353 Fix unsupportable data ref treatment.
23355 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23357 * tree-vect-data-refs.c (vect_get_data_access_cost):
23358 Workaround for SLP handling.
23359 (vect_enhance_data_refs_alignment):
23360 Compute costs for doing no peeling at all, compare to the best
23361 peeling costs so far and avoid peeling if cheaper.
23363 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23365 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23366 Return peeling info and set costs to zero for unlimited cost
23368 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23369 with unknown misalignment. Compute and costs for unknown
23370 misalignment, compare them to the costs for known misalignment
23371 and choose the cheapest for peeling.
23373 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23375 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23376 (vect_get_peeling_costs_all_drs): Create function.
23377 (vect_peeling_hash_get_lowest_cost):
23378 Use vect_get_peeling_costs_all_drs.
23379 (vect_peeling_supportable): Create function.
23380 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23382 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23384 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23385 DR_HAS_NEGATIVE_STEP.
23386 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23387 (vect_enhance_data_refs_alignment): Use.
23388 (vect_duplicate_ssa_name_ptr_info): Use.
23389 * tree-vectorizer.h (dr_misalignment): Use.
23390 (known_alignment_for_access_p): Use.
23392 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23395 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23397 (has_section_name): New function.
23399 2017-05-30 Martin Liska <mliska@suse.cz>
23402 * auto-profile.c (get_function_decl_from_block): Fix
23405 2017-05-30 Richard Biener <rguenther@suse.de>
23407 PR middle-end/80876
23408 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23410 2017-05-30 Martin Liska <mliska@suse.cz>
23412 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23413 * dumpfile.h (struct dump_file_info): Remove ctors.
23415 2017-05-30 Martin Liska <mliska@suse.cz>
23417 * predict.def: Fix GNU coding style.
23419 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23421 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23422 Mark 'to' argument with ATTRIBUTE_UNUSED.
23424 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23426 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23427 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23428 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23431 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23433 * doc/install.texi (Options specification): Restore entry of
23434 --enable-sjlj-exceptions.
23436 2017-05-27 Michael Eager <eager@eagercon.com>
23439 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23441 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23443 * config/microblaze/microblaze.h
23444 (FIXED_REGISTERS): Update in macro.
23445 (CALL_USED_REGISTERS): Update in macro.
23447 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23449 * doc/install.texi: Add links to macOS binary distributions.
23451 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23455 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23457 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23459 2017-05-26 Martin Liska <mliska@suse.cz>
23461 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23463 2017-05-26 Martin Liska <mliska@suse.cz>
23465 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23466 always leading ';; '.
23467 (dump_bb_info): Likewise.
23468 (brief_dump_cfg): Likewise.
23469 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23470 * dumpfile.c: Remove usage of TDF_VERBOSE.
23471 * dumpfile.h (enum dump_kind): Likewise.
23472 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23473 * print-tree.c (debug_verbose): Remove.
23474 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23475 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23476 * tree-diagnostic.c (default_tree_printer): Replace
23477 TDF_DIAGNOSTIC with TDF_SLIM.
23479 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23481 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23482 in parameter loop, rather than loop_vinfo.
23483 (create_intersect_range_checks): Ditto.
23484 (vect_create_cond_for_alias_checks): Update call to above functions.
23486 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23488 PR tree-optimization/80815
23489 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23490 for merging runtime alias checks. Handle negative DR_STEPs.
23492 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23494 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23496 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23497 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23498 out code pruning runtime alias checks.
23499 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23500 factored out from above.
23501 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23503 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23505 (prune_runtime_alias_test_list): New decalaration.
23507 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23509 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23510 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23511 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23512 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23513 (operator==, comp_dr_with_seg_len_pair): Ditto.
23514 (vect_prune_runtime_alias_test_list): Ditto.
23516 2017-05-26 Martin Liska <mliska@suse.cz>
23519 * params.def: Bound partial-inlining-entry-probability param.
23521 2017-05-26 Marek Polacek <polacek@redhat.com>
23524 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23527 2017-05-26 Richard Biener <rguenther@suse.de>
23529 PR tree-optimization/80842
23530 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23533 2017-05-26 Richard Biener <rguenther@suse.de>
23535 PR tree-optimization/80844
23536 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23538 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23540 * doc/md.texi (Machine Constraints): Update x86 family
23541 machine constraints section to match 'config/i386/constraints.md'.
23543 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23545 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23547 2017-05-25 Nathan Sidwell <nathan@acm.org>
23549 * doc/invoke.texi (--enable-languages): Update documentation.
23551 2017-05-25 Martin Liska <mliska@suse.cz>
23553 * dumpfile.c: Add TDF_FOLDING.
23554 * dumpfile.h (enum dump_kind): Likewise.
23555 * genmatch.c (dt_simplify::gen_1): Use it.
23557 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23559 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23561 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23563 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23564 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23566 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23568 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23569 * match.pd (X == C): Rewrite it here.
23570 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23571 with_certain_nonzero_bits2): New predicates.
23572 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23574 2017-05-24 Nathan Sidwell <nathan@acm.org>
23576 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23579 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23582 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23584 * config/powerpcspe: New port. Files are copied from the rs6000
23585 port, with "rs6000" in filenames replaced by "powerpcspe".
23587 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23589 PR rtl-optimization/80754
23590 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23592 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23594 * config/sparc/sparc.md (length): Return the correct value for -mflat
23595 sibcalls to match output_sibcall.
23597 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23601 * config/rs6000/rs6000.c (struct machine_function): Add new field
23603 (rs6000_get_separate_components): Init that field, use it.
23604 (rs6000_components_for_bb): Use the field.
23606 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23608 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23610 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23612 PR middle-end/80823
23613 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23615 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23618 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23619 address against address_operand predicate.
23620 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23622 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23624 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23625 parameters passed indirectly.
23627 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23629 * config/i386/i386.md (*movdi_internal): Remove SSE4
23630 alternative 18 (?r, *v). Update insn attributes.
23631 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23632 Update insn attributes.
23633 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23634 Update insn attributes.
23635 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23636 alternative 1 (r, v). Remove isa attribute.
23637 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23638 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23639 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23641 2017-05-23 Tom de Vries <tom@codesourcery.com>
23643 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23646 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23648 * cgraphunit.c (symbol_table::process_new_functions): Update.
23649 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23650 (inline_generate_summary): Rename to ...
23651 (ipa_fn_summary_generate): ... this one.
23652 (inline_read_summary): Rename to ...
23653 (ipa_fn_summary_read): ... this one.
23654 (inline_write_summary): Rename to ...
23655 (ipa_fn_summary_write): ... this one.
23656 (inline_free_summary): Rename to ...
23657 (ipa_free_fn_summary): ... this one.
23658 (pass_data_local_fn_summary, pass_local_fn_summary,
23659 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23660 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23661 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23662 make_pass_ipa_fn_summary): New.
23663 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23664 inline_write_summary, inline_free_summary): Remove.
23665 (ipa_free_fn_summary) : New.
23666 * ipa-inline.c (ipa_inline): Update.
23667 (pass_ipa_inline): Do not generate summaries.
23668 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23670 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23671 and add pass_ipa_fn_summary.
23672 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23674 (make_pass_inline_parameters): Remove.
23676 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23678 * omp-low.c (struct omp_context): Remove "default_kind" member.
23681 * omp-offload.c (execute_oacc_device_lower): Remove the
23682 parallelism dimensions function attributes for unparallelized
23683 OpenACC kernels constructs.
23685 2017-05-23 Martin Liska <mliska@suse.cz>
23687 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23689 (cgraph_edge::make_speculative): Likewise.
23690 (cgraph_edge::resolve_speculation): Likewise.
23691 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23692 (cgraph_node::dump): Likewise.
23693 * cgraph.h: Likewise.
23694 * cgraphunit.c (analyze_functions): Likewise.
23695 (symbol_table::compile): Likewise.
23696 * ipa-cp.c (print_all_lattices): Likewise.
23697 (determine_versionability): Likewise.
23698 (initialize_node_lattices): Likewise.
23699 (ipcp_verify_propagated_values): Likewise.
23700 (estimate_local_effects): Likewise.
23701 (update_profiling_info): Likewise.
23702 (create_specialized_node): Likewise.
23703 (perhaps_add_new_callers): Likewise.
23704 (decide_about_value): Likewise.
23705 (decide_whether_version_node): Likewise.
23706 (identify_dead_nodes): Likewise.
23707 (ipcp_store_bits_results): Likewise.
23708 * ipa-devirt.c (dump_targets): Likewise.
23709 (ipa_devirt): Likewise.
23710 * ipa-icf.c (sem_item::dump): Likewise.
23711 (sem_function::equals): Likewise.
23712 (sem_variable::equals): Likewise.
23713 (sem_item_optimizer::read_section): Likewise.
23714 (sem_item_optimizer::execute): Likewise.
23715 (congruence_class::dump): Likewise.
23716 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23717 (dump_inline_summary): Likewise.
23718 (estimate_node_size_and_time): Likewise.
23719 (inline_analyze_function): Likewise.
23720 * ipa-inline-transform.c (inline_call): Likewise.
23721 * ipa-inline.c (report_inline_failed_reason): Likewise.
23722 (want_early_inline_function_p): Likewise.
23723 (edge_badness): Likewise.
23724 (update_edge_key): Likewise.
23725 (inline_small_functions): Likewise.
23726 * ipa-profile.c (ipa_profile): Likewise.
23727 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23728 (ipa_make_edge_direct_to_target): Likewise.
23729 (remove_described_reference): Likewise.
23730 (ipa_impossible_devirt_target): Likewise.
23731 (propagate_controlled_uses): Likewise.
23732 (ipa_print_node_params): Likewise.
23733 (ipcp_transform_function): Likewise.
23734 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23735 (propagate_pure_const): Likewise.
23736 * ipa-reference.c (generate_summary): Likewise.
23737 (read_write_all_from_decl): Likewise.
23738 (propagate): Likewise.
23739 (ipa_reference_read_optimization_summary): Likewise.
23740 * ipa-utils.c (ipa_merge_profiles): Likewise.
23741 * ipa.c (walk_polymorphic_call_targets): Likewise.
23742 (symbol_table::remove_unreachable_nodes): Likewise.
23743 (ipa_single_use): Likewise.
23744 * passes.c (execute_todo): Likewise.
23745 * predict.c (drop_profile): Likewise.
23746 * symtab.c (symtab_node::get_dump_name): New function.
23747 (symtab_node::dump_name): Likewise.
23748 (symtab_node::dump_asm_name): Likewise.
23749 (symtab_node::dump_references): Likewise.
23750 (symtab_node::dump_referring): Likewise.
23751 (symtab_node::dump_base): Likewise.
23752 (symtab_node::debug_symtab): Likewise.
23753 * tree-sra.c (convert_callers_for_node): Likewise.
23754 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23755 * value-prof.c (init_node_map): Likewise.
23757 2017-05-23 Martin Liska <mliska@suse.cz>
23759 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23760 and symtab_node::debug_symtab to symbol_table::debug.
23761 * cgraphunit.c (analyze_functions): Use the renamed function.
23762 (symbol_table::compile): Likewise.
23763 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23764 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23765 * passes.c (execute_todo): Likewise.
23766 * symtab.c (symbol_table::dump): New function.
23767 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23769 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23771 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23772 that nonconst implies exec.
23774 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23776 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23777 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23778 (inline_edge_summary_vec): Turn into ...
23779 (ipa_call_summaries): ... this one.
23780 (redirect_to_unreachable, edge_set_predicate,
23781 evaluate_properties_for_edge, inline_summary_alloc,
23782 reset_ipa_call_summary, reset_inline_summary,
23783 inline_summary_t::duplicate): Update.
23784 (inline_edge_duplication_hook): Turn to ...
23785 (ipa_call_summary_t::duplicate): ... this one.
23786 (inline_edge_removal_hook): Turn to ...
23787 (ipa_call_summary_t::remove): ... this one.
23788 (dump_inline_edge_summary): Turn to ...
23789 (dump_ipa_call_summary): ... this one.
23790 (estimate_function_body_sizes): Update.
23791 (inline_update_callee_summaries): Update.
23792 (remap_edge_change_prob): Update.
23793 (remap_edge_summaries): Update.
23794 (inline_merge_summary): Update.
23795 (do_estimate_edge_time): Update.
23796 (inline_generate_summary): Update.
23797 (inline_read_section): Update.
23798 (inline_read_summary): Update.
23799 (inline_free_summary): Update.
23800 * ipa-inline.c (can_inline_edge_p): Update.
23801 (compute_inlined_call_time): Update.
23802 (want_inline_small_function_p): Update.
23803 (edge_badness): Update.
23804 (early_inliner): Update.
23805 * ipa-inline.h (inline_edge_summary): Turn to ...
23806 (ipa_call_summary): ... this one.
23807 (ipa_call_summary_t): New class.
23808 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23809 (ipa_call_summaries): New.
23810 (inline_edge_summary): Remove.
23811 (estimate_edge_growth): Update.
23812 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23813 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23814 * ipa-split.c (execute_split_functions): Update.
23815 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23817 2017-05-23 Tom de Vries <tom@codesourcery.com>
23819 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23820 attributes): Document rdrand effective target.
23822 2017-05-23 Tom de Vries <tom@codesourcery.com>
23824 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23825 attributes): Sort alphabetically.
23827 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23829 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23831 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23834 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23835 V2DF/V2DI splat into two separate patterns, one that handles
23836 registers, and the other that only handles memory. Drop support
23837 for splatting from a GPR on ISA 2.07 and then splitting the
23838 splat into direct move and splat.
23839 (vsx_splat_<mode>_reg): Likewise.
23840 (vsx_splat_<mode>_mem): Likewise.
23842 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23844 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23846 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23848 PR middle-end/80809
23849 * omp-low.c (finish_taskreg_remap): New function.
23850 (finish_taskreg_scan): If unit size of ctx->record_type
23851 is non-constant, unshare the size expression and replace
23852 decls in it with possible outer var refs.
23854 PR middle-end/80809
23855 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23856 GOVD_SHARED rather than GOVD_PRIVATE with it.
23857 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23858 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23860 PR middle-end/80853
23861 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23862 as last argument to build_outer_var_ref for pointer bases of array
23863 section reductions.
23865 2017-05-19 Martin Sebor <msebor@redhat.com>
23867 * print-tree.c (print_node): Print DECL_READ_P flag.
23869 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23871 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23872 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23873 * cgraph.c: Likewise.
23874 * cgraphunit.c: Likewise.
23875 * gengtype.c: Likewise.
23876 * ipa-cp.c: Likewise.
23877 * ipa-devirt.c: Likewise.
23878 * ipa-icf.c: Likewise.
23879 * ipa-predicate.c: Likewise.
23880 * ipa-profile.c: Likewise.
23881 * ipa-prop.c: Likewise.
23882 * ipa-split.c: Likewise.
23884 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23885 edge_predicate_pool, dump_inline_hints,
23886 inline_summary::account_size_time, redirect_to_unreachable,
23887 edge_set_predicate, set_hint_predicate,
23888 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23889 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23890 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23891 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23892 ipa_call_summary_t::remove, initialize_growth_caches,
23893 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23894 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23895 mark_modified, unmodified_parm_1, unmodified_parm,
23896 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23897 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23898 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23899 will_be_nonconstant_predicate, record_modified_bb_info,
23900 get_minimal_bb, record_modified, param_change_prob,
23901 phi_result_unknown_predicate, predicate_for_phi_result,
23902 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23903 estimate_function_body_sizes, compute_inline_parameters,
23904 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23905 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23906 inline_update_callee_summaries, remap_edge_change_prob,
23907 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23908 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23909 inline_analyze_function, inline_summary_t::insert,
23910 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23911 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23912 inline_free_summary): Move to ipa-fnsummary.h
23913 (predicate_t): Remove.
23914 * ipa-fnsummary.c: New file.
23915 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23916 (enum inline_hints_vals, inline_hints, agg_position_info,
23917 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23918 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23919 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23920 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23921 inline_read_summary, inline_write_summary, inline_free_summary,
23922 inline_analyze_function, initialize_inline_failed,
23923 inline_merge_summary, inline_update_overall_summary,
23924 compute_inline_parameters): Move to ipa-fnsummary.h
23925 * ipa-fnsummary.h: New file.
23926 * ipa-inline-transform.h: Include ipa-inline.h.
23927 * ipa-inline.c: LIkewise.
23929 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23931 * ipa-inline.c (edge_badness): Use inlined_time instead of
23932 inline_summaries->get.
23934 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23936 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23938 2017-05-22 Nathan Sidwell <nathan@acm.org>
23940 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23941 (fdump-lang): Document 'raw' option.
23942 * dumpfile.h (TDI_tu): Delete.
23943 * dumpfile.c (dump_files): Remove translation-unit.
23944 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23946 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23948 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23949 command option from $(AWK) call.
23950 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23951 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23952 [FORMAT]: Remove handling of variable.
23953 * config/avr/t-multilib: Regenerate.
23955 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23957 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23959 (dump_inline_summary): Do not print self_time.
23960 (estimate_function_body_sizes): Do not set self_time.
23961 (compute_inline_parameters): Likewise.
23962 (inline_read_section, inline_write_summary): Do not stream self_time.
23963 * ipa-inline.h (inline_summary): Drop self_time.
23965 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23967 * ipa-inline-analysis.c (account_size_time): Rename to ...
23968 (inline_summary::account_size_time): ... this one.
23969 (reset_ipa_call_summary): Turn to ...
23970 (ipa_call_summary::reset): ... this one.
23971 (reset_inline_summary): Turn to ...
23972 (inline_summary::reset): ... this one.
23973 (inline_summary_t::remove): Update.
23974 (inline_summary_t::duplicate): Update.
23975 (ipa_call_summary_t::remove): Update.
23976 (dump_inline_summary): Update.
23977 (estimate_function_body_sizes): Update.
23978 (compute_inline_parameters): Update.
23979 (estimate_node_size_and_time): Update.
23980 (inline_merge_summary): Update.
23981 (inline_update_overall_summary): Update.
23982 (inline_read_section): Update.
23983 (inline_write_summary): Update.
23984 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23985 add account_size_time and reset member functions.
23986 (ipa_call_summary): Add reset function.
23987 * ipa-predicate.h (predicate::operator &): Constify.
23989 2017-05-22 Richard Biener <rguenther@suse.de>
23991 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23993 2017-05-19 Jason Merrill <jason@redhat.com>
23995 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23997 2017-05-19 Marek Polacek <polacek@redhat.com>
24000 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24001 TYPE_OVERFLOW_WRAPS checks.
24003 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24005 * tree-core.h (enum omp_clause_default_kind): Add
24006 "OMP_CLAUSE_DEFAULT_PRESENT".
24007 * tree-pretty-print.c (dump_omp_clause): Handle it.
24008 * gimplify.c (enum gimplify_omp_var_data): Add
24009 "GOVD_MAP_FORCE_PRESENT".
24010 (gimplify_adjust_omp_clauses_1): Map it to
24011 "GOMP_MAP_FORCE_PRESENT".
24012 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24014 * gimplify.c (oacc_default_clause): Clarify.
24016 2017-05-19 Nathan Sidwell <nathan@acm.org>
24018 LANG_HOOK_REGISTER_DUMPS
24019 * toplev.c (general_init): Call register dump lang hook.
24020 * doc/invoke.texi: Document -fdump-lang option family.
24021 * dumpfile.c (dump_files): Remove class dump here.
24022 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24023 * dumpfile.h (tree_dump_index): Remove TDI_class.
24024 * langhooks-def.h (lhd_register_dumps): Declare.
24025 (LANG_HOOKS_REGISTER_DUMPS): Define.
24026 (LANG_HOOKS_INITIALIZER): Add it.
24027 * langhooks.c (lhd_register_dumps): Define.
24028 * langhooks.h (struct lang_hooks): Add register_dumps.
24030 2017-05-19 Nathan Sidwell <nathan@acm.org>
24032 * context.h (context::set_passes): New.
24033 * context.c (context::context): Do not create pass manager.
24034 * toplev.c (general_init): Create pass manager here.
24036 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24038 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24039 use this splitter if two add or or instructions would also work for
24040 the constant we want to generate.
24042 2017-05-19 Richard Biener <rguenther@suse.de>
24045 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24046 predicate evaluation.
24048 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24050 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24052 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24055 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24057 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24058 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24059 (inline_edge_summary_vec): Turn into ...
24060 (ipa_call_summaries): ... this one.
24061 (redirect_to_unreachable, edge_set_predicate,
24062 evaluate_properties_for_edge, inline_summary_alloc,
24063 reset_ipa_call_summary, reset_inline_summary,
24064 inline_summary_t::duplicate): Update.
24065 (inline_edge_duplication_hook): Turn to ...
24066 (ipa_call_summary_t::duplicate): ... this one.
24067 (inline_edge_removal_hook): Turn to ...
24068 (ipa_call_summary_t::remove): ... this one.
24069 (dump_inline_edge_summary): Turn to ...
24070 (dump_ipa_call_summary): ... this one.
24071 (estimate_function_body_sizes): Update.
24072 (inline_update_callee_summaries): Update.
24073 (remap_edge_change_prob): Update.
24074 (remap_edge_summaries): Update.
24075 (inline_merge_summary): Update.
24076 (do_estimate_edge_time): Update.
24077 (inline_generate_summary): Update.
24078 (inline_read_section): Update.
24079 (inline_read_summary): Update.
24080 (inline_free_summary): Update.
24081 * ipa-inline.c (can_inline_edge_p): Update.
24082 (compute_inlined_call_time): Update.
24083 (want_inline_small_function_p): Update.
24084 (edge_badness): Update.
24085 (early_inliner): Update.
24086 * ipa-inline.h (inline_edge_summary): Turn to ...
24087 (ipa_call_summary): ... this one.
24088 (ipa_call_summary_t): New class.
24089 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24090 (ipa_call_summaries): New.
24091 (inline_edge_summary): Remove.
24092 (estimate_edge_growth): Update.
24093 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24094 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24095 * ipa-split.c (execute_split_functions): Update.
24096 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24098 2017-05-19 Richard Biener <rguenther@suse.de>
24100 PR middle-end/80764
24101 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24103 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24105 * config/rs6000/rs6000.c (struct machine_function): Add field
24106 fpr_is_wrapped_separately.
24107 (rs6000_get_separate_components): Use 64 components. Handle the
24108 new FPR components.
24109 (rs6000_components_for_bb): Handle the FPR components.
24110 (rs6000_emit_prologue_components): Handle the FPR components.
24111 (rs6000_emit_epilogue_components): Handle the FPR components.
24112 (rs6000_set_handled_components): Handle the FPR components.
24113 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24114 that are already separately shrink-wrapped.
24115 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24116 that are already separately shrink-wrapped.
24118 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24121 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24124 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24125 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24126 cases where the register allocator uses a move and an offsettable
24127 memory operation to/from a FPR register on ISA 2.06/2.07.
24128 (define_peephole2 for Altivec d-form store): Likewise.
24130 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24133 * config/i386/mmx.md (*mov<mode>_internal): Enable
24134 alternatives 11, 12, 13 and 14 also for 32bit targets.
24135 Remove alternatives 15, 16, 17 and 18.
24136 * config/i386/sse.md (vec_concatv2di): Change
24137 alternative (!x, *y) to (x, ?!*Yn).
24139 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24141 * dumpfile.h (enum dump_kind): Remove stray comma.
24143 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24145 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24146 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24147 predicate::num_conditions
24148 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24149 (CHANGED): turn into predicate::changed.
24150 (agg_position_info): Move to ipa-predicate.h
24151 (add_condition, predicate::add_clause, predicate::operator &=,
24152 predicate::or_with, predicate::evaluate, predicate::probability,
24153 dump_condition, dump_clause, predicate::dump,
24154 predicate::remap_after_duplication, predicate::remap_after_inlining,
24155 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24156 (evaluate_conditions_for_known_args): Update.
24157 (set_cond_stmt_execution_predicate): Update.
24158 * ipa-inline.h: Include ipa-predicate.h
24159 (condition, inline_param_summary, conditions, agg_position_info,
24160 predicate): Move to ipa-predicate.h
24161 * ipa-predicate.c: New file.
24162 * ipa-predicate.h: New file.
24164 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24166 * final.c (leaf_function_p): Check we are not in a sequence.
24168 2017-05-18 Martin Liska <mliska@suse.cz>
24170 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24171 * dumpfile.c (dump_register): Use new enum dump_kind.
24172 (get_dump_file_name): Likewise.
24173 (dump_enable_all): Likewise.
24174 (dump_switch_p_1): Likewise.
24175 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24176 * dumpfile.h (enum dump_kind): New enum type.
24177 (struct dump_file_info): Create constructor and
24178 format fields and comments.
24179 * passes.c (pass_manager::register_one_dump_file):
24181 * statistics.c (statistics_early_init): Likewise.
24182 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24183 TDF_TREE with TDF_SLIM.
24184 (gather_memory_references_ref): Likewise.
24186 2017-05-18 Martin Liska <mliska@suse.cz>
24188 * vec.h (struct vnull): Use it.
24190 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24192 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24193 (true_predicate, false_predicate, true_predicate_p,
24194 false_predicate_p): Remove.
24195 (single_cond_predicate, not_inlined_predicate): Turn to member function
24197 (add_condition): Update.
24198 (add_clause): Turn to...
24199 (predicate::add_clause): ... this one; update; allow passing NULL
24201 (and_predicates): Turn to ...
24202 (predicate::operator &=): ... this one.
24203 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24204 (or_predicates): Turn to ...
24205 (predicate::or_with): ... this one.
24206 (evaluate_predicate): Turn to ...
24207 (predicate::evaluate): ... this one.
24208 (predicate_probability): Turn to ...
24209 (predicate::probability): ... this one.
24210 (dump_condition): Update.
24211 (dump_predicate): Turn to ...
24212 (predicate::dump): ... this one.
24213 (account_size_time): Update.
24214 (edge_set_predicate): Update.
24215 (set_hint_predicate): UPdate.
24216 (evaluate_conditions_for_known_args): Update.
24217 (evaluate_properties_for_edge): Update.
24218 (remap_predicate_after_duplication): Turn to...
24219 (predicate::remap_after_duplication): ... this one.
24220 (remap_hint_predicate_after_duplication): Update.
24221 (inline_summary_t::duplicate): UPdate.
24222 (dump_inline_edge_summary): Update.
24223 (dump_inline_summary): Update.
24224 (set_cond_stmt_execution_predicate): Update.
24225 (set_switch_stmt_execution_predicate): Update.
24226 (compute_bb_predicates): Update.
24227 (will_be_nonconstant_expr_predicate): Update.
24228 (will_be_nonconstant_predicate): Update.
24229 (phi_result_unknown_predicate): Update.
24230 (predicate_for_phi_result): Update.
24231 (array_index_predicate): Update.
24232 (estimate_function_body_sizes): Update.
24233 (estimate_node_size_and_time): Update.
24234 (estimate_ipcp_clone_size_and_time): Update.
24235 (remap_predicate): Rename to ...
24236 (predicate::remap_after_inlining): ... this one.
24237 (remap_hint_predicate): Update.
24238 (inline_merge_summary): Update.
24239 (inline_update_overall_summary): Update.
24240 (estimate_size_after_inlining): Update.
24241 (read_predicate): Rename to ...
24242 (predicate::stream_in): ... this one.
24243 (read_inline_edge_summary): Update.
24244 (write_predicate): Rename to ...
24245 (predicate::stream_out): ... this one.
24246 (write_inline_edge_summary): Update.
24247 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24248 (clause_t): Turn to uint32_t
24249 (predicate): Turn to class; implement constructor and operators
24251 (size_time_entry): Update.
24252 (inline_summary): Update.
24253 (inline_edge_summary): Update.
24255 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24257 * fold-const.c (fold_binary_loc): Move transformation...
24258 * match.pd (C - X CMP X): ... here.
24260 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24262 * config/sparc/sparc.c (sparc_option_override): Set function
24263 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24264 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24266 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24268 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24270 2017-05-18 Marek Polacek <polacek@redhat.com>
24273 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24274 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24275 gimple_assign_load_p.
24277 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24279 PR middle-end/80692
24280 * real.c (do_compare): Give decimal_do_compare preference over
24281 comparing just the signs.
24283 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24285 * doc/md.texi (Canonicalization of Instructions): Describe the
24286 canonical form of instructions that inherently set a condition
24289 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24291 PR middle-end/80775
24292 * tree-cfg.c: Move deletion of unreachable case statements to after
24293 the merging of consecutive case labels.
24295 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24297 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24298 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24299 restoring of callee-saved registers.
24301 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24303 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24304 * config/visium/visium.c (single_set_and_flags): Likewise.
24305 * config/visium/visium.md (Substitutions): Likewise.
24307 2017-05-17 Martin Liska <mliska@suse.cz>
24309 * cfg.c: Introduce dump_flags_t type and
24310 use it instead of int type.
24312 * cfghooks.c: Likewise.
24313 * cfghooks.h (struct cfg_hooks): Likewise.
24314 * cfgrtl.c: Likewise.
24315 * cfgrtl.h: Likewise.
24316 * cgraph.c (cgraph_node::get_body): Likewise.
24317 * coretypes.h: Likewise.
24318 * domwalk.c: Likewise.
24319 * domwalk.h: Likewise.
24320 * dumpfile.c (struct dump_option_value_info): Likewise.
24321 (dump_enable_all): Likewise.
24322 (dump_switch_p_1): Likewise.
24323 (opt_info_switch_p): Likewise.
24324 * dumpfile.h (enum tree_dump_index): Likewise.
24325 (struct dump_file_info): Likewise.
24326 * genemit.c: Likewise.
24327 * generic-match-head.c: Likewise.
24328 * gengtype.c (open_base_files): Likewise.
24329 * gimple-pretty-print.c: Likewise.
24330 * gimple-pretty-print.h: Likewise.
24331 * graph.c (print_graph_cfg): Likewise.
24332 * graphite-scop-detection.c (dot_all_sese): Likewise.
24333 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24334 * loop-unroll.c (report_unroll): Likewise.
24335 * passes.c (pass_manager::register_one_dump_file): Likewise.
24336 * print-tree.c: Likewise.
24337 * statistics.c: Likewise.
24338 * tree-cfg.c: Likewise.
24339 * tree-cfg.h: Likewise.
24340 * tree-dfa.c: Likewise.
24341 * tree-dfa.h: Likewise.
24342 * tree-dump.c (dump_function): Likewise.
24343 * tree-dump.h (struct dump_info): Likewise.
24344 * tree-pretty-print.c: Likewise.
24345 * tree-pretty-print.h: Likewise.
24346 * tree-ssa-live.c: Likewise.
24347 * tree-ssa-live.h: Likewise.
24348 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24349 * tree-vect-loop.c: Likewise.
24350 * tree-vect-slp.c: Likewise.
24352 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24353 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24355 PR tree-optimization/80457
24356 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24357 of all arguments to a statement as scalar_to_vec operations.
24358 (vectorizable_call): Adjust call to vect_model_simple_cost for
24360 (vectorizable_conversion): Likewise.
24361 (vectorizable_assignment): Likewise.
24362 (vectorizable_shift): Likewise.
24363 (vectorizable_operation): Likewise.
24364 (vectorizable_comparison): Likewise.
24365 (vect_is_simple_cond): Record the def types for operands.
24366 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24367 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24368 for statement argument count.
24370 2017-05-16 Carl Love <cel@us.ibm.com>
24372 * config/rs6000/rs6000-c: Add support for built-in functions
24373 vector unsigned long long vec_bperm (vector unsigned long long,
24374 vector unsigned char)
24375 vector signed long long vec_mule (vector signed int,
24377 vector unsigned long long vec_mule (vector unsigned int,
24378 vector unsigned int)
24379 vector signed long long vec_mulo (vector signed int,
24381 vector unsigned long long vec_mulo (vector unsigned int,
24382 vector unsigned int)
24383 vector signed char vec_sldw (vector signed char,
24384 vector signed char,
24386 vector unsigned char vec_sldw (vector unsigned char,
24387 vector unsigned char,
24389 vector signed short vec_sldw (vector signed short,
24390 vector signed short,
24392 vector unsigned short vec_sldw (vector unsigned short,
24393 vector unsigned short,
24395 vector signed int vec_sldw (vector signed int,
24398 vector unsigned int vec_sldw (vector unsigned int,
24399 vector unsigned int,
24401 vector signed long long vec_sldw (vector signed long long,
24402 vector signed long long,
24404 vector unsigned long long vec_sldw (vector unsigned long long,
24405 vector unsigned long long,
24407 * config/rs6000/rs6000-c: Add support for built-in functions
24408 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24409 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24410 * doc/extend.texi: Update the built-in documentation for the
24411 new built-in functions.
24413 2017-05-16 Marek Polacek <polacek@redhat.com>
24417 * tree.c (save_expr): Don't fold the expression.
24419 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24421 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24422 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24423 and (?*y,m). Update insn attributes.
24425 2017-05-16 Martin Liska <mliska@suse.cz>
24427 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24428 flags argument of print_gimple_stmt, print_gimple_expr,
24429 print_generic_stmt and print_generic_expr.
24430 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24431 * coretypes.h: Likewise.
24432 * except.c (dump_eh_tree): Likewise.
24433 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24434 * gimple-pretty-print.h: Likewise.
24435 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24436 (backprop::push_to_worklist): Likewise.
24437 (backprop::pop_from_worklist): Likewise.
24438 (backprop::process_use): Likewise.
24439 (backprop::intersect_uses): Likewise.
24440 (note_replacement): Likewise.
24441 * gimple-ssa-store-merging.c
24442 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24443 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24444 (pass_store_merging::execute): Likewise.
24445 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24446 (ssa_base_cand_dump_callback): Likewise.
24447 (dump_incr_vec): Likewise.
24448 (replace_refs): Likewise.
24449 (replace_mult_candidate): Likewise.
24450 (create_add_on_incoming_edge): Likewise.
24451 (create_phi_basis): Likewise.
24452 (insert_initializers): Likewise.
24453 (all_phi_incrs_profitable): Likewise.
24454 (introduce_cast_before_cand): Likewise.
24455 (replace_one_candidate): Likewise.
24456 * gimplify.c (gimplify_expr): Likewise.
24457 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24458 (set_rename): Likewise.
24459 (rename_uses): Likewise.
24460 (copy_loop_phi_nodes): Likewise.
24461 (add_close_phis_to_merge_points): Likewise.
24462 (copy_loop_close_phi_args): Likewise.
24463 (copy_cond_phi_args): Likewise.
24464 (graphite_copy_stmts_from_block): Likewise.
24465 (translate_pending_phi_nodes): Likewise.
24466 * graphite-poly.c (print_pdr): Likewise.
24467 (dump_gbb_cases): Likewise.
24468 (dump_gbb_conditions): Likewise.
24469 (print_scop_params): Likewise.
24470 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24471 (build_cross_bb_scalars_use): Likewise.
24472 (gather_bbs::before_dom_children): Likewise.
24473 * hsa-dump.c (dump_hsa_immed): Likewise.
24474 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24475 (get_replacement_map): Likewise.
24476 * ipa-inline-analysis.c (dump_condition): Likewise.
24477 (estimate_function_body_sizes): Likewise.
24478 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24479 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24480 * ipa-prop.c (ipa_dump_param): Likewise.
24481 (ipa_print_node_jump_functions_for_edge): Likewise.
24482 (ipa_modify_call_arguments): Likewise.
24483 (ipa_modify_expr): Likewise.
24484 (ipa_dump_param_adjustments): Likewise.
24485 (ipa_dump_agg_replacement_values): Likewise.
24486 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24487 * ipa-pure-const.c (check_stmt): Likewise.
24488 (pass_nothrow::execute): Likewise.
24489 * ipa-split.c (execute_split_functions): Likewise.
24490 * omp-offload.c (dump_oacc_loop_part): Likewise.
24491 (dump_oacc_loop): Likewise.
24492 * trans-mem.c (tm_log_emit): Likewise.
24493 (tm_memopt_accumulate_memops): Likewise.
24494 (dump_tm_memopt_set): Likewise.
24495 (dump_tm_memopt_transform): Likewise.
24496 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24497 (print_loop): Likewise.
24498 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24499 (chkp_gather_checks_info): Likewise.
24500 (chkp_get_check_result): Likewise.
24501 (chkp_remove_check_if_pass): Likewise.
24502 (chkp_use_outer_bounds_if_possible): Likewise.
24503 (chkp_reduce_bounds_lifetime): Likewise.
24504 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24505 (chkp_mark_completed_bounds): Likewise.
24506 (chkp_register_incomplete_bounds): Likewise.
24507 (chkp_mark_invalid_bounds): Likewise.
24508 (chkp_maybe_copy_and_register_bounds): Likewise.
24509 (chkp_build_returned_bound): Likewise.
24510 (chkp_get_bound_for_parm): Likewise.
24511 (chkp_build_bndldx): Likewise.
24512 (chkp_get_bounds_by_definition): Likewise.
24513 (chkp_generate_extern_var_bounds): Likewise.
24514 (chkp_get_bounds_for_decl_addr): Likewise.
24515 * tree-chrec.c (chrec_apply): Likewise.
24516 * tree-data-ref.c (dump_data_reference): Likewise.
24517 (dump_subscript): Likewise.
24518 (dump_data_dependence_relation): Likewise.
24519 (analyze_overlapping_iterations): Likewise.
24520 * tree-inline.c (expand_call_inline): Likewise.
24521 (tree_function_versioning): Likewise.
24522 * tree-into-ssa.c (dump_defs_stack): Likewise.
24523 (dump_currdefs): Likewise.
24524 (dump_names_replaced_by): Likewise.
24525 (dump_update_ssa): Likewise.
24526 (update_ssa): Likewise.
24527 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24528 * tree-parloops.c (build_new_reduction): Likewise.
24529 (try_create_reduction_list): Likewise.
24530 (ref_conflicts_with_region): Likewise.
24531 (oacc_entry_exit_ok_1): Likewise.
24532 (oacc_entry_exit_single_gang): Likewise.
24533 * tree-pretty-print.h: Likewise.
24534 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24535 (get_scalar_evolution): Likewise.
24536 (add_to_evolution): Likewise.
24537 (get_loop_exit_condition): Likewise.
24538 (analyze_evolution_in_loop): Likewise.
24539 (analyze_initial_condition): Likewise.
24540 (analyze_scalar_evolution): Likewise.
24541 (instantiate_scev): Likewise.
24542 (number_of_latch_executions): Likewise.
24543 (gather_chrec_stats): Likewise.
24544 (final_value_replacement_loop): Likewise.
24545 (scev_const_prop): Likewise.
24546 * tree-sra.c (dump_access): Likewise.
24547 (disqualify_candidate): Likewise.
24548 (create_access): Likewise.
24549 (reject): Likewise.
24550 (maybe_add_sra_candidate): Likewise.
24551 (create_access_replacement): Likewise.
24552 (analyze_access_subtree): Likewise.
24553 (analyze_all_variable_accesses): Likewise.
24554 (sra_modify_assign): Likewise.
24555 (initialize_constant_pool_replacements): Likewise.
24556 (find_param_candidates): Likewise.
24557 (decide_one_param_reduction): Likewise.
24558 (replace_removed_params_ssa_names): Likewise.
24559 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24560 * tree-ssa-copy.c (dump_copy_of): Likewise.
24561 (copy_prop_visit_cond_stmt): Likewise.
24562 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24563 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24564 (record_equivalences_from_stmt): Likewise.
24565 * tree-ssa-dse.c (compute_trims): Likewise.
24566 (delete_dead_call): Likewise.
24567 (delete_dead_assignment): Likewise.
24568 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24569 (forward_propagate_into_cond): Likewise.
24570 (pass_forwprop::execute): Likewise.
24571 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24572 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24574 (move_computations_worker): Likewise.
24575 (execute_sm): Likewise.
24576 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24577 (remove_exits_and_undefined_stmts): Likewise.
24578 (remove_redundant_iv_tests): Likewise.
24579 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24580 (adjust_iv_update_pos): Likewise.
24581 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24582 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24583 (value_replacement): Likewise.
24584 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24585 * tree-ssa-pre.c (print_pre_expr): Likewise.
24586 (get_representative_for): Likewise.
24587 (create_expression_by_pieces): Likewise.
24588 (insert_into_preds_of_block): Likewise.
24589 (eliminate_insert): Likewise.
24590 (eliminate_dom_walker::before_dom_children): Likewise.
24591 (eliminate): Likewise.
24592 (remove_dead_inserted_code): Likewise.
24593 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24594 * tree-ssa-reassoc.c (get_rank): Likewise.
24595 (eliminate_duplicate_pair): Likewise.
24596 (eliminate_plus_minus_pair): Likewise.
24597 (eliminate_not_pairs): Likewise.
24598 (undistribute_ops_list): Likewise.
24599 (eliminate_redundant_comparison): Likewise.
24600 (update_range_test): Likewise.
24601 (optimize_range_tests_var_bound): Likewise.
24602 (optimize_vec_cond_expr): Likewise.
24603 (rewrite_expr_tree): Likewise.
24604 (rewrite_expr_tree_parallel): Likewise.
24605 (linearize_expr): Likewise.
24606 (break_up_subtract): Likewise.
24607 (linearize_expr_tree): Likewise.
24608 (attempt_builtin_powi): Likewise.
24609 (attempt_builtin_copysign): Likewise.
24610 (transform_stmt_to_copy): Likewise.
24611 (transform_stmt_to_multiply): Likewise.
24612 (dump_ops_vector): Likewise.
24613 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24614 (print_scc): Likewise.
24615 (set_ssa_val_to): Likewise.
24616 (visit_reference_op_store): Likewise.
24617 (visit_use): Likewise.
24618 (sccvn_dom_walker::before_dom_children): Likewise.
24619 (run_scc_vn): Likewise.
24620 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24622 (expr_hash_elt::print): Likewise.
24623 (const_and_copies::pop_to_marker): Likewise.
24624 (const_and_copies::record_const_or_copy_raw): Likewise.
24625 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24626 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24627 (dump_predicates): Likewise.
24628 (find_uninit_use): Likewise.
24629 (warn_uninitialized_phi): Likewise.
24630 (pass_late_warn_uninitialized::execute): Likewise.
24631 * tree-ssa.c (verify_vssa): Likewise.
24632 (verify_ssa): Likewise.
24633 (maybe_optimize_var): Likewise.
24634 * tree-vrp.c (dump_value_range): Likewise.
24635 (dump_all_value_ranges): Likewise.
24636 (dump_asserts_for): Likewise.
24637 (register_edge_assert_for_2): Likewise.
24638 (vrp_visit_cond_stmt): Likewise.
24639 (vrp_visit_switch_stmt): Likewise.
24640 (vrp_visit_stmt): Likewise.
24641 (vrp_visit_phi_node): Likewise.
24642 (simplify_cond_using_ranges_1): Likewise.
24643 (fold_predicate_in): Likewise.
24644 (evrp_dom_walker::before_dom_children): Likewise.
24645 (evrp_dom_walker::push_value_range): Likewise.
24646 (evrp_dom_walker::pop_value_range): Likewise.
24647 (execute_early_vrp): Likewise.
24649 2017-05-16 Richard Biener <rguenther@suse.de>
24651 * dwarf2out.c (loc_list_from_tree_1): Do not create
24652 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24654 2017-05-16 Richard Biener <rguenther@suse.de>
24656 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24658 (note_variable_value_in_expr): If we resolved the decl ref
24659 do not push to the stack.
24661 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24663 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24664 operations in fast-math mode.
24665 (vaddq_f16): Likewise.
24666 (vmul_f16): Likewise.
24667 (vmulq_f16): Likewise.
24668 (vsub_f16): Likewise.
24669 (vsubq_f16): Likewise.
24670 * config/arm/neon.md (add<mode>3): New.
24672 (fma:<VH:mode>3): New. Also remove outdated comment.
24675 2017-05-16 Martin Liska <mliska@suse.cz>
24679 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24680 (odr_types_equivalent_p): Likewise.
24682 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24684 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24686 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24689 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24690 non-interunit SSE move alternatives with '?'.
24691 (zero-extendsidi peephole2): New peephole to skip intermediate
24692 general register in SSE zero-extend sequence.
24694 2017-05-15 Jeff Law <law@redhat.com>
24696 * reorg.c (relax_delay_slots): Create a new variable to hold
24697 the temporary target rather than clobbering TARGET_LABEL.
24699 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24700 missing argument to extract_bit_field call.
24701 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24703 2017-05-15 Martin Liska <mliska@suse.cz>
24706 * gcc.c (process_command): Do not allow empty argument of -o option.
24708 2017-05-15 Renlin Li <renlin.li@arm.com>
24710 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24711 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24712 * config/aarch64/constraints.md (Usf): Add long call check.
24713 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24714 (call_value): Likewise.
24715 (sibcall): Likewise.
24716 (sibcall_value): Likewise.
24718 (call_value_insn): New.
24719 (sibcall_insn): Update rtx pattern.
24720 (sibcall_value_insn): Likewise.
24721 (call_internal): Remove.
24722 (call_value_internal): Likewise.
24723 (sibcall_internal): Likewise.
24724 (sibcall_value_internal): Likewise.
24725 (call_reg): Likewise.
24726 (call_symbol): Likewise.
24727 (call_value_reg): Likewise.
24728 (call_value_symbol): Likewise.
24730 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24733 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24735 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24737 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24738 compatible with CCGOCmode and with CCZmode.
24740 2017-05-14 Martin Sebor <msebor@redhat.com>
24742 PR middle-end/77671
24743 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24744 (gimple_fold_builtin_snprintf): Same.
24745 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24746 (gimple_fold_builtin_snprintf): Same.
24747 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24748 of character types.
24749 (is_call_safe): New function.
24750 (try_substitute_return_value): Call it.
24751 (try_simplify_call): New function.
24752 (pass_sprintf_length::handle_gimple_call): Call it.
24754 2017-05-14 Martin Sebor <msebor@redhat.com>
24756 PR middle-end/80669
24757 * builtins.c (expand_builtin_stpncpy): Simplify.
24759 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24761 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24762 * config/i386/i386.h
24763 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24764 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24765 (struct machine_function): Add new members call_ms2sysv,
24766 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24767 (struct machine_frame_state): New fields sp_realigned and
24768 sp_realigned_offset.
24769 * config/i386/i386.c
24770 (enum xlogue_stub): New enum.
24771 (enum xlogue_stub_sets): New enum.
24772 (class xlogue_layout): New class.
24773 (struct ix86_frame): New fields stack_realign_allocate_offset,
24774 stack_realign_offset and outlined_save_offset. Modify comments to
24775 detail stack layout when using out-of-line stubs.
24776 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24777 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24778 -mcall-ms2sysv-xlogues.
24779 (stub_managed_regs): New static variable.
24780 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24781 registers managed by out-of-line stub.
24782 (disable_call_ms2sysv_xlogues): New function.
24783 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24784 m->call_ms2sysv when appropriate and compute frame layout for
24786 (sp_valid_at, fp_valid_at): New inline functions.
24787 (choose_basereg): New function.
24788 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24790 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24791 Use align parameter of choose_baseaddr to generated aligned SSE movs
24793 (pro_epilogue_adjust_stack): Modify to track
24794 machine_frame_state::sp_realigned.
24795 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24796 (ix86_nsaved_sseregs): Likewise.
24797 (ix86_emit_save_regs): Likewise.
24798 (ix86_emit_save_regs_using_mov): Likewise.
24799 (ix86_emit_save_sse_regs_using_mov): Likewise.
24800 (get_scratch_register_on_entry): Likewise.
24801 (gen_frame_set): New function.
24802 (gen_frame_load): Likewise.
24803 (gen_frame_store): Likewise.
24804 (emit_outlined_ms2sysv_save): Likewise.
24805 (emit_outlined_ms2sysv_restore): Likewise.
24806 (ix86_expand_prologue): Modify stack re-alignment code and call
24807 emit_outlined_ms2sysv_save when appropriate.
24808 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24809 parameter rtx_insn *insn, which allows the function to be used to only
24810 generate the notes.
24811 (ix86_expand_epilogue): Modify validity checks of frame and stack
24812 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24813 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24814 * config/i386/predicates.md
24815 (save_multiple): New predicate.
24816 (restore_multiple): Likewise.
24817 * config/i386/sse.md
24818 (save_multiple<mode>): New pattern.
24819 (save_multiple_realign<mode>): Likewise.
24820 (restore_multiple<mode>): Likewise.
24821 (restore_multiple_and_return<mode>): Likewise.
24822 (restore_multiple_leave_return<mode>): Likewise.
24823 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24825 2017-05-14 Julia Koval <julia.koval@intel.com>
24827 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24828 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24829 (__builtin_ia32_xsetbv): New builtins.
24830 * config/i386/i386.c (ix86_expand_special_args_builtin):
24832 (ix86_expand_builtin): Special expand for new intrinsics.
24833 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24834 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24835 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24837 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24839 * cfganal.c (inverted_post_order_compute): Change argument type
24841 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24842 * df-core.c (rest_of_handle_df_initialize): Adjust.
24843 (rest_of_handle_df_finish): Likewise.
24844 (df_analyze_1): Likewise.
24845 (df_analyze): Likewise.
24846 (loop_inverted_post_order_compute): Change argument to be a vec *.
24847 (df_analyze_loop): Adjust.
24848 (df_get_n_blocks): Likewise.
24849 (df_get_postorder): Likewise.
24850 * df.h (struct df_d): Change field to be a vec.
24851 * lcm.c (compute_laterin): Adjust.
24852 (compute_available): Likewise.
24853 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24854 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24855 * tree-ssa-pre.c (compute_antic): Likewise.
24857 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24859 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24860 (depth_first_search::depth_first_search): Change structure init
24861 function to this constructor.
24862 (depth_first_search::add_bb): Rename function to this member.
24863 (depth_first_search::execute): Likewise.
24864 (flow_dfs_compute_reverse_finish): Adjust.
24866 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24868 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24869 (longest_simple_path): Likewise.
24870 * shrink-wrap.c (spread_components): Likewise.
24871 (disqualify_problematic_components): Likewise.
24872 (emit_common_heads_for_components): Likewise.
24873 (emit_common_tails_for_components): Likewise.
24874 (insert_prologue_epilogue_for_components): Likewise.
24876 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24878 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24881 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24883 * df-core.c (df_set_blocks): Start using auto_bitmap.
24884 (df_compact_blocks): Likewise.
24885 * df-problems.c (df_rd_confluence_n): Likewise.
24886 * df-scan.c (df_insn_rescan_all): Likewise.
24887 (df_process_deferred_rescans): Likewise.
24888 (df_update_entry_block_defs): Likewise.
24889 (df_update_exit_block_uses): Likewise.
24890 (df_entry_block_bitmap_verify): Likewise.
24891 (df_exit_block_bitmap_verify): Likewise.
24892 (df_scan_verify): Likewise.
24893 * lra-constraints.c (lra_constraints): Likewise.
24894 (undo_optional_reloads): Likewise.
24895 (lra_undo_inheritance): Likewise.
24896 * lra-remat.c (calculate_gen_cands): Likewise.
24897 (do_remat): Likewise.
24898 * lra-spills.c (assign_spill_hard_regs): Likewise.
24899 (spill_pseudos): Likewise.
24900 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24901 (bitmap_set_subtract_values): Likewise.
24903 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24905 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24906 management with auto_bitmap.
24907 (fix_inter_tick): Likewise.
24908 (fix_recovery_deps): Likewise.
24909 * ira.c (add_store_equivs): Likewise.
24910 (find_moveable_pseudos): Likewise.
24911 (split_live_ranges_for_shrink_wrap): Likewise.
24912 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24913 (rtx_reuse_manager::seen_def_p): Likewise.
24914 (rtx_reuse_manager::set_seen_def): Likewise.
24915 * print-rtl.h (class rtx_reuse_manager): Likewise.
24917 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24919 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24921 (migrate_btr_def): Likewise.
24922 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24923 * df-core.c (loop_post_order_compute): Likewise.
24924 (loop_inverted_post_order_compute): Likewise.
24925 * hsa-common.h: Likewise.
24926 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24927 * init-regs.c (initialize_uninitialized_regs): Likewise.
24928 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24929 (inline_small_functions): Likewise.
24930 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24931 * ira.c (combine_and_move_insns): Likewise.
24932 (build_insn_chain): Likewise.
24933 * loop-invariant.c (find_invariants): Likewise.
24934 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24935 * predict.c (tree_predict_by_opcode): Likewise.
24936 (predict_paths_leading_to): Likewise.
24937 (predict_paths_leading_to_edge): Likewise.
24938 (estimate_loops_at_level): Likewise.
24939 (estimate_loops): Likewise.
24940 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24941 (spread_components): Likewise.
24942 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24943 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24944 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24945 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24946 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24947 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24948 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24949 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24950 (mark_threaded_blocks): Likewise.
24951 (thread_through_all_blocks): Likewise.
24952 * tree-ssa.c (verify_ssa): Likewise.
24953 (execute_update_addresses_taken): Likewise.
24954 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24956 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24958 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24960 (post_order_compute): Likewise.
24961 (inverted_post_order_compute): Likewise.
24962 (pre_and_rev_post_order_compute_fn): Likewise.
24964 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24966 * genrecog.c (int_set::int_set): Explicitly construct our
24967 auto_vec base class.
24968 * vec.h (auto_vec::auto_vec): New constructor.
24970 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24972 * bitmap.h (class auto_bitmap): New constructor taking
24973 bitmap_obstack * argument.
24975 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24977 * bitmap.h (class auto_bitmap): Change type of m_bits to
24978 bitmap_head, and adjust ctor / dtor and member operators.
24980 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24982 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24983 when returned register mode doesn't match original mode.
24985 2017-05-12 Jeff Law <law@redhat.com>
24986 Jakub Jelinek <jakub@redhat.com>
24988 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24989 we look for cc setter after the compare-elim changes.
24990 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24991 within the vector to match what compare-elim now expects.
24992 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24993 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24995 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24996 after the compare-elim changes.
24997 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24998 the vector to match what compare-elim now expects.
24999 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25000 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25001 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25002 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25003 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25005 * config/visium/visium.c (single_set_and_flags): Fix where
25006 we look for cc setter after the compare-elim changes.
25007 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25008 with the vector to match what compare-elim now expects.
25009 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25010 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25011 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25012 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25013 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25014 (neg<mode>2_insn_set_overflow): Likewise.
25016 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25018 PR middle-end/79794
25019 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25020 maybe_expand_insn call, set ops[0].target. If still set after call,
25021 set alt_rtl. Add extra arg to recursive calls.
25022 (extract_bit_field): Add alt_rtl argument. Pass to
25024 * expmed.h (extract_bit_field): Fix prototype.
25025 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25026 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25027 to extract_bit_field_calls.
25028 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25029 Pass alt_rtl to extract_bit_field calls.
25030 * calls.c (store_unaligned_arguments_into_psuedos)
25031 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25032 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25034 * optabs.h (struct expand_operand): Add target bitfield.
25036 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25038 * compare-elim.c (try_eliminate_compare): Canonicalize
25039 operation with embedded compare to
25040 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25041 (set (reg) (operation)].
25043 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25045 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25048 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25049 cost of adding a carry flag for ADC instruction.
25050 [case MINUS]: Ignore the cost of subtracting a carry flag
25051 for SBB instruction.
25053 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25055 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25057 * config/rs6000/bmiintrin.h: New file.
25058 * config/rs6000/bmi2intrin.h: New file.
25059 * config/rs6000/x86intrin.h: New file.
25061 2017-05-12 Jeff Law <law@redhat.com>
25063 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25066 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25068 PR middle-end/80707
25069 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25071 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25073 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25074 early expansion of vector divide builtins.
25075 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25076 builtins identified as having unsigned arguments.
25078 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25080 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25081 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25082 expansion of vector logical operations (and, andc, or, xor,
25085 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25087 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25088 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25090 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25092 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25093 early GIMPLE expansion of vector multiplies.
25095 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25097 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25098 TARGET_HAVE_MOVT conditional.
25099 (movt splitter): Likewise.
25101 2017-05-12 Richard Biener <rguenther@suse.de>
25103 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25104 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25105 Fold all stmts not inplace.
25107 2017-05-12 Richard Biener <rguenther@suse.de>
25109 PR tree-optimization/80713
25110 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25111 inserted_exprs bit for not removed stmts.
25113 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25115 PR middle-end/69921
25116 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25117 parallelized" attribute for parallelized OpenACC kernels.
25118 * omp-offload.c (execute_oacc_device_lower): Use it.
25120 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25121 Set "oacc kernels" attribute.
25122 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25123 parameter. Adjust all users.
25124 (oacc_fn_attrib_kernels_p): Remove function.
25125 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25126 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25127 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25128 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25129 assert "oacc kernels" attribute is set.
25131 2017-05-11 Carl Love <cel@us.ibm.com>
25133 * config/rs6000/rs6000-c: Add support for built-in functions
25134 vector unsigned char vec_popcnt (vector signed char)
25135 vector unsigned char vec_popcnt (vector unsigned char)
25136 vector unsigned short vec_popcnt (vector signed short)
25137 vector unsigned short vec_popcnt (vector unsigned short)
25138 vector unsigned int vec_popcnt (vector signed int)
25139 vector unsigned int vec_popcnt (vector unsigned int)
25140 vector unsigned long long vec_popcnt (vector signed long long)
25141 vector unsigned long long vec_popcnt (vector unsigned long long)
25142 vector signed long long vec_slo (vector signed long long,
25143 vector signed char)
25144 vector signed long long vec_slo (vector signed long long,
25145 vector unsigned char)
25146 vector unsigned long long vec_slo (vector unsigned long long,
25147 vector signed char)
25148 vector unsigned long long vec_slo (vector unsigned long long,
25149 vector unsigned char)
25150 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25151 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25152 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25153 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25154 * doc/extend.texi: Update the built-in documentation file for the
25155 new built-in functions.
25157 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25159 * attribs.h (sorted_attr_string): Move machine independent
25160 functions for target clone support from the i386 port to common
25161 code. Rename ix86_function_versions to common_function_versions.
25162 Rename make_name to make_unique_name.
25163 (common_function_versions): Likewise.
25164 (make_unique_name): Likewise.
25165 (make_dispatcher_decl): Likewise.
25166 (is_function_default_version): Likewise.
25167 * attribs.c (attr_strcmp): Likewise.
25168 (sorted_attr_string): Likewise.
25169 (common_function_versions): Likewise.
25170 (make_unique_name): Likewise.
25171 (make_dispatcher_decl): Likewise.
25172 (is_function_default_version): Likewise.
25173 * config/i386/i386.c (attr_strcmp): Likewise.
25174 (sorted_attr_string): Likewise.
25175 (ix86_function_versions): Likewise.
25176 (make_name): Likewise.
25177 (make_dispatcher_decl): Likewise.
25178 (is_function_default_version): Likewise.
25179 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25181 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25184 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25185 Account for direct move costs for vec_construct of integer
25188 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25191 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25192 (UNSPEC_STX_ATOMIC): Ditto.
25193 (loaddi_via_sse): New insn.
25194 (storedi_via_sse): Ditto.
25195 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25196 Update corresponding peephole2 patterns.
25197 (atomic_storedi_fpu): Ditto.
25199 2017-05-11 Julia Koval <julia.koval@intel.com>
25201 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25202 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25204 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25205 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25206 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25208 2017-05-11 Nathan Sidwell <nathan@acm.org>
25210 * graphite-poly.c: Include dumpfile.h.
25212 * dumpfle.h (dump_function): Declare here ...
25213 * tree-dump.h (dump_function): ... not here.
25214 * dumpfile.c: #include tree-cfg.h.
25215 (dump_function): Move here from ...
25216 * tree-dump.c (dump_function): ... here.
25217 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25218 * graphite-poly.c: Don't include tree-dump.h.
25219 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25220 * print-tree.c: Likewise.
25221 * stor-layout.c: Likewise.
25222 * tree-nested.c: Likewise.
25224 * dumpfile.c (dump_start): Use TDF_FLAGS.
25225 (dump_enable_all): Fix TDF_KIND check thinko.
25227 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25229 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25230 array entries to represent two legal parameterizations of the
25231 overloaded __builtin_cmpb function, as represented by the
25232 P6_OV_BUILTIN_CMPB constant.
25233 (altivec_resolve_overloaded_builtin): Add special case handling
25234 for the __builtin_cmpb function, as represented by the
25235 P6_OV_BUILTIN_CMPB constant.
25236 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25237 (BU_P6_64BIT_2): New macro.
25238 (BU_P6_OVERLOAD_2): New macro
25239 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25240 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25241 (CMPB): Add overload support to represent both 32-bit and 64-bit
25242 compare-bytes function.
25243 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25244 support for TARGET_CMPB.
25245 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25246 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25247 documentation of the __builtin_cmpb overloaded built-in function.
25249 2017-05-11 Richard Biener <rguenther@suse.de>
25251 PR tree-optimization/80705
25252 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25253 bases are not vectorizable.
25255 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25257 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25258 when counting register pressure.
25260 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25262 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25263 (iv_ca_more_deps): Renamed to ...
25264 (iv_ca_compare_deps): ... this.
25265 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25267 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25269 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25271 (determine_group_iv_costs): ... here.
25272 (find_inv_vars_cb): Record inv var if it's not recorded before.
25274 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25276 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25277 (get_shiftadd_cost): Ditto.
25279 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25281 * tree-ssa-address.c: Include header file.
25282 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25284 (add_to_parts): Refactor.
25285 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25286 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25289 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25291 PR tree-optimization/53090
25292 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25294 (extract_cond_operands): Detect condition with IV on both sides
25295 and return COMP_IV_EXPR_2.
25296 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25297 (rewrite_use_compare): Simplify by removing call to function
25298 extract_cond_operands.
25300 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25302 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25303 (extract_cond_operands): Detect condition comparing against non-
25304 invariant bound and return appropriate enum value.
25305 (find_interesting_uses_cond): Update use of extract_cond_operands.
25306 Handle its return value accordingly.
25307 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25309 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25311 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25312 nonlinear iv_use computation in loop invariant sensitive way.
25314 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25316 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25317 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25319 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25321 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25322 (dump_cand): Support iv_cand.inv_exprs.
25323 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25325 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25328 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25330 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25332 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25333 as local function. Include necessary header files.
25334 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25336 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25338 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25340 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25342 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25343 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25344 RSHIFT_EXPR and BIT_NOT_EXPR.
25346 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25348 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25349 (adjust_setup_cost): New parameter supporting round up adjustment.
25350 (struct address_cost_data): Delete.
25351 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25352 (split_address_cost, ptr_difference_cost): Delete.
25353 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25354 (struct ainc_cost_data): New struct.
25355 (get_address_cost_ainc): New function.
25356 (get_address_cost, get_computation_cost): Reimplement.
25357 (determine_group_iv_cost_address): Record inv_expr for all uses of
25359 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25360 (iv_ca_has_deps): Reimplemented to ...
25361 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25363 (iv_ca_extend): Call iv_ca_more_deps.
25365 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25367 * tree-ssa-address.c (struct mem_address): Move to header file.
25368 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25369 * tree-ssa-address.h (struct mem_address): Move from C file.
25370 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25372 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25374 * tree-affine.h (aff_combination_type): New interface.
25375 (aff_combination_zero_p): Remove static.
25376 (aff_combination_const_p): New interface.
25377 (aff_combination_singleton_var_p): New interfaces.
25379 2017-05-11 Richard Biener <rguenther@suse.de>
25381 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25382 Skip unreachable blocks and destinations.
25383 (eliminate): Move stmt removal and fixup ...
25384 (fini_eliminate): ... here. Skip inserted exprs.
25385 (pass_pre::execute): Move fini_pre after fini_eliminate.
25386 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25387 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25388 PRE to get rid of dead code that has invalid SSA form and
25389 split critical edges again.
25391 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25393 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25395 2017-05-11 Richard Biener <rguenther@suse.de>
25397 * passes.c (execute_function_todo): Verify loops if they are
25398 said to be up-to-date.
25399 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25400 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25402 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25405 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25406 handle calling assemble_external ourself.
25409 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25410 modes with zero size. Enhance comment.
25412 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25414 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25415 built-ins for vec_xl and vec_xst with short and char pointer
25418 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25420 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25421 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25422 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25423 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25424 (_mm_maskz_min_round_ss): New intrinsics.
25425 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25426 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25427 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25428 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25429 (__builtin_ia32_minss_mask_round): New builtins.
25430 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25431 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25432 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25434 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25435 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25437 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25440 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25442 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25443 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25444 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25445 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25446 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25447 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25448 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25449 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25450 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25451 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25452 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25453 (__builtin_ia32_mulss_mask_round): New builtins.
25454 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25455 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25456 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25458 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25459 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25461 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25464 2017-05-10 Julia Koval <julia.koval@intel.com>
25466 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25467 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25468 (_mm256_setr_m128i): New intrinsics.
25470 2017-05-10 Julia Koval <julia.koval@intel.com>
25472 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25473 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25474 (_mm_maskz_rcp14_ss): New intrinsics.
25475 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25476 (__builtin_ia32_rcp14ss_mask): New builtins.
25477 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25479 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25481 PR tree-optimization/51513
25482 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25483 (assert_unreachable_fallthru_edge_p): Use it.
25484 (group_case_labels_stmt): Likewise.
25485 * tree-cfg.h: Prototype it.
25486 * stmt.c: Include cfghooks.h and tree-cfg.h.
25487 (emit_case_dispatch_table) <gap_label>: New local variable.
25488 Use it to fill dispatch table gaps.
25489 Test for default_label before updating probabilities.
25490 (expand_case) <default_label>: Remove unneeded initialization.
25491 Test for unreachable default case statement and remove its edge.
25492 Set default_label accordingly.
25493 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25495 2017-05-10 Carl Love <cel@us.ibm.com>
25497 * config/rs6000/rs6000-c: Add support for built-in functions
25498 vector signed char vec_neg (vector signed char)
25499 vector signed short int vec_neg (vector short int)
25500 vector signed int vec_neg (vector signed int)
25501 vector signed long long vec_neg (vector signed long long)
25502 vector float vec_neg (vector float)
25503 vector double vec_neg (vector double)
25504 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25506 * config/rs6000/altivec.h: Add define for vec_neg
25507 * doc/extend.texi: Update the built-in documentation for the
25508 new built-in functions.
25510 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25512 PR tree-optimization/77644
25513 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25515 2017-05-10 Nathan Sidwell <nathan@acm.org>
25517 * dumpfile.h (TDI_lang_all): New.
25518 (TDF_KIND): New. Renumber others
25519 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25521 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25523 (get_dump_file_name): Adjust suffix generation.
25524 (dump_enable_all): Use TDF_KIND.
25525 * doc/invoke.texi (-fdump-lang-all): Document.
25527 * dumpfile.h: Tabify.
25529 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25532 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25533 Move member access before delete.
25535 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25537 * tree-inline.c (expand_call_inline): Split block at stmt
25540 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25543 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25544 are now unused after splitting mov{sf,sd}_hardfloat.
25545 (f32_lr2): Likewise.
25546 (f32_lm): Likewise.
25547 (f32_lm2): Likewise.
25548 (f32_li): Likewise.
25549 (f32_li2): Likewise.
25550 (f32_lv): Likewise.
25551 (f32_sr): Likewise.
25552 (f32_sr2): Likewise.
25553 (f32_sm): Likewise.
25554 (f32_sm2): Likewise.
25555 (f32_si): Likewise.
25556 (f32_si2): Likewise.
25557 (f32_sv): Likewise.
25558 (f32_dm): Likewise.
25559 (f32_vsx): Likewise.
25560 (f32_av): Likewise.
25561 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25562 For movsf, order stores so the VSX stores occur before the GPR
25563 store which encourages the register allocator to use a traditional
25564 FPR instead of a GPR. For movsd, order the stores so that the GPR
25565 store comes before the VSX stores to allow the power6 to work.
25566 This is due to the power6 not having a 32-bit integer store
25567 instruction from a FPR.
25568 (movsf_hardfloat): Likewise.
25569 (movsd_hardfloat): Likewise.
25571 2017-05-09 Martin Sebor <msebor@redhat.com>
25573 PR translation/80280
25574 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25577 PR translation/80280
25578 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25579 data member added in r247778.
25580 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25582 2017-05-09 Nathan Sidwell <nathan@acm.org>
25584 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25586 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25589 2017-05-09 Marek Polacek <polacek@redhat.com>
25591 * doc/invoke.texi: Fix typo.
25593 2017-05-09 Richard Biener <rguenther@suse.de>
25595 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25596 (vrp_val_is_min): Likewise.
25597 (set_value_range_to_value): Likewise.
25598 (set_value_range_to_nonnegative): Likewise.
25599 (gimple_assign_nonzero_p): Likewise.
25600 (gimple_stmt_nonzero_p): Likewise.
25601 (vrp_int_const_binop): Likewise. Remove unreachable case.
25602 (adjust_range_with_scev): Adjust comments.
25603 (compare_range_with_value): Likewise.
25604 (extract_range_from_phi_node): Likewise.
25605 (test_for_singularity): Likewise.
25607 2017-05-09 Richard Biener <rguenther@suse.de>
25609 * tree-vrp.c (get_single_symbol): Add assert that we don't
25610 get overflowed constants as invariant part.
25611 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25612 checks. Use wi::cmp instead of recursing for integer constants.
25613 (compare_values): Just ignore whether we assumed undefined
25614 overflow instead of failing the compare.
25615 (extract_range_for_var_from_comparison_expr): Add comment before the
25616 TREE_NO_WARNING sets.
25617 (test_for_singularity): Likewise.
25618 (extract_range_from_comparison): Do not disable optimization
25619 when we assumed undefined overflow.
25620 (extract_range_basic): Remove init of unused var.
25622 2017-05-09 Richard Biener <rguenther@suse.de>
25624 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25625 (extract_range_from_multiplicative_op_1): Adjust.
25626 (extract_range_from_binary_expr_1): Use int_const_binop.
25628 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25631 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25632 rs6000_store_data_bypass_p in seven define_bypass directives and
25633 in several comments.
25634 * config/rs6000/rs6000-protos.h: Add prototype for
25635 rs6000_store_data_bypass_p function.
25636 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25637 function implements slightly different (rs6000-specific) semantics
25638 than store_data_bypass_p, returning false rather than aborting
25639 with assertion error when arguments do not satisfy the
25640 requirements of store data bypass.
25641 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25642 rs6000_store_data_bypass_p.
25644 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25646 * config/xtensa/xtensa-protos.h
25647 (xtensa_initial_elimination_offset): New declaration.
25648 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25649 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25650 macro definition, add case for FRAME_POINTER_REGNUM when
25651 FRAME_GROWS_DOWNWARD.
25652 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25653 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25654 xtensa_initial_elimination_offset.
25656 2017-05-08 Nathan Sidwell <nathan@acm.org>
25658 * doc/invoke.texi: Alphabetize -fdump options.
25660 2017-05-08 Martin Sebor <msebor@redhat.com>
25662 PR translation/80280
25663 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25665 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25667 * target.def (compute_frame_layout): New optional target hook.
25668 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25669 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25670 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25672 * reload1.c (verify_initial_elim_offsets): Likewise.
25673 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25674 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25675 (arm_get_frame_offsets): Split up into this ...
25676 (arm_compute_frame_layout): ... and this function.
25678 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25680 * config/aarch64/constraints.md (Usa): New constraint.
25681 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25683 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25685 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25686 with_multilib_list after it has been checked.
25688 2017-05-08 Richard Biener <rguenther@suse.de>
25690 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25691 (bitmap_set_subtract_values): Likewise.
25693 2017-05-08 Richard Biener <rguenther@suse.de>
25695 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25696 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25698 (gimple_stmt_nonzero_warnv_p): Rename to ...
25699 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25701 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25702 (extract_range_basic): Adjust, do not disable propagation on
25703 strict overflow sensitive simplification.
25704 (vrp_visit_cond_stmt): Likewise.
25706 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25708 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25709 body size unconditionally.
25711 2017-05-07 Jeff Law <law@redhat.com>
25714 2017-05-06 Jeff Law <law@redhat.com>
25715 PR tree-optimization/78496
25716 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25719 PR tree-optimization/78496
25720 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25721 (simplify_stmt_using_ranges): Call it.
25722 (vrp_dom_walker::before_dom_children): Extract equivalences
25723 from an ASSERT_EXPR with an equality comparison against a
25726 2017-05-06 Jeff Law <law@redhat.com>
25728 PR tree-optimization/78496
25729 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25732 PR tree-optimization/78496
25733 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25734 (simplify_stmt_using_ranges): Call it.
25735 (vrp_dom_walker::before_dom_children): Extract equivalences
25736 from an ASSERT_EXPR with an equality comparison against a
25739 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25741 * lra-constraints.c (lra_copy_reg_equiv): New function.
25742 (split_reg): Use it to copy equivalence information from the
25743 original register to the spill register.
25745 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25747 PR rtl-optimization/75964
25748 * simplify-rtx.c (simplify_const_relational_operation): Remove
25749 invalid handling of comparisons of integer ABS.
25751 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25753 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25754 initialize to zero.
25755 (init_regs): Remove declaration.
25756 (function_arg_advance_32): Initialize error_p as boolean variable.
25758 2017-05-05 Nathan Sidwell <nathan@acm.org>
25760 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25761 lines. Use for (;;).
25763 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25765 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25766 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25767 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25768 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25769 VF=2 that require versioning.
25771 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25773 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25776 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25778 * diagnostic.h (diagnostic_override_option_index): Convert from
25779 macro to inline function.
25781 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25783 * diagnostic.c (last_module_changed_p): New function.
25784 (set_last_module): New function.
25785 (diagnostic_report_current_module): Convert macro usage to
25786 the above functions.
25787 * diagnostic.h (diagnostic_context::last_module): Strengthen
25788 from const line_map * to const line_map_ordinary *.
25789 (diagnostic_last_module_changed): Delete macro.
25790 (diagnostic_set_last_module): Delete macro.
25792 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25794 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25795 with diagnostic_report_diagnostic.
25796 (diagnostic_n_impl_richloc): Likewise.
25797 * diagnostic.h (report_diagnostic): Delete macro.
25798 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25799 with diagnostic_report_diagnostic.
25800 * substring-locations.c (format_warning_va): Likewise.
25802 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25804 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25805 save/restor of format_spec. Move option-printing code to...
25806 (print_option_information): ...this new function, and
25807 reimplement by simply printing to the pretty_printer,
25808 rather than appending to the format string.
25810 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25812 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25813 handling logic into...
25814 (update_effective_level_from_pragmas): ...this new function.
25816 2017-05-04 Andrew Waterman <andrew@sifive.com>
25818 * config/riscv/riscv.opt (mstrict-align): New option.
25819 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25820 (SLOW_UNALIGNED_ACCESS): Define.
25821 (riscv_slow_unaligned_access): Declare.
25822 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25824 (riscv_slow_unaligned_access): New variable.
25825 (rocket_tune_info): Set slow_unaligned_access to true.
25826 (optimize_size_tune_info): Set slow_unaligned_access to false.
25827 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25828 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25829 (riscv_option_override): Set riscv_slow_unaligned_access.
25830 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25832 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25834 * config/riscv/riscv.md: Unify indentation.
25836 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25841 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25843 (extendsi<mode>2): Add support for doing sign extension via
25844 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25845 don't have ISA 3.0 instructions.
25846 (extendsi<mode>2 splitter): Likewise.
25847 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25848 generate the normal insns since SImode can now go in vector
25849 registers. Disallow the special UNSPECs needed for previous
25850 machines to hide SImode being used. Add new insns
25851 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25852 (fix_trunc<mode>si2_stfiwx): Likewise.
25853 (fix_trunc<mode>si2_internal): Likewise.
25854 (fixuns_trunc<mode>si2): Likewise.
25855 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25856 (fctiw<u>z_<mode>_smallint): Likewise.
25857 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25858 of floating point to 32-bit integer from doing a direct move to
25859 the GPR registers to do a store.
25860 (fctiwz_<mode>): Break long line.
25862 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25864 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25865 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25866 (addr_list, addr_offset_valid_p): New.
25867 (split_address_groups): Check offset validity with above function.
25868 (gt-tree-ssa-loop-ivopts.h): Include header file.
25870 2017-05-05 Nathan Sidwell <nathan@acm.org>
25872 * config.gcc (arm*-*-*): Add missing 'fi'.
25874 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25876 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25877 included in -fopt-info does not matter.
25878 * doc/optinfo.texi (-fopt-info): Fix description of default
25879 behavour. Explicitly say order of options included in -fopt-info
25882 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25884 * config.gcc: Allow combinations of aprofile and rmprofile values for
25885 --with-multilib-list.
25886 * config/arm/t-multilib: New file.
25887 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25888 variables. Remove setting of ISA and floating-point ABI in
25889 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25890 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25891 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25892 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25894 * config/arm/t-rmprofile: Likewise except for the matches changes.
25895 * doc/install.texi (--with-multilib-list): Document the combination of
25896 aprofile and rmprofile values and warn about pitfalls in doing that.
25898 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25900 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25901 (movdi_aarch64): Likewise.
25903 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25905 PR tree-optimization/80632
25906 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25908 (build_arrays): Initialize it for virtual phis.
25909 (fix_phi_nodes): Use it for virtual phis.
25911 PR tree-optimization/80558
25912 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25913 [x, y] op z into [x op, y op z] for op & or | if conditions
25916 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25917 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25920 * config/arm/arm.md (use_literal_pool): Remove.
25921 (64-bit immediate split): No longer takes cost into consideration
25922 if arm_disable_literal_pool is enabled.
25923 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25924 used when arm_disable_literal_pool is enabled.
25925 (arm_max_const_double_inline_cost): Remove use of
25926 arm_disable_literal_pool.
25927 (push_minipool_fix): Add assert.
25928 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25929 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25930 (no_literal_pool_sf_immediate): New.
25932 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25934 PR tree-optimization/80613
25935 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25936 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25938 2017-05-05 Richard Biener <rguenther@suse.de>
25940 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25942 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25944 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25945 of this flag from insn conditions due to removal from r247495.
25947 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25949 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25951 (arm_early_store_addr_dep_ptr): Likewise.
25952 * config/arm/aarch-common-protos.h
25953 (arm_early_load_addr_dep_ptr): Add prototype.
25954 (arm_early_store_addr_dep_ptr): Likewise.
25955 * config/arm/cortex-a53.md: Add new bypasses.
25957 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25959 * tree.c (next_type_uid): Change type to unsigned.
25960 (type_hash_canon): Decrement back next_type_uid if
25961 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25962 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25965 2017-05-04 Martin Sebor <msebor@redhat.com>
25967 * builtins.c: Fix a trivial typo in a comment.
25969 PR middle-end/79234
25970 * builtins.c (check_sizes): Adjust to handle reading past the end.
25971 Avoid printing excessive upper bound of ranges. Use %E to print
25972 tree nodes instead of converting them to %wu.
25973 (expand_builtin_memchr): New function.
25974 (compute_dest_size): Rename...
25975 (compute_objsize): ...to this.
25976 (expand_builtin_memcpy): Adjust.
25977 (expand_builtin_mempcpy): Adjust.
25978 (expand_builtin_strcat): Adjust.
25979 (expand_builtin_strcpy): Adjust.
25980 (check_strncat_sizes): Adjust.
25981 (expand_builtin_strncat): Adjust.
25982 (expand_builtin_strncpy): Adjust and simplify.
25983 (expand_builtin_memset): Adjust.
25984 (expand_builtin_bzero): Adjust.
25985 (expand_builtin_memcmp): Adjust.
25986 (expand_builtin): Handle memcmp.
25987 (maybe_emit_chk_warning): Check strncat just once.
25989 2017-05-04 Martin Sebor <msebor@redhat.com>
25991 PR preprocessor/79214
25992 PR middle-end/79222
25993 PR middle-end/79223
25994 * builtins.c (check_sizes): Add inlining context and issue
25995 warnings even when -Wno-system-headers is set.
25996 (check_strncat_sizes): Same.
25997 (expand_builtin_strncat): Same.
25998 (expand_builtin_memmove): New function.
25999 (expand_builtin_stpncpy): Same.
26000 (expand_builtin): Handle memmove and stpncpy.
26002 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26004 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26005 which is not used any more.
26007 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26009 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26011 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26013 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26014 (cortexa53_tunings): Likewise.
26015 (cortexa57_tunings): Likewise.
26016 (cortexa72_tunings): Likewise.
26017 (cortexa73_tunings): Likewise.
26019 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26021 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26022 Set loop alignment to 8.
26024 2017-05-04 Martin Sebor <msebor@redhat.com>
26026 PR translation/80280
26027 * builtins.c (expand_builtin_object_size): Add missing quoting to
26028 %D and like directives.
26029 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26030 (hsa_type_for_tree_type): Same.
26031 (verify_function_arguments): Same.
26032 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26033 * varasm.c (get_section): Same.
26036 2017-05-04 Martin Sebor <msebor@redhat.com>
26038 PR translation/80280
26039 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26041 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26043 * config/aarch64/aarch64.c (generic_addrcost_table):
26044 Change HI/TI mode setting.
26046 2017-05-04 Martin Jambor <mjambor@suse.cz>
26048 PR tree-optimization/80622
26049 * tree-sra.c (comes_initialized_p): New function.
26050 (build_accesses_from_assign): Only set write lazily when
26051 comes_initialized_p is false.
26052 (analyze_access_subtree): Use comes_initialized_p.
26053 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26054 instead of testing for PARM_DECL.
26056 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26058 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26059 constraint on operand 0 to allow more general addressing modes.
26060 Adjust output template.
26061 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26063 * config/aarch64/aarch64-protos.h
26064 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26065 * config/aarch64/constraints.md (Dp): New address constraint.
26066 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26069 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26071 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26072 update use of estimate_ipcp_clone_size_and_time.
26073 (estimate_local_effects): Update use of
26074 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26075 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26076 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26077 Return nonspecialized time.
26079 2017-05-04 Richard Biener <rguenther@suse.de>
26081 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26082 for the last VUSE which def dominates the PHI. Directly call
26084 (get_continuation_for_phi_1): Remove.
26086 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26088 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26089 to explain the use of truncating division. Cap the number of
26090 iterations to the maximum given by nb_iterations_upper_bound,
26093 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26095 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26096 * configure: Regenerate.
26097 * config.in: Regenerate.
26098 * config/i386/driver-mingw32.c: new file.
26099 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26100 * config.host: Link driver-mingw32.o on MinGW host.
26101 * doc/install.texi: Document new --enable-mingw-wildcard configure
26104 2017-05-04 Marek Polacek <polacek@redhat.com>
26106 PR tree-optimization/80612
26107 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26109 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26110 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26112 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26113 (movt splitter): Likewise.
26114 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26115 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26116 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26117 block for Thumb-1 with MOVT.
26118 (thumb2_legitimate_address_p): Move code block ...
26119 (can_avoid_literal_pool_for_label_p): ... into this new function.
26120 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26122 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26123 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26124 "M-profile targets with the MOVT instruction".
26126 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26128 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26129 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26130 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26132 2017-05-04 Martin Liska <mliska@suse.cz>
26134 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26135 variable cond_code.
26137 2017-05-04 Richard Biener <rguenther@suse.de>
26139 * tree.c (array_at_struct_end_p): Handle arrays at struct
26140 end with flexarrays more conservatively. Refactor and treat
26141 arrays of arrays or aggregates more strict. Fix
26142 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26143 * tree.c (array_at_struct_end_p): Adjust prototype.
26144 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26145 * gimple-fold.c (get_range_strlen): Likewise.
26146 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26148 2017-05-04 Richard Biener <rguenther@suse.de>
26150 PR tree-optimization/31130
26151 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26153 (supports_overflow_infinity): Likewise.
26154 (is_negative_overflow_infinity): Likewise.
26155 (is_positive_overflow_infinity): Likewise.
26156 (is_overflow_infinity): Likewise.
26157 (stmt_overflow_infinity): Likewise.
26158 (overflow_infinity_range_p): Likewise.
26159 (usable_range_p): Remove as always returning true.
26160 (make_overflow_infinity): Remove.
26161 (negative_overflow_infinity): Likewise.
26162 (positive_overflow_infinity): Likewise.
26163 (avoid_overflow_infinity): Likewise.
26164 (set_value_range): Adjust accordingly.
26165 (set_value_range_to_nonnegative): Likewise, remove now unused
26166 overflow_infinity arg.
26167 (vrp_operand_equal_p): Adjust.
26168 (update_value_range): Likewise.
26169 (range_int_cst_singleton_p): Likewise.
26170 (operand_less_p): Likewise.
26171 (compare_values_warnv): Likewise.
26172 (extract_range_for_var_from_comparison_expr): Likewise.
26173 (vrp_int_const_binop): Likewise.
26174 (zero_nonzero_bits_from_vr): Likewise.
26175 (extract_range_from_multiplicative_op_1): Likewise.
26176 (extract_range_from_binary_expr_1): Likewise.
26177 (extract_range_from_unary_expr): Likewise.
26178 (extract_range_from_comparison): Likewise.
26179 (extract_range_basic): Likewise.
26180 (adjust_range_with_scev): Likewise.
26181 (compare_ranges): Likewise.
26182 (compare_range_with_value): Likewise.
26183 (dump_value_range): Likewise.
26184 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26186 (simplify_cond_using_ranges): Adjust.
26188 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26190 * brig-builtins.def: Added a builtin for class_f64.
26191 * builtin-types.def: Added a builtin type needed by class_f64.
26193 2017-05-03 Jason Merrill <jason@redhat.com>
26195 * timevar.def: Add TV_CONSTEXPR.
26197 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26199 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26201 2017-05-03 Martin Jambor <mjambor@suse.cz>
26203 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26204 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26205 * ipa-cp.c (ipcp_propagate_stage): Do not call
26206 ipa_update_after_lto_read.
26207 * ipa-inline.c (ipa_inline): Likewise.
26209 2017-05-03 Martin Jambor <mjambor@suse.cz>
26211 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26212 tag. Added a default constructor and a destructor.
26213 (ipa_edge_args_sum_t): New class;
26214 (ipa_edge_args_sum): Declare.
26215 (ipa_edge_args_vector): Remove declaration.
26216 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26217 (ipa_free_edge_args_substructures): Remove declaration.
26218 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26219 (ipa_edge_args_info_available_for_edge_p): Likewise.
26220 * ipa-prop.c (ipa_edge_args_vector): Removed.
26221 (edge_removal_hook_holder): Likewise.
26222 (edge_duplication_hook_holder): Likewise.
26223 (ipa_edge_args_sum): New variable.
26224 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26225 ipa_edge_args_vector.
26226 (ipa_free_edge_args_substructures): Likewise.
26227 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26228 ipa_edge_args_vector.
26229 (ipa_edge_removal_hook): Turned into method
26230 ipa_edge_args_sum_t::remove.
26231 (ipa_edge_duplication_hook): Turned into method
26232 ipa_edge_args_sum_t::duplicate.
26233 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26234 registering edge hooks.
26235 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26236 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26237 ipa_edge_args_sum instead of ipa_edge_args_vector.
26238 * ipa-profile.c (ipa_profile): Likewise.
26240 2017-05-03 Martin Jambor <mjambor@suse.cz>
26242 * symbol-summary.h (function_summary): New method exists.
26243 (function_summary::symtab_removal): Deallocate through release.
26244 (call_summary): New class.
26245 (gt_ggc_mx): New overload.
26246 (gt_pch_nx): Likewise.
26247 (gt_pch_nx): Likewise.
26249 2017-05-03 Jeff Law <law@redhat.com>
26251 PR tree-optimization/78496
26252 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26253 from simplify_cond_using_ranges. Split off code to walk
26254 backwards through casts into ...
26255 (simplify_cond_using_ranges_2): New function.
26256 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26257 (execute_vrp): After identifying jump threads, call
26258 simplify_cond_using_ranges_2.
26260 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26263 * ipa-inline.h (inline_summary): Add ctor.
26264 (create_ggc): Do not use ggc_cleared_alloc.
26266 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26267 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26269 * gcc.c (handle_braces): Support escaping in switch matching
26271 * doc/invoke.texi (Spec Files): Document it.
26272 Remove superfluous @code markup in items.
26274 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26276 * diagnostic-show-locus.c (struct column_range): New struct.
26277 (get_affected_columns): New function.
26278 (get_printed_columns): New function.
26279 (struct correction): New struct.
26280 (correction::ensure_capacity): New function.
26281 (correction::ensure_terminated): New function.
26282 (struct line_corrections): New struct.
26283 (line_corrections::~line_corrections): New dtor.
26284 (line_corrections::add_hint): New function.
26285 (layout::print_trailing_fixits): Reimplement in terms of the new
26287 (selftest::test_overlapped_fixit_printing): New function.
26288 (selftest::diagnostic_show_locus_c_tests): Call it.
26290 2017-05-03 Nathan Sidwell <nathan@acm.org>
26292 Canonicalize canonical type hashing
26293 * tree.h (type_hash_canon_hash): Declare.
26294 * tree.c (type_hash_list, attribute_hash_list): Move into
26295 type_hash_canon_hash.
26296 (build_type_attribute_qual_variant): Break out hash code calc into
26297 type_hash_canon_hash.
26298 (type_hash_canon_hash): New. Generic type hash computation.
26299 (build_range_type_1, build_array_type_1, build_function_type,
26300 build_method_type_directly, build_offset_type, build_complex_type,
26301 make_vector_type): Call it.
26303 2017-05-03 Richard Biener <rguenther@suse.de>
26305 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26306 When all DRs have unknown misaligned do not always peel
26307 when there is a store but apply the same costing model as if
26308 there were only loads.
26310 2017-05-03 Richard Biener <rguenther@suse.de>
26313 PR tree-optimization/80492
26314 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26315 compare_base_decls returning dont-know properly.
26317 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26319 * config/arm/iterators.md (CCSI): New mode iterator.
26320 (arch): New mode attribute.
26321 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26322 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26323 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26324 code iterator for success result mode.
26325 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26326 the corresponding new insn generators.
26328 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26331 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26332 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26334 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26336 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26337 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26338 (DDR_A): Wrap DDR argument in brackets.
26339 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26340 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26341 (DDR_REVERSED_P): Likewise.
26343 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26345 PR tree-optimization/79472
26346 * tree-switch-conversion.c (struct switch_conv_info): Add
26347 contiguous_range and default_case_nonstandard fields.
26348 (collect_switch_conv_info): Compute contiguous_range and
26349 default_case_nonstandard fields, don't clear final_bb if
26350 contiguous_range and only the default case doesn't have the required
26352 (check_all_empty_except_final): Set default_case_nonstandard instead
26353 of failing if contiguous_range and the default case doesn't have empty
26355 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26356 and only the default case doesn't have the required constants. Skip
26358 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26359 if default_case_nonstandard.
26360 (build_constructors): Build constant 1 just once. Assert that default
26361 values aren't inserted in between cases if contiguous_range. Skip
26363 (build_arrays): Skip virtual phis.
26364 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26365 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26366 Handle virtual phis.
26367 (gen_inbound_check): Handle default_case_nonstandard case.
26368 (process_switch): Adjust check_final_bb caller. Call
26369 gather_default_values with the first non-default case instead of
26370 default case if default_case_nonstandard.
26372 2017-05-02 Nathan Sidwell <nathan@acm.org>
26374 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26375 check. Fix formatting.
26377 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26379 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26380 errors when comparing specialized and unspecialized times.
26382 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26384 * diagnostic-show-locus.c
26385 (layout::should_print_annotation_line_p): Make private.
26386 (layout::print_annotation_line): Make private.
26387 (layout::annotation_line_showed_range_p): Make private.
26388 (layout::show_ruler): Make private.
26389 (layout::print_source_line): Make private. Pass in line and
26390 line_width, rather than calling location_get_source_line. Drop
26392 (layout::print_leading_fixits): New method.
26393 (layout::print_any_fixits): Rename to...
26394 (layout::print_trailing_fixits): ...this, and make private.
26395 Don't print newline fixits.
26396 (diagnostic_show_locus): Move logic for printing one row into...
26397 (layout::print_line): ...this new function. Move the
26398 location_get_source_line call and error-handling from
26399 print_source_line to here. Call print_leading_fixits, and rename
26400 print_any_fixits to print_trailing_fixits.
26401 (selftest::test_fixit_insert_containing_newline): Update now that
26402 newlines are partially supported.
26403 (selftest::test_fixit_insert_containing_newline_2): New test.
26404 (selftest::test_fixit_replace_containing_newline): Update comments.
26405 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26406 * edit-context.c (class added_line): New class.
26407 (class edited_line): Describe newline handling in comment.
26408 (edited_line::actually_edited_p): New method.
26409 (edited_line::print_content): Delete redundant decl.
26410 (edited_line::m_predecessors): New field.
26411 (edited_file::print_content): Call edited_line::print_content.
26412 (edited_file::print_diff): Update to support newlines.
26413 (edited_file::print_diff_hunk): Likewise.
26414 (edited_file::print_run_of_changed_lines): New function.
26415 (edited_file::print_diff_line): Convert to...
26416 (print_diff_line): ...this.
26417 (edited_file::get_effective_line_count): New function.
26418 (edited_line::edited_line): Initialize new field m_predecessors.
26419 (edited_line::~edited_line): Clean up m_predecessors.
26420 (edited_line::apply_fixit): Handle newlines.
26421 (edited_line::get_effective_line_count): New function.
26422 (edited_line::print_content): New function.
26423 (edited_line::print_diff_lines): New function.
26424 (selftest::test_applying_fixits_insert_containing_newline): New
26426 (selftest::test_applying_fixits_replace_containing_newline): New
26428 (selftest::insert_line): New function.
26429 (selftest::test_applying_fixits_multiple_lines): Add example of
26431 (selftest::edit_context_c_tests): Call the new tests.
26433 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26435 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26436 parameter cand. Update dump information.
26437 (get_computation_cost): Update uses.
26439 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26441 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26442 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26443 (get_computation_at, rewrite_use_address): Update use of
26444 get_computation_aff.
26446 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26448 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26449 (get_computation): Delete.
26450 (get_computation_cost): Implement like get_computation_cost_at.
26451 Use get_computation_at.
26452 (get_computation_cost_at): Delete.
26453 (rewrite_use_nonlinear_expr): Use get_computation_at.
26454 (rewrite_use_compare, remove_unused_ivs): Ditto.
26456 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26458 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26460 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26462 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26463 (ivopts_global_cost_for_size): Rename parameter and update uses.
26464 (iv_ca_recount_cost): Update uses.
26465 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26466 candidates seperately in n_invs and n_cands.
26467 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26469 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26471 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26472 (find_inv_vars_cb): New.
26473 (find_depends): Renamed to ...
26474 (find_inv_vars): ... this.
26475 (add_candidate_1, force_var_cost): Call find_inv_vars.
26476 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26478 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26480 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26481 inv_vars. Add inv_exprs.
26482 (struct iv_cand): Rename depends_on to inv_vars.
26483 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26484 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26485 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26486 (dump_cand): Dump inv_vars.
26487 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26488 (record_invariant, find_depends, add_candidate_1): Ditto.
26489 (set_group_iv_cost, force_var_cost): Ditto.
26490 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26491 (get_computation_cost_at, get_computation_cost): Ditto.
26492 (determine_group_iv_cost_generic): Ditto.
26493 (determine_group_iv_cost_address): Ditto.
26494 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26495 (determine_group_iv_costs): Ditto.
26496 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26497 (iv_ca_set_remove_invariants): Renamed to ...
26498 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26499 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26500 (iv_ca_set_add_invariants): Renamed to ...
26501 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26502 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26503 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26504 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26505 (create_new_ivs): Remove useless dump.
26507 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26509 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26511 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26512 (iv_ca_set_no_cp, create_new_iv): Ditto.
26514 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26516 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26518 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26520 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26521 function tree_check2.
26523 2017-05-02 Martin Liska <mliska@suse.cz>
26525 * doc/gcov.texi: Add missing preposition.
26526 * gcov.c (function_info::function_info): Properly fill up
26527 all member variables.
26529 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26531 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26533 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26535 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26537 2017-05-02 Martin Liska <mliska@suse.cz>
26540 * lto-streamer-in.c (lto_read_tree_1): Remove
26541 LTO_STREAMER_DEBUG.
26542 * lto-streamer.c (struct tree_hash_entry): Likewise.
26543 (struct tree_entry_hasher): Likewise.
26544 (tree_entry_hasher::hash): Likewise.
26545 (tree_entry_hasher::equal): Likewise.
26546 (lto_streamer_init): Likewise.
26547 (lto_orig_address_map): Likewise.
26548 (lto_orig_address_get): Likewise.
26549 (lto_orig_address_remove): Likewise.
26550 * lto-streamer.h: Likewise.
26551 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26552 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26554 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26556 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26557 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26558 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26559 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26560 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26561 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26562 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26563 (mm_maskz_sub_ss): New intrinsics.
26564 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26565 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26566 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26567 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26568 (__builtin_ia32_subss_mask_round): New builtins.
26569 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26570 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26571 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26573 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26574 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26576 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26579 2017-05-02 Martin Jambor <mjambor@suse.cz>
26581 PR tree-optimization/78687
26582 * tree-sra.c (access): New field parent.
26583 (process_subtree_disqualification): New function.
26584 (disqualify_candidate): Call it.
26585 (build_accesses_from_assign): Reset write flag if creating an
26587 (build_access_subtree): Fill in parent field and also prpagate
26588 down grp_write flag.
26589 (create_artificial_child_access): New parameter set_grp_write, set
26590 grp_write to its value.
26591 (propagate_subaccesses_across_link): Also propagate grp_write flag
26593 (propagate_all_subaccesses): Push the closest parent back to work
26594 queue if add_access_to_work_queue returned true.
26596 2017-05-02 Richard Biener <rguenther@suse.de>
26598 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26599 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26600 -fstrict-overflow documentation.
26601 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26602 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26603 flag_strict_overflow.
26604 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26605 * lto-opts.c (lto_write_options): Do not stream it.
26606 * lto-wrapper.c (merge_and_complain): Do not handle it.
26607 * opts.c (default_options_table): Do not set -fstrict-overflow.
26608 (finish_options): Likewise do not clear it when sanitizing.
26609 * simplify-rtx.c (simplify_const_relational_operation): Do not
26610 test flag_strict_overflow.
26612 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26614 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26615 using enabled attribute.
26616 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26617 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26618 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26619 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26620 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26621 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26622 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26623 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26624 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26625 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26627 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26629 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26631 2017-05-02 Richard Biener <rguenther@suse.de>
26633 PR tree-optimization/80591
26635 2017-04-10 Richard Biener <rguenther@suse.de>
26637 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26640 2017-05-02 Richard Biener <rguenther@suse.de>
26642 PR tree-optimization/80549
26643 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26644 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26645 headers if they do not have a preheader.
26647 2017-05-02 Martin Liska <mliska@suse.cz>
26650 * common.opt: Fix typo.
26651 * doc/invoke.texi: Likewise.
26653 2017-05-01 Jan Beulich <jbeulich@suse.com>
26655 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26656 swapping, add (x,x,m,x,n) alternative.
26658 2017-05-01 Nathan Sidwell <nathan@acm.org>
26660 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26661 unnecessary unadjusted_alignment check.
26663 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26666 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26668 * gimplify.c (gimplify_cilk_detach): New function.
26669 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26670 * tree-core.h: Document EXPR_CILK_SPAWN.
26671 * tree.h (EXPR_CILK_SPAWN): Define.
26673 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26675 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26676 to use new fixit_hint representation, using the "replace" logic.
26677 (get_line_span_for_fixit_hint): Likewise.
26678 (layout::print_any_fixits): Likewise.
26679 (selftest::test_one_liner_many_fixits): Rename to...
26680 (selftest::test_one_liner_many_fixits_1): ...this, and update
26681 comment and expected output to reflect that the multiple fix-it
26682 hints are now consolidated into one insertion.
26683 (selftest::test_one_liner_many_fixits_2): New test.
26684 (selftest::test_diagnostic_show_locus_one_liner): Update for
26686 (selftest::test_fixit_consolidation): Update for fix-it API
26688 * diagnostic.c (print_parseable_fixits): Likewise.
26689 * edit-context.c (edited_line::m_line_events): Convert from
26690 auto_vec <line_event *> to auto_vec <line_event>.
26691 (class line_event): Convert from abstract base class to a concrete
26692 class, taking over the role of replace_event.
26693 (class insert_event): Delete.
26694 (class replace_event): Rename to class line_event. Convert to
26696 (edit_context::add_fixits): Reimplement.
26697 (edit_context::apply_insert): Delete.
26698 (edit_context::apply_replace): Rename to...
26699 (edit_context::apply_fixit): ...this. Convert to half-open range.
26700 (edited_file::apply_insert): Delete.
26701 (edited_file::apply_replace): Rename to...
26702 (edited_file::apply_fixit): ...this.
26703 (edited_line::~edited_line): Drop deletion of events.
26704 (edited_line::apply_insert): Delete.
26705 (edited_line::apply_replace): Rename to...
26706 (edited_line::apply_fixit): ...this. Convert to half-open range.
26707 Update for change to type of m_line_events.
26708 * edit-context.h (edit_context::apply_insert): Delete.
26709 (edit_context::apply_replace): Rename to...
26710 (edit_context::apply_fixit): ...this.
26712 2017-05-01 Martin Sebor <msebor@redhat.com>
26714 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26717 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26720 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26721 __get_cpuid_max returns 0.
26722 (__get_cpuid_count): Ditto.
26724 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26726 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26727 replacement expression is another instance of one of its arguments.
26729 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26732 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26733 check for stack push/pop autoinc.
26734 * config/i386/i386.c (ix86_agi_dependent): Return false
26735 if the only reason why modified_in_p returned true is that
26736 addr is SP based and set_insn is a push or pop.
26738 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26740 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26743 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26746 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26747 (account_size_time): Use two predicates - exec_pred and
26749 (evaluate_conditions_for_known_args): Compute both clause and
26751 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26752 (inline_summary_t::duplicate): Update.
26753 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26755 (compute_inline_parameters): Likewise.
26756 (estimate_edge_size_and_time): Update caluclation of time.
26757 (estimate_node_size_and_time): Compute both time and nonspecialized
26759 (estimate_ipcp_clone_size_and_time): Update.
26760 (inline_merge_summary): Update.
26761 (do_estimate_edge_time): Update.
26762 (do_estimate_edge_size): Update.
26763 (do_estimate_edge_hints): Update.
26764 (inline_read_section, inline_write_summary): Stream both new predicates.
26765 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26767 (compute_inlined_call_time): Cleanup.
26768 (big_speedup_p): Update.
26769 (edge_badness): Update.
26770 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26771 (size_time_entry): Replace predicate by exec_predicate and
26772 nonconst_predicate.
26773 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26774 (estimate_edge_time): Return also nonspec_time.
26775 (reset_edge_growth_cache): Update.
26777 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26779 PR rtl-optimization/80491
26780 * ifcvt.c (noce_process_if_block): When looking for x setter
26781 with missing else_bb, don't check only the insn right before
26782 cond_earliest, but look for the last insn that x is modified in
26783 within the same bb.
26785 PR rtl-optimization/80491
26786 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26788 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26790 PR tree-optimization/80487
26791 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26793 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26795 PR tree-optimization/79697
26796 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26797 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26798 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26800 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26801 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26803 2017-04-28 Martin Sebor <msebor@redhat.com>
26805 PR tree-optimization/80523
26806 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26807 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26809 (maybe_warn, format_directive, parse_directive): Use new functions.
26810 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26812 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26814 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26816 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26818 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26819 target_header_dir): Set correctly.
26820 * configure: Regenerated.
26821 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26822 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26823 instead of SYSTEM_HEADER_DIR.
26825 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26827 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26828 (estimate_local_effects): Likewise.
26829 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26830 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26831 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26832 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26833 do_estimate_edge_time, estimate_edge_time): Likewise.
26834 * ipa-inline-analysis.c (estimate_node_size_and_time,
26835 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26836 (estimate_time_after_inlining): Remove.
26838 2017-04-28 Martin Liska <mliska@suse.cz>
26840 * doc/gcov.texi: Enhance documentation of gcov.
26842 2017-04-28 Martin Liska <mliska@suse.cz>
26844 * doc/gcov.texi: Sort options in alphabetic order.
26845 * doc/gcov-dump.texi: Likewise.
26846 * doc/gcov-tool.texi: Likewise.
26847 * gcov.c (print_usage): Likewise.
26848 * gcov-dump.c (print_usage): Likewise.
26849 * gcov-tool.c (print_merge_usage_message): Likewise.
26850 (print_rewrite_usage_message): Likewise.
26851 (print_overlap_usage_message): Likewise.
26853 2017-04-28 Martin Liska <mliska@suse.cz>
26855 PR gcov-profile/53915
26856 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26858 2017-04-28 Martin Liska <mliska@suse.cz>
26860 PR gcov-profile/79891
26861 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26862 is marked by compiler as living on a line.
26863 (get_cycles_count): Remove usage of the union.
26864 (output_intermediate_file): Likewise.
26865 (find_source): Fix GNU coding style.
26866 (accumulate_line_counts): Remove old non-all block mode.
26867 (output_lines): Remove usage of the union.
26868 * profile.c (output_location): Include all BBs, even if
26869 belonging to a same line (and file) as a previous BB.
26871 2017-04-28 Martin Liska <mliska@suse.cz>
26873 * gcov.c (process_args): Handle new argument 'w'.
26874 (read_graph_file): Assign ID to BBs.
26875 (output_branch_count): Display BB # if verbose flag is set.
26876 (output_lines): Likewise for arcs.
26877 (print_usage): Add '--verbose' option help.
26878 * doc/gcov.texi: Document --verbose (-w) option.
26880 2017-04-28 Martin Liska <mliska@suse.cz>
26882 * gcov.c (struct block_location_info): New struct.
26883 (process_file): Fill up the new structure.
26884 (read_graph_file): Replace usage of encoding by the newly added
26886 (add_line_counts): Likewise.
26887 (accumulate_line_counts): Remove usage of the union.
26888 (function_info::function_info): New function.
26889 (function_info::~function_info): Likewise.
26890 (process_file): Call delete instead of release_function.
26891 (release_function): Release the function.
26892 (release_structures): Call delete instead of release_function.
26893 (solve_flow_graph): Replace usage of num_blocks.
26894 (find_exception_blocks): Likewise.
26895 (output_lines): Fix GNU coding style.
26897 2017-04-28 Martin Liska <mliska@suse.cz>
26900 * coverage.c (coverage_remove_note_file): New function.
26901 * coverage.h: Declare the function.
26902 * toplev.c (finalize): Clean if an error has been seen.
26904 2017-04-28 Martin Liska <mliska@suse.cz>
26906 PR gcov-profile/80031
26907 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26908 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26909 * gcov.c (read_graph_file): Read just number of blocks.
26910 * profile.c (branch_prob): Do not stream 0 flags per a basic
26913 2017-04-28 Martin Liska <mliska@suse.cz>
26915 * gcov-dump.c (tag_*): Add new argument to declarations.
26916 (dump_gcov_file): Likewise.
26917 (tag_blocks): Add and use new argument depth.
26918 (tag_arcs): Likewise.
26919 (tag_lines): Likewise.
26920 (tag_counters): Likewise.
26921 (tag_summary): Likewise.
26922 (dump_working_sets): Use depth to do a proper indentation.
26924 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26927 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26928 * symtab.c (symtab_node::debug_symtab): Move definition here.
26930 2017-04-28 Richard Biener <rguenther@suse.de>
26932 * lto-streamer.h (LTO_major_version): Bump to 7.
26934 2017-04-28 Richard Biener <rguenther@suse.de>
26936 * tree-vrp.c (assert_info): New struct.
26937 (add_assert_info): New helper.
26938 (register_edge_assert_for_2): Refactor to add asserts to a vector
26940 (register_edge_assert_for_1): Likewise.
26941 (register_edge_assert_for): Likewise.
26942 (finish_register_edge_assert_for): New helper actually registering
26943 asserts where live on edge.
26944 (find_conditional_asserts): Adjust.
26945 (find_switch_asserts): Likewise.
26946 (evrp_dom_walker::try_find_new_range): Generalize.
26947 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26949 2017-04-27 Marek Polacek <polacek@redhat.com>
26952 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26953 arg10 and arg11 to itype.
26955 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26957 * doc/extend.texi (Object Size Checking): Improve grammar.
26959 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26962 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26963 that the logic for permitting reciprocal estimates matches that
26966 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26969 * tree.c (type_cache_hasher::equal): Only compare
26970 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26971 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26972 non-aggregate element types.
26973 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26974 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26976 2017-04-27 Richard Biener <rguenther@suse.de>
26978 PR middle-end/80533
26979 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26980 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26981 keeping a reference to a trailing array.
26983 2017-04-27 Richard Biener <rguenther@suse.de>
26985 PR middle-end/80539
26986 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26987 being in loop-closed SSA form conservatively.
26988 (chrec_fold_multiply_poly_poly): Likewise.
26990 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26992 PR middle-end/79665
26993 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26994 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26996 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26999 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27000 (aarch64_function_arg_alignment): Return unsigned int again, but still
27001 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27002 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27003 Don't emit -Wpsabi note.
27004 (aarch64_function_arg_boundary): Likewise.
27005 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27008 2017-04-26 Nathan Sidwell <nathan@acm.org>
27010 * tree.h (crc32_unsigned_n): Declare.
27011 (crc32_unsigned, crc32_unsigned): Make inline.
27012 * tree.c (crc32_unsigned_bits): Replace with ...
27013 (crc32_unsigned_n): ... this.
27014 (crc32_unsigned, crc32_byte): Remove.
27015 (crc32_string): Remove unnecessary braces.
27017 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27019 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27020 * ipa-inline-analysis.c (MAX_TIME): Remove.
27021 (account_size_time): Use sreal for time.
27022 (dump_inline_summary): Update.
27023 (estimate_function_body_sizes): Update.
27024 (estimate_edge_size_and_time): Update.
27025 (estimate_calls_size_and_time): Update.
27026 (estimate_node_size_and_time): Update.
27027 (inline_merge_summary): Update.
27028 (inline_update_overall_summary): Update.
27029 (estimate_time_after_inlining): Update.
27030 (inline_read_section): Update.
27031 (inline_write_summary): Update.
27032 * ipa-inline.c (compute_uninlined_call_time): Update.
27033 (compute_inlined_call_time): Update.
27034 (recursive_inlining): Update.
27035 (inline_small_functions): Update.
27036 (dump_overall_stats): Update.
27037 * ipa-inline.h: Include sreal.h.
27038 (size_time_entry): Turn time to sreal.
27039 (inline_summary): Turn self_time nad time to sreal.
27041 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27043 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27045 (sreal::stream_out, sreal::stream_in): New.
27046 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27048 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27050 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27053 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27056 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27058 (dimode_scalar_chain::compute_convert_gain): Ditto.
27059 (dimode_scalar_chain::make_vector_copies): Ditto.
27060 (dimode_scalar_chain::convert_reg): Ditto.
27061 (dimode_scalar_chain::convert_insn): Ditto.
27062 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27063 (VI248_AVX512BW_1): New mode iterator.
27064 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27065 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27068 2017-04-25 Martin Sebor <msebor@redhat.com>
27070 PR tree-optimization/80497
27071 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27072 constants are representable in HOST_WIDE_INT.
27073 (parse_directive): Ditto.
27075 2017-04-25 Martin Sebor <msebor@redhat.com>
27078 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27079 (new_zero_array): Adjust signature.
27080 (dom_info::dom_init): Used unsigned rather that size_t.
27081 (dom_info::dom_info): Same.
27083 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27084 Jakub Jelinek <jakub@redhat.com>
27087 * config/arm/arm.c: Include gimple.h.
27088 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27089 returns negative, increment ncrn only if it returned positive.
27090 (arm_needs_doubleword_align): Return int instead of bool,
27091 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27092 members, but if there is any such non-FIELD_DECL
27093 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27094 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27095 returns negative, increment nregs only if it returned positive.
27096 (arm_setup_incoming_varargs): Likewise.
27097 (arm_function_arg_boundary): Emit -Wpsabi note if
27098 arm_needs_doubleword_align returns negative, return
27099 DOUBLEWORD_ALIGNMENT only if it returned positive.
27101 2017-04-25 Marek Polacek <polacek@redhat.com>
27104 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27105 first argument to type.
27107 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27110 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27111 type checks to test for compatibility instead of equality.
27113 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27114 Jakub Jelinek <jakub@redhat.com>
27117 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27119 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27120 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27121 the alignment computation, but return their maximum in warn_alignment.
27122 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27123 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27125 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27126 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27129 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27131 * config/arc/simdext.md (dmpyh): Fix typo.
27133 2017-04-25 Richard Biener <rguenther@suse.de>
27135 PR tree-optimization/80492
27136 * alias.c (compare_base_decls): Handle registers with asm
27137 specification conservatively.
27138 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27139 compare_base_decls returning dont-know properly.
27141 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27143 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27144 (legitimate_offset_address_p): New function.
27145 (arc_legitimate_address_p): Use above function.
27147 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27149 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27151 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27153 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27154 ACCH registers whenever they are available.
27156 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27158 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27159 double regs fix when not used.
27161 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27163 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27165 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27166 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27168 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27170 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27171 when emitting short ADD instructions.
27173 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27175 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27177 (cmpsi_cc_c_insn): Likewise.
27178 (cbranchsi4_scratch): Compute proper instruction length using
27179 compact_hreg_operand.
27180 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27182 2017-04-25 Richard Biener <rguenther@suse.de>
27184 PR middle-end/80509
27185 * passes.c (pass_manager::pass_manager): Initialize
27186 m_name_to_pass_map.
27188 2017-04-25 Richard Biener <rguenther@suse.de>
27190 PR tree-optimization/79201
27191 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27193 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27196 * config/s390/vector.md: Split MEM->GPR vector moves for
27197 non-s_operand addresses.
27199 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27202 * config/s390/predicates.md (reload_const_wide_int_operand): New
27204 * config/s390/s390.md ("movti"): Remove d/P alternative.
27205 ("movti_bigconst"): New pattern definition.
27207 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27210 * s390-protos.h (s390_expand_cs_hqi): Removed.
27211 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27212 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27213 modes as well as CCZ1mode and CCZmode.
27214 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27215 signature of s390_emit_compare_and_swap.
27216 (s390_expand_cs_hqi): Likewise, make static.
27217 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27218 compare-and-swap, in some cases.
27219 (s390_expand_cs): Wrapper function.
27220 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27222 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27223 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27224 patterns for small and large integers. Forbid symref memory operands.
27225 Move expander to s390.c. Require cc register.
27226 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27227 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27228 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27229 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27230 symref memory operands. Remove CC mode and call s390_match_ccmode
27232 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27234 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27236 * config/s390/s390.md (define_peephole2): New peephole to help
27237 combining the load-and-test pattern with volatile memory.
27239 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27241 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27242 with CCZmode for TARGET_Z196.
27244 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27246 PR rtl-optimization/80501
27247 * combine.c (make_compound_operation_int): Set subreg_code to SET
27248 even for AND with mask of the sign bit of mode.
27250 PR rtl-optimization/80500
27251 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27252 sum's initial value.
27254 2017-04-25 Julian Brown <julian@codesourcery.com>
27255 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27257 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27259 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27261 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27263 2017-04-25 Julian Brown <julian@codesourcery.com>
27264 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27266 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27267 (thunderx2t99_sha): New Reservation.
27269 2017-04-25 Julian Brown <julian@codesourcery.com>
27270 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27272 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27273 type for 1-element load.
27275 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27277 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27279 2017-04-24 Martin Jambor <mjambor@suse.cz>
27281 PR tree-optimization/80293
27282 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27283 char arrays not totally scalarizable if it is false.
27284 (analyze_all_variable_accesses): Pass correct value in the new
27285 parameter. Add a statistics counter.
27287 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27289 PR middle-end/79931
27290 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27292 2017-04-24 Richard Biener <rguenther@suse.de>
27294 PR tree-optimization/80494
27295 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27296 out for complex types.
27298 2017-04-24 Richard Biener <rguenther@suse.de>
27300 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27301 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27302 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27303 (DFS): Adjust and never fail.
27304 (sccvn_dom_walker::fail): Remove.
27305 (sccvn_dom_walker::before_dom_children): Adjust.
27306 (run_scc_vn): Likewise and never fail.
27307 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27308 (pass_fre::execute): Likewise.
27310 2017-04-24 Richard Biener <rguenther@suse.de>
27312 PR tree-optimization/79725
27313 * tree-ssa-sink.c (statement_sink_location): Return whether
27314 failure reason was zero uses. Move that check later.
27315 (sink_code_in_bb): Deal with zero uses by removing the stmt
27318 2017-04-24 Richard Biener <rguenther@suse.de>
27321 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27322 pointer-based references.
27324 2017-04-24 Richard Biener <rguenther@suse.de>
27327 * pass_manager.h (pass_manager::operator new): Remove.
27328 (pass_manager::operator delete): Likewise.
27329 * passes.c (pass_manager::operator new): Remove.
27330 (pass_manager::operator delete): Likewise.
27331 (pass_manager::pass_manager): Zero individual pass members.
27333 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27336 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27337 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27338 Check "XEXP (src, 1)" operand here.
27339 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27340 Check "XEXP (src, 1)" operand here.
27341 (dimode_scalar_chain::make_vector_copies): Detect count register
27342 of a shift instruction. Zero extend count register from QImode
27343 to DImode to satisfy vector shift pattern count operand predicate.
27344 Substitute vector shift count operand with a DImode copy.
27345 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27348 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27350 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27351 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27352 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27353 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27354 (UNSPEC_NOREX_MEM): Remove definition.
27356 2017-04-21 Richard Biener <rguenther@suse.de>
27358 PR tree-optimization/79547
27359 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27360 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27361 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27362 without any constraints.
27364 2017-04-21 Richard Biener <rguenther@suse.de>
27366 PR tree-optimization/78847
27367 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27369 2017-04-21 Richard Biener <rguenther@suse.de>
27371 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27372 (build_distinct_type_copy): Likewise.
27373 (build_variant_type_copy): Likewise.
27374 * tree.c (build_qualified_type): Pass down mem-stat info.
27375 (build_distinct_type_copy): Likewise.
27376 (build_variant_type_copy): Likewise.
27378 2017-04-21 Richard Biener <rguenther@suse.de>
27380 PR tree-optimization/80237
27381 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27383 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27384 for a simplified result.
27386 2016-04-21 Richard Biener <rguenther@suse.de>
27388 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27389 sth as strict as a simple_iv but a chrec without symbols and an
27390 operand defined in the loop we are peeling (and not some subloop).
27391 (propagate_constants_for_unrolling): Propagate all constants.
27393 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27396 * config/i386/i386.c (print_reg): Remove assert for disalowed
27397 regno values, call output_operand_lossage instead.
27399 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27402 * config/i386/constraints.md (Yc): New register constraint.
27403 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27404 Use Yc constraint for alternative 2 of operand 0. Remove
27405 preferred_for_speed attribute.
27407 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27409 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27410 lastprivate clauses in SIMT case.
27412 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27414 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27416 2017-04-20 Richard Biener <rguenther@suse.de>
27418 PR tree-optimization/57796
27419 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27420 as N scalar stores.
27421 (vect_model_load_cost): Cost gathers as N scalar loads.
27423 2017-04-20 Richard Biener <rguenther@suse.de>
27425 * ggc-page.c (ggc_allocated_p): Rename to ...
27426 (safe_lookup_page_table_entry): ... this and return the lookup
27428 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27430 2017-04-20 Richard Biener <rguenther@suse.de>
27432 PR tree-optimization/80453
27433 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27434 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27435 from the conditions.
27436 (vn_phi_eq): Pass them down.
27437 (vn_phi_lookup): Record them.
27438 (vn_phi_insert): Likewise.
27440 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27442 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27443 uninitialized variable warning to avoid buffer overrun.
27445 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27448 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27449 is suppressed for '{ 0 }' in C.
27451 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27453 * BASE-VER: Set to 8.0.0.
27455 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27457 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27458 priority .init_array and .fini_array section with SECTION_NOTYPE
27461 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27463 PR middle-end/80423
27464 * tree.h (build_array_type): Add typeless_storage default argument.
27465 * tree.c (type_cache_hasher::equal): Also compare
27466 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27467 (build_array_type): Add typeless_storage argument, set
27468 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27470 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27471 (build_array_type): Likewise. Add typeless_storage argument.
27473 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27474 Jakub Jelinek <jakub@redhat.com>
27476 PR tree-optimization/80426
27477 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27478 operation on symbolic operands, also compute the overflow for the
27479 invariant part when the operation degenerates into a negation.
27481 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27484 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27485 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27488 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27490 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27493 * config/avr/avr.c (tree.h): Include it.
27494 (cgraph.h): Include it.
27495 (avr_encode_section_info): Don't warn for uninitialized progmem
27496 variable if it's just an alias.
27498 2017-04-19 Richard Biener <rguenther@suse.de>
27501 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27502 when needed by AutoPGO.
27504 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27507 * doc/lto.texi: Remove an extra 'that'.
27509 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27511 PR rtl-optimization/80429
27512 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27513 are only used in debug insns.
27515 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27516 Vladimir Makarov <vmakarov@redhat.com>
27518 * config/sparc/predicates.md (input_operand): Add comment. Return
27519 true for any memory operand when LRA is in progress.
27520 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27522 2017-04-18 Jeff Law <law@redhat.com>
27525 * mips.md ({return,simple_return}_internal): Do not overwrite
27528 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27530 PR tree-optimization/80443
27531 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27532 instead of adding 1, subtract -1 and similarly instead of subtracting
27535 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27537 PR rtl-optimization/80357
27538 * haifa-sched.c (tmp_bitmap): New variable.
27539 (model_recompute): Handle duplicate use records.
27540 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27541 (free_global_sched_pressure_data): Free it.
27543 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27546 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27547 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27548 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27549 instead of SYSTEM_HEADER_DIR.
27551 2017-04-18 Jeff Law <law@redhat.com>
27553 PR middle-end/80422
27554 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27555 predecessors after walking up the insn chain.
27557 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27560 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27561 sizetype type into debug info.
27563 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27566 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27567 unneeded test for TARGET_UPPER_REGS_SF.
27568 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27570 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27573 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27574 instead of gsi_after_labels.
27576 2017-04-18 Jeff Law <law@redhat.com>
27578 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27582 2017-04-13 Jeff Law <law@redhat.com>
27583 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27584 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27586 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27589 * config/avr/avr.c (intl.h): Include it.
27590 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27592 2017-04-18 Martin Liska <mliska@suse.cz>
27594 PR gcov-profile/78783
27595 * gcov-tool.c (gcov_output_files): Validate that destination
27596 file is either removed by the tool or by a user.
27598 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27599 Guy Benyei <guybe@mellanox.com>
27601 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27602 block, and do not negate it, the stored id is already negative.
27604 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27606 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27608 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27611 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27612 masks of options that should be turned off if the VSX vector
27613 options are turned off.
27614 (OTHER_P8_VECTOR_MASKS): Likewise.
27615 (OTHER_VSX_VECTOR_MASKS): Likewise.
27616 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27617 rs6000_disable_incompatible_switches to validate no type switches
27619 (rs6000_incompatible_switch): New function to disallow turning on
27620 other vector options if -mno-vsx, -mno-power8-vector, or
27621 -mno-power9-vector are specified.
27623 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27625 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27627 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27629 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27630 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27631 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27632 (ARG_POINTER_CFA_OFFSET): Likewise.
27634 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27636 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27637 conditions to take advantage of various optimizations.
27639 2017-04-13 Jeff Law <law@redhat.com>
27641 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27642 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27643 (zero_extendsidi2_dext): Likewise.
27645 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27648 * fold-const.c (fold_ternary_loc): Revert
27649 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27652 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27654 PR rtl-optimization/80343
27655 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27656 new scratch pseudo.
27658 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27661 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27662 to ubsan_encode_value.
27664 2017-04-13 Jeff Law <law@redhat.com>
27666 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27667 appearing in DEBUG_INSNs.
27669 2017-04-13 Martin Liska <mliska@suse.cz>
27671 PR gcov-profile/80413
27672 * gcov-io.c (gcov_write_string): Copy to buffer just when
27673 allocated size is greater than zero.
27675 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27678 * dwarf2out.c (decls_for_scope): Ignore declarations of
27679 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27681 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27684 * ipa-visibility.c (non_local_p): Fix typos.
27685 (localize_node): When localizing symbol in same comdat group,
27686 dissolve the group only when we know external symbols are going
27688 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27690 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27692 PR tree-optimization/79390
27693 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27694 order does not result in usable sequence, retry with reversed operand
27700 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27701 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27702 op0 instead of fold_convert_loc (loc, type, arg0).
27704 2017-04-12 Jeff Law <law@redhat.com>
27706 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27707 has a delay slot in the generated code.
27709 * config/cris/cris.md (cris_preferred_reload_class): Return
27710 GENNONACR_REGS rather than GENERAL_REGS.
27712 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27715 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27716 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27717 signedness of the result type.
27719 2017-04-12 Richard Biener <rguenther@suse.de>
27720 Jeff Law <law@redhat.com>
27722 PR tree-optimization/80359
27723 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27724 trim stores to TARGET_MEM_REFs.
27726 2017-04-12 Richard Biener <rguenther@suse.de>
27728 PR tree-optimization/79390
27729 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27730 threading case even more.
27732 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27735 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27736 for quad_address_p for TImode, instead of just not indexed_address.
27738 2017-04-12 Richard Biener <rguenther@suse.de>
27739 Bernd Edlinger <bernd.edlinger@hotmail.de>
27741 PR middle-end/79671
27742 * alias.c (component_uses_parent_alias_set_from): Handle
27743 TYPE_TYPELESS_STORAGE.
27744 (get_alias_set): Likewise.
27745 * tree-core.h (tree_type_common): Add typeless_storage flag.
27746 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27747 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27748 for types containing members with TYPE_TYPELESS_STORAGE.
27749 (place_field): Likewise.
27750 (layout_type): Likewise for ARRAY_TYPE.
27751 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27752 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27753 TYPE_TYPELESS_STORAGE.
27754 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27756 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27759 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27760 first argument to type.
27762 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27766 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27767 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27768 (rs6000_expand_binop_builtin): Likewise.
27769 (rs6000_expand_ternop_builtin): Likewise; also add missing
27770 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27771 vshasigma built-ins.
27772 * doc/extend.texi: Document that vec_xxpermdi's third argument
27773 must be a constant.
27775 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27777 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27778 Use shift_const cost parameter when calculating gain of STV shifts.
27780 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27782 PR rtl-optimization/70478
27783 * lra-constraints.c (process_alt_operands): Check memory for
27784 disfavoring memory insn operand.
27786 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27788 PR middle-end/80100
27789 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27790 left shift in unsigned HOST_WIDE_INT type.
27792 PR rtl-optimization/80385
27793 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27794 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27797 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27798 if they have any depend clauses.
27800 2017-04-11 Martin Liska <mliska@suse.cz>
27803 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27804 * ipa-split.c (split_function): Create a local comdat symbol
27805 if caller is in a comdat group.
27807 2017-04-11 Martin Liska <mliska@suse.cz>
27810 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27813 2017-04-11 Martin Sebor <msebor@redhat.com>
27815 PR middle-end/80364
27816 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27817 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27819 (directive::set_width, directive::set_precision, format_character):
27821 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27824 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27827 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27828 conflict, set target->arch_name instead of target->cpu_name.
27830 2017-04-11 Richard Biener <rguenther@suse.de>
27832 PR tree-optimization/80374
27833 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27834 build_zero_cst, remove fold_convertible_p check again.
27836 2017-04-11 Martin Liska <mliska@suse.cz>
27839 * ubsan.c (instrument_object_size): Do not instrument register
27842 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27845 * config/i386/i386-builtin-types.def
27846 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27847 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27848 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27849 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27850 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27851 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27852 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27853 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27854 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27855 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27856 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27857 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27858 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27859 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27860 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27861 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27862 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27863 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27864 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27865 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27866 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27867 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27868 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27869 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27870 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27871 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27872 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27873 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27874 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27875 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27876 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27877 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27878 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27879 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27880 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27881 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27882 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27883 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27884 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27885 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27886 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27887 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27888 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27889 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27890 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27891 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27893 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27894 flag to second_arg_count, handle 4 argument function type _COUNT
27895 aliases, handle second_arg_count on second argument rather than last.
27897 2017-04-10 Jeff Law <law@redhat.com>
27899 PR tree-optimization/80374
27900 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27901 record anything if we can not convert integer_zero_node to the
27904 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27907 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27908 Enhance special handling given to the TARGET_P9_MINMAX option in
27909 relation to certain other options.
27911 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27913 PR tree-optimization/80153
27914 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27915 remove POINTER_PLUS_EXPR's base part directly, rather than through
27918 2017-04-10 Richard Biener <rguenther@suse.de>
27919 Bin Cheng <bin.cheng@arm.com>
27921 PR tree-optimization/80153
27922 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27923 the first element of pointer type aff_tree. Build result expr in
27925 (add_elt_to_tree): Convert to type unconditionally. Remove other
27926 fold_convert calls.
27927 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27928 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27930 2017-04-10 Richard Biener <rguenther@suse.de>
27932 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27935 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27937 PR rtl-optimization/70478
27938 * lra-constraints.c (curr_small_class_check): New.
27939 (update_and_check_small_class_inputs): New.
27940 (process_alt_operands): Update curr_small_class_check. Disfavor
27941 alternative insn memory operands. Check available regs for small
27944 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27947 * config/mips/mips.opt (-mvirt): Update description.
27948 * doc/invoke.texi (-mvirt): Likewise.
27950 2017-04-10 Richard Biener <rguenther@suse.de>
27952 PR middle-end/80362
27953 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27954 looking for NEGATE_EXPR in -A / -B to A / B folding.
27956 2017-04-10 Martin Liska <mliska@suse.cz>
27958 PR gcov-profile/80224
27959 * gcov.c (print_usage): Fix usage string.
27960 (get_gcov_intermediate_filename): Remove.
27961 (output_gcov_file): Use both for normal and intermediate format.
27962 (generate_results): Do not initialize special file for
27963 intermediate format.
27965 2017-04-10 Richard Biener <rguenther@suse.de>
27967 PR tree-optimization/80304
27968 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27971 2017-04-10 Nathan Sidwell <nathan@acm.org>
27974 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27975 (rs6000_init_builtins): Use it.
27977 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27979 * config/arm/arm.md (<mrc>): Add mode to SET source.
27980 (<mrrc>): Likewise.
27982 2017-04-10 Richard Biener <rguenther@suse.de>
27984 PR middle-end/80344
27985 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27987 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27990 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27991 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27992 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27993 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27994 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27995 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27996 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27997 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27998 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27999 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28000 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28001 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28002 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28003 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28004 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28005 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28006 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28007 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28008 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28009 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28010 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28011 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28012 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28014 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28016 PR rtl-optimization/70478
28017 * lra-constraints.c: Reverse the last patch.
28019 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28021 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28022 Add comment for WCHAR_T.
28024 2017-04-08 Martin Liska <mliska@suse.cz>
28027 2017-04-07 Martin Liska <mliska@suse.cz>
28030 * ipa-split.c (split_function): Add function part to a same comdat
28033 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28036 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28038 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28040 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28041 (vec_pairing): ...to this.
28042 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28043 (rs6000_sched_init): Adjust for name change.
28044 (struct rs6000_sched_context): Likewise.
28045 (rs6000_init_sched_context): Likewise.
28046 (rs6000_set_sched_context): Likewise.
28048 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28054 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28056 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28057 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28058 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28060 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28062 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28064 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28066 PR rtl-optimization/70703
28067 * ira-color.c (update_conflict_hard_regno_costs): Use
28068 int64_t instead of HOST_WIDE_INT.
28070 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28072 PR rtl-optimization/70478
28073 * lra-constraints.c (process_alt_operands): Disfavor alternative
28074 insn memory operands.
28076 2017-04-07 Jeff Law <law@redhat.com>
28078 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28079 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28081 2017-04-07 Martin Liska <mliska@suse.cz>
28084 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28085 Show error message instead of an ICE.
28087 2017-04-07 Martin Liska <mliska@suse.cz>
28090 * ipa-split.c (split_function): Add function part to a same comdat
28093 2017-04-07 Richard Biener <rguenther@suse.de>
28095 PR middle-end/80341
28096 * tree.c (get_unwidened): Also handle ! for_type case for
28098 * convert.c (do_narrow): Split out from ...
28099 (convert_to_integer_1): ... here. Do not pass final truncation
28100 type to get_unwidened for TRUNC_DIV_EXPR.
28102 2017-04-07 Richard Biener <rguenther@suse.de>
28104 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28106 (aff_combination_const): Adjust.
28107 (aff_combination_scale): Likewise.
28108 (aff_combination_add_elt): Likewise.
28109 (aff_combination_add_cst): Likewise.
28110 (aff_combination_convert): Likewise.
28111 (add_elt_to_tree): Likewise. Remove unused argument.
28112 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28114 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28116 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28118 * config/arm/arm.c (arm_default_short_enums): Use
28119 ARM_DEFAULT_SHORT_ENUMS.
28120 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28122 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28125 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28126 members with redundant out-of-class redeclaration.
28128 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28131 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28132 * config/i386/i386.md (*zero_extendsidi2):
28133 Add (?*x,*x) and (?*v,*v) alternatives.
28135 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28138 * config/i386/i386.c (ix86_expand_builtin)
28139 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28140 mode from insn data. Convert operands to insn operand mode.
28141 Copy operands that don't satisfy insn predicate to a register.
28143 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28145 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28148 2017-04-06 Richard Biener <rguenther@suse.de>
28150 PR tree-optimization/80334
28151 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28152 preserve alignment of accesses.
28154 2017-04-06 Richard Biener <rguenther@suse.de>
28156 PR tree-optimization/80262
28157 * tree-sra.c (build_ref_for_offset): Preserve address-space
28159 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28160 Drop useless address-space information on MEM_REF offsets.
28162 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28164 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28166 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28168 PR rtl-optimization/70703
28169 * ira-color.c (update_conflict_hard_regno_costs): Use
28170 HOST_WIDE_INT instead of long.
28172 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28175 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28176 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28178 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28179 for x86_64 target. Handle -m3dnowa option.
28181 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28183 PR rtl-optimization/70703
28184 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28185 (update_conflict_hard_regno_costs): Use long instead of unsigned
28186 arithmetic for cost calculation.
28188 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28189 Bernd Edlinger <bernd.edlinger@hotmail.de>
28192 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28195 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28198 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28199 ptr_mode with Pmode throughout.
28200 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28201 into probe_stack_range and use DImode.
28203 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28206 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28207 call_eh_return is true.
28209 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28211 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28212 Initialize last_match_fntype_index.
28214 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28217 * tree-nvr.c: Include internal-fn.h.
28218 (pass_return_slot::execute): Ignore internal calls without
28221 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28222 Richard Biener <rguenther@suse.de>
28225 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28226 captures used multiple times, except for the last use.
28227 * generic-match-head.c: Include gimplify.h.
28229 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28231 PR tree-optimization/79390
28232 * target.h (struct noce_if_info): Declare.
28233 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28234 * target.def (noce_conversion_profitable_p): New target hook.
28235 * ifcvt.h (struct noce_if_info): New type, moved from ...
28236 * ifcvt.c (struct noce_if_info): ... here.
28237 (noce_conversion_profitable_p): Renamed to ...
28238 (default_noce_conversion_profitable_p): ... this. No longer
28240 (noce_try_store_flag_constants, noce_try_addcc,
28241 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28242 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28243 instead of noce_conversion_profitable_p.
28244 * config/i386/i386.c: Include ifcvt.h.
28245 (ix86_option_override_internal): Don't override
28246 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28247 (ix86_noce_conversion_profitable_p): New function.
28248 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28249 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28250 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28251 * doc/tm.texi: Regenerated.
28253 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28255 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28258 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28261 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28262 instructions for small multiply cores.
28264 2017-04-04 Jeff Law <law@redhat.com>
28266 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28268 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28269 that are not set by the loop over the elements.
28271 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28274 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28275 int mode, convert_modes it to mode as unsigned, otherwise use
28276 lowpart_subreg to mode rather than SImode.
28277 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28278 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28279 Use DImode instead of SImode for the shift count operand.
28280 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28283 2017-04-04 Richard Biener <rguenther@suse.de>
28285 PR middle-end/80281
28286 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28287 arithmetic done for the negate or the plus. Simplify.
28288 (A - (-B) -> A + B): Likewise.
28289 * fold-const.c (split_tree): Make sure to not negate pointers.
28291 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28293 PR rtl-optimization/60818
28294 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28295 a compare of comparisons with the thing compared if this results
28296 in a different machine mode.
28298 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28300 * alias.c (base_alias_check): Fix typo in comment.
28301 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28302 * cgraphunit.c (symbol_table::compile): Likewise.
28303 * collect2.c (maybe_run_lto_and_relink): Likewise.
28304 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28305 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28306 * config/avr/avr.c (avr_map_op_t): Likewise.
28307 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28308 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28309 * config/epiphany/epiphany.md (movcc): Likewise.
28310 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28311 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28313 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28314 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28315 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28316 * config/sh/sh.c (sh_rtx_costs): Likewise.
28317 * fold-const.c (fold_truth_andor): Likewise.
28318 * genautomata.c (collapse_flag): Likewise.
28319 * gengtype.h (struct type::u::s): Likewise.
28320 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28321 * input.c (FORMAT_AMOUNT): Likewise.
28322 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28323 (known_aggs_to_agg_replacement_list): Likewise.
28324 * ipa-inline-analysis.c: Likewise.
28325 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28326 * ipa-polymorphic-call.c
28327 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28328 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28329 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28331 * modulo-sched.c (apply_reg_moves): Likewise.
28332 * omp-expand.c (build_omp_regions_1): Likewise.
28333 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28334 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28335 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28336 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28337 * value-prof.c: Likewise.
28338 * var-tracking.c (val_reset): Likewise.
28340 2017-04-03 Richard Biener <rguenther@suse.de>
28342 PR tree-optimization/80275
28343 * fold-const.c (split_address_to_core_and_offset): Handle
28346 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28348 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28349 descriptors is at least equal to that of functions.
28351 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28353 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28355 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28358 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28359 (mov<IMOD4:mode>): New expander.
28360 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28362 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28364 PR rtl-optimization/79405
28365 * fwprop.c (propagations_left): New variable.
28366 (forward_propagate_into): Decrement it.
28367 (fwprop_init): Initialize it.
28368 (fw_prop): If the variable has reached zero, stop propagating.
28369 (fwprop_addr): Ditto.
28371 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28374 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28375 a FUNCTION_DECL, pass it as decl instead of origin to
28378 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28380 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28383 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28386 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28387 TARGET_VSX_SMALL_INTEGER.
28389 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28391 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28392 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28394 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28396 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28397 extraction from odd-numbered MSA register.
28399 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28401 PR middle-end/80173
28402 * expmed.c (store_bit_field_1): Don't attempt to create
28403 a word subreg out of hard registers wider than word if they
28404 have HARD_REGNO_NREGS of 1 for their mode.
28406 PR middle-end/80163
28407 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28408 conversions to integer types wider than word and pointer.
28411 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28412 (rtx_equal_for_cselib_p): Pass 0 to it.
28413 * cselib.c (cselib_hasher::equal): Likewise.
28414 (rtx_equal_for_cselib_1): Add depth argument. If depth
28415 is 128, don't look up VALUE locs and punt. Increment
28416 depth in recursive calls when walking VALUE locs.
28418 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28420 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28421 (make_gcov_file_name): Use the canonical path name for generating
28423 (read_line): Fix handling of files with ascii null bytes.
28425 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28427 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28428 to initialise a vector register instead
28429 of using a const_int.
28431 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28433 PR translation/80189
28434 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28435 diagnostic messages.
28437 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28440 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28441 (dfp_diex_<mode>): Update mode of operand 1.
28442 * doc/extend.texi (dxex, dxexq): Document change to return type.
28443 (diex, diexq): Document change to argument type.
28445 2017-03-30 Martin Jambor <mjambor@suse.cz>
28448 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28449 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28450 it reflects the signature changes performed at the callee side.
28451 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28452 to cgraph_build_function_type_skip_args.
28453 (build_function_decl_skip_args): Adjust call to the above function.
28455 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28458 * config/i386/sse.md
28459 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28460 register as dest whenever it is a MEM not rtx_equal_p to the
28461 corresponding dup operand, and when forcing into reg move the
28462 reg into the memory afterwards.
28463 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28464 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28465 for the force_reg mode.
28466 (avx512vl_vextractf128<mode>): Use register as dest either
28467 always when a MEM, or when it is a MEM not rtx_equal_p to the
28468 corresponding dup operand, or even not when it is a CONST_VECTOR
28469 depending on the mode and lo vs. hi.
28470 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28472 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28473 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28474 Likewise. Require that operands[2] is even.
28475 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28476 Remove extraneous parens. Require that operands[2] is a multiple
28478 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28479 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28480 disallow memory then.
28482 2017-03-30 Richard Biener <rguenther@suse.de>
28484 PR tree-optimization/77498
28485 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28486 to non-constants over backedges.
28488 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28490 PR rtl-optimization/80233
28491 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28492 as last_combined_insn. Do not test for BARRIER_P separately.
28494 2017-03-29 Andreas Schwab <schwab@suse.de>
28497 * calls.c (prepare_call_address): Convert funexp to Pmode before
28498 copying to temp reg.
28500 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28502 PR tree-optimization/80158
28503 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28504 Handle possible future case of more than one alternate
28506 (replace_rhs_if_not_dup): Likewise.
28507 (replace_one_candidate): Likewise.
28509 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28511 PR rtl-optimization/80193
28512 * ira.c (ira): Do not check allocation for LRA.
28514 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28516 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28517 (nvptx_output_simt_exit): Declare.
28518 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28519 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28520 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28521 (nvptx_declare_function_name): ...here. Emit declaration of local
28522 memory space buffer for omp_simt_enter insn.
28523 (nvptx_output_unisimt_switch): New.
28524 (nvptx_output_softstack_switch): New.
28525 (nvptx_output_simt_enter): New.
28526 (nvptx_output_simt_exit): New.
28527 * config/nvptx/nvptx.h (struct machine_function): New fields
28528 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28529 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28530 (UNSPECV_SIMT_EXIT): Ditto.
28531 (omp_simt_enter_insn): New insn.
28532 (omp_simt_enter): New expansion.
28533 (omp_simt_exit): New insn.
28534 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28536 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28537 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28538 (expand_GOMP_SIMT_EXIT): New.
28539 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28540 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28541 (GOMP_SIMT_EXIT): Ditto.
28542 * target-insns.def (omp_simt_enter): New insn.
28543 (omp_simt_exit): Ditto.
28544 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28546 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28547 (lower_rec_input_clauses): Likewise.
28548 (lower_lastprivate_clauses): Handle SIMT privatization.
28550 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28551 (ompdevlow_adjust_simt_enter): New.
28552 (find_simtpriv_var_op): New.
28553 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28554 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28556 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28557 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28558 (copy_decl_for_dup_finish): Ditto.
28560 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28562 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28565 * config/i386/i386.c (ix86_option_override_internal): Always
28566 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28568 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28570 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28572 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28574 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28575 mark new edge's irreducible flag accordign to it.
28576 (vect_do_peeling): Check loop preheader edge's irreducible flag
28577 and pass it to function slpeel_add_loop_guard.
28579 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28581 PR tree-optimization/80218
28582 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28583 Update block frequencies and counts.
28585 2017-03-28 Richard Biener <rguenther@suse.de>
28587 PR tree-optimization/78644
28588 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28589 of a simplification result we may not use it at all.
28591 2017-03-28 Richard Biener <rguenther@suse.de>
28594 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28595 without arguments, generate default definition of a SSA name.
28597 2017-03-28 Richard Biener <rguenther@suse.de>
28599 PR middle-end/80222
28600 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28601 TYPE_REF_CAN_ALIAS_ALL references.
28602 * fold-const.c (fold_indirect_ref_1): Likewise.
28604 2017-03-28 Martin Liska <mliska@suse.cz>
28607 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28608 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28610 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28611 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28613 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28614 (EXTRA_SPECS): Define.
28615 (SUBTARGET_EXTRA_SPECS): Likewise.
28616 (SUBTARGET_CPP_SPEC): Likewise.
28617 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28618 SUBTARGET_EXTRA_SPECS.
28619 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28621 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28623 * config/arc/simdext.md (vst64_insn): Update pattern.
28624 (vld32wh_insn): Likewise.
28625 (vld32wl_insn): Likewise.
28626 (vld64_insn): Likewise.
28627 (vld32_insn): Likewise.
28629 2017-03-28 Marek Polacek <polacek@redhat.com>
28632 * fold-const.c (fold_comparison): Use protected_set_expr_location
28633 instead of SET_EXPR_LOCATION.
28635 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28637 * tree.c (add_expr): Avoid name lookup warning.
28639 2017-03-27 Jeff Law <law@redhat.com>
28641 PR tree-optimization/80216
28642 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28643 function name. Limit recursion depth.
28644 (record_temporary_equivalences): Corresponding changes.
28646 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28648 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28651 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28654 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28656 * cfgcleanup.c (reg_note_cfa_p): New array.
28657 (insns_have_identical_cfa_notes): New function.
28658 (old_insns_match_p): Don't cross-jump in between /f
28659 and non-/f instructions. If both i1 and i2 are frame related,
28660 verify all CFA notes, their order and content.
28662 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28665 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28666 HImode and SImode with zero extend to DImode to one insn.
28667 (bswap<mode>2_extenddi): Likewise.
28668 (bswapsi2_extenddi): Likewise.
28669 (bswaphi2_extendsi): Likewise.
28670 (bswaphi2): Combine bswap HImode and SImode into one insn.
28671 Separate memory insns from swapping register.
28672 (bswapsi2): Likewise.
28673 (bswap<mode>2): Likewise.
28674 (bswaphi2_internal): Delete, no longer used.
28675 (bswapsi2_internal): Likewise.
28676 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28677 store, and gpr<-gpr swap insns.
28678 (bswap<mode>2_store): Likewise.
28679 (bswaphi2_reg): Register only splitter, combine with the splitter.
28680 (bswaphi2 splitter): Likewise.
28681 (bswapsi2_reg): Likewise.
28682 (bswapsi2 splitter): Likewise.
28683 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28684 the insns into load, store, and register/register insns.
28685 (bswapdi2_ldbrx): Likewise.
28686 (bswapdi2_load): Likewise.
28687 (bswapdi2_store): Likewise.
28688 (bswapdi2_reg): Likewise.
28690 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28692 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28693 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28695 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28698 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28700 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28701 special handling for target option conflicts between dform
28702 options (-mpower9-dform, -mpower9-dform-vector,
28703 -mpower9-dform-scalar) and -mno-direct-move.
28705 2017-03-27 Richard Biener <rguenther@suse.de>
28707 PR tree-optimization/80181
28708 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28710 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28712 * config/arc/predicates.md (move_double_src_operand): Replace the
28713 call to move_double_src_operand with a call to address_operand.
28715 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28717 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28718 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28719 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28721 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28723 * config/arc/predicates.md (long_immediate_loadstore_operand):
28724 Consider scaled addresses cases.
28726 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28728 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28729 restored when in interrupt.
28730 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28731 doesn't have delay slot.
28733 2017-03-27 Richard Biener <rguenther@suse.de>
28736 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28737 inlined thunk clones.
28739 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28742 * asan.c (instrument_derefs): Copy over last operand from
28743 original COMPONENT_REF to the new COMPONENT_REF with
28744 DECL_BIT_FIELD_REPRESENTATIVE.
28745 * ubsan.c (instrument_object_size): Likewise.
28747 2017-03-27 Richard Biener <rguenther@suse.de>
28749 PR tree-optimization/80170
28750 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28751 sure DR/SCEV didnt fold in constants we do not see when looking
28752 at the reference base alignment.
28754 2017-03-27 Richard Biener <rguenther@suse.de>
28756 PR middle-end/80171
28757 * gimple-fold.c (fold_ctor_reference): Properly guard against
28758 NULL return value from canonicalize_constructor_val.
28760 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28763 * config/i386/i386.c (ix86_expand_builtin)
28764 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28765 flags reg setting and flags reg using instructions.
28766 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28767 clobbering instructions to zero extend op2.
28769 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28771 * doc/install.texi (Configuration) <--with-aix-soname>:
28772 Update link to AIX ld.
28774 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28776 PR rtl-optimization/80160
28777 PR rtl-optimization/80159
28778 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28779 reg_alternate_class into account.
28781 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28784 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28785 to consider in curr_insn_transform.
28787 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28789 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28790 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28791 and emit_mode_inner.
28793 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28795 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28796 argument to the overloaded builtin variants. Use the new flag to
28797 deprecate certain builtin variants.
28798 * config/s390/s390-builtin-types.def: Add new builtin types.
28799 * config/s390/s390-builtins.h: Support new flags field for
28800 overloaded builtins.
28801 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28802 (s390_macro_to_expand): Enable vector float data type.
28803 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28804 builtins by incrementing the __VEC__ version number.
28805 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28807 (s390_resolve_overloaded_builtin): Emit error messages depending
28808 on the builtin flags.
28809 * config/s390/s390.c (s390_expand_builtin): Support additional
28810 flags argument. Change error message to match the messages
28811 emitted in s390-c.c.
28812 * config/s390/s390.md: New UNSPEC_* constants.
28813 (op_type): Add new instruction types.
28814 * config/s390/vecintrin.h: Add new builtins and test data class
28816 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28817 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28818 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28819 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28820 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28821 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28823 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28824 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28825 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28826 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28828 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28829 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28830 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28831 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28833 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28834 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28835 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28836 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28837 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28838 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28839 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28841 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28842 ("vec_scatter_element<V_HW_4:mode>_DI")
28843 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28844 ("vec_fpint<mode>", "vflls")
28845 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28846 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28847 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28848 ("*vec_cmphe<mode>_cc"): ... these.
28850 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28851 mode constant instead of magic value.
28853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28855 * config/s390/s390.c (s390_expand_vec_compare): Support other
28856 vector floating point modes than just V2DF.
28857 (s390_expand_vcond): Likewise.
28858 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28859 (s390_cannot_change_mode_class): Prevent mode changes between TF
28860 and V1TF in vector registers.
28861 * config/s390/s390.md (DF, SF): New mode attributes.
28862 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28863 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28864 SFmode support for VRs.
28865 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28867 (VFT, VF_HW): New mode iterators.
28868 (vw, sdx): New mode attributes.
28869 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28870 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28871 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28872 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28873 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28874 also the new vector floating point modes. Renaming to ...
28876 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28877 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28878 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28879 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28880 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28881 ("vec_unordered<mode>"): ... these.
28883 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28884 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28885 ("*vec_extendv2df"): New insn definitions.
28887 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28889 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28890 ("mulditi3_2", "*muldi3_sign"): New patterns.
28891 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28892 rename the pattern definition.
28894 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28896 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28898 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28900 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28902 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28903 instruction if possible.
28904 * config/s390/vector.md (vec_halfnumelts): New mode
28906 ("*vec_vllezlf<mode>"): New pattern.
28908 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28910 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28911 ("popcountv4si2", "popcountv2di2"): Rename to ...
28912 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28913 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28915 ("popcount<mode>2_vxe"): New pattern.
28917 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28919 * common/config/s390/s390-common.c (processor_flags_table): Add
28921 * config.gcc: Add arch12.
28922 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28923 Default to arch12 for unknown CPU model numbers.
28924 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28925 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28926 PROCESSOR_max sanity check.
28927 * config/s390/s390-opts.h (enum processor_type): Add
28929 * config/s390/s390.c (processor_table): Add arch12.
28930 (s390_expand_builtin): Add check for B_VXE flag.
28931 (s390_issue_rate): Add PROCESSOR_ARCH12.
28932 (s390_get_sched_attrmask): Likewise.
28933 (s390_get_unit_mask): Likewise.
28934 (s390_sched_score): Enable z13 scheduling for arch12.
28935 (s390_sched_reorder): Likewise.
28936 (s390_sched_variable_issue): Likewise.
28937 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28939 (s390_tune_attr): Use z13 scheduling also for arch12.
28940 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28941 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28942 (TARGET_VXE_P): New macros.
28943 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28944 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28945 * config/s390/s390.opt: Add arch12 as processor_type.
28947 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28949 * config/s390/s390.md
28950 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28951 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28952 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28954 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28955 Rename expanders to ...
28957 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28958 ("fixuns_truncdddi2_emu"): ... these.
28960 ("fixuns_trunc<mode>si2_emu"): New expander.
28962 ("*fixuns_truncdfdi2_z13"): Rename to ...
28963 ("*fixuns_truncdfdi2_vx"): ... this.
28965 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28967 * config/s390/2964.md: Remove the single element vector compare
28968 instructions which are no longer used.
28969 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28971 (s390_canonicalize_comparison): Remove handling of DFmode
28973 (s390_expand_vec_compare_scalar): Remove function.
28974 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28975 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28977 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28979 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28981 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28983 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28984 will anyway by matched by mov<mode>_64dfp.
28986 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28988 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28989 vlef/vstef. Add missing operand to vleif.
28991 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28993 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28994 pair for all vector types with 64 bit elements.
28995 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28996 * config/s390/vector.md (V_HW_64): ... here.
28997 (V_128_NOSINGLE): New mode iterator.
28998 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28999 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29000 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29001 ("*vec_load_pairv2di"): Change to ...
29002 ("*vec_load_pair<mode>"): ... this one.
29004 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29006 * config/s390/constraints.md: Add comments.
29007 (jKK): Reject element sizes > 8 bytes.
29008 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29010 * config/s390/s390.md: Add the s_operand checks formerly in
29011 s390_split_ok_p to various splitters where they are still
29013 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29014 for 128 bit vectors. Plus two splitters.
29016 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29018 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29021 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29024 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29025 error if the boundary argument is not constant.
29027 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29029 PR rtl-optimization/80112
29030 * loop-doloop.c (doloop_condition_get): Don't check condition
29031 if cmp isn't SET with IF_THEN_ELSE src.
29033 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29035 PR tree-optimization/80158
29036 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29037 replacing a candidate statement, also replace it for the
29038 candidate's alternate interpretation.
29039 (replace_rhs_if_not_dup): Likewise.
29040 (replace_one_candidate): Likewise.
29042 2017-03-24 Richard Biener <rguenther@suse.de>
29044 PR tree-optimization/80167
29045 * graphite-isl-ast-to-gimple.c
29046 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29048 (translate_isl_ast_to_gimple::get_rename): Likewise.
29050 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29052 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29053 handling of certain combinations of target options, including the
29054 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29055 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29057 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29060 * config/arm/arm.md (*load_multiple): Add reload_completed to
29061 matching condition.
29063 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29064 Richard Biener <rguenth@suse.de>
29066 PR tree-optimization/79908
29067 PR tree-optimization/80136
29068 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29069 been cast away, gimplify_and_add suffices.
29071 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29073 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29075 2017-03-23 Richard Biener <rguenther@suse.de>
29077 PR tree-optimization/80032
29078 * gimplify.c (gimple_push_cleanup): Forced unconditional
29079 cleanups still have to go to the conditional_cleanups
29082 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29084 PR tree-optimization/80072
29085 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29087 (next_operand_entry_id): Change type to unsigned int.
29088 (sort_by_operand_rank): Make sure to return the right return value
29089 even if unsigned fields are bigger than INT_MAX.
29090 (struct oecount): Change cnt and id type to unsigned int.
29091 (oecount_hasher::equal): Formatting fix.
29092 (oecount_cmp): Make sure to return the right return value
29093 even if unsigned fields are bigger than INT_MAX.
29094 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29097 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29098 TREE_READONLY on result if writing it more than once.
29101 * doc/invoke.texi (-fsanitize=thread): Document that with
29102 -fnon-call-exceptions atomics are not able to throw
29106 * tsan.c: Include tree-eh.h.
29107 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29108 maybe_clean_or_replace_eh_stmt where needed.
29109 (instrument_memory_accesses): Add cfg_changed argument.
29110 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29111 if it returned true.
29112 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29114 PR rtl-optimization/63191
29115 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29116 wrapper function, moved the whole old content into ...
29117 (ix86_delegitimize_address_1): ... this. New inline function.
29118 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29119 true as last argument instead of ix86_delegitimize_address.
29121 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29123 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29124 cortexa57_branch_cost.
29126 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29128 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29130 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29133 * doc/md.texi (Constraints): Document wA constraint.
29134 * config/rs6000/constraints.md (wA): New.
29135 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29136 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29137 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29138 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29140 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29143 * gimplify.c (is_oacc_declared): New function.
29144 (oacc_default_clause): Use it to set default flags for acc declared
29145 variables inside parallel regions.
29146 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29147 declared variables.
29148 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29149 declare attribute to any decl as necessary.
29151 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29154 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29155 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29156 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29157 (arm_arch_lpae): This.
29158 * config/arm/arm.c (arm_arch7ve): Rename into ...
29159 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29160 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29163 2017-03-22 Martin Liska <mliska@suse.cz>
29166 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29167 error message instead of an ICE.
29169 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29171 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29173 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29175 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29177 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29180 2017-03-21 Martin Sebor <msebor@redhat.com>
29182 * doc/extend.texi: Use "cannot" instead of "can't."
29183 * doc/hostconfig.texi: Same.
29184 * doc/install.texi: Same.
29185 * doc/invoke.texi: Same.
29186 * doc/loop.texi: Same.
29187 * doc/md.texi: Same.
29188 * doc/objc.texi: Same.
29189 * doc/rtl.texi: Same.
29190 * doc/tm.texi: Same.
29191 * doc/tm.texi.in: Same.
29192 * doc/trouble.texi: Same.
29194 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29197 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29198 (collect_checksum_attributes): Set it.
29199 (die_checksum_ordered): Use it.
29201 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29203 PR tree-optimization/79908
29204 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29205 change: For a VA_ARG whose LHS has been cast away, use
29206 force_gimple_operand to construct the side effects.
29208 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29210 PR translation/80001
29211 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29212 more amenable to translation.
29213 (oacc_loop_auto_partitions): Likewise.
29215 2017-03-21 Marek Polacek <polacek@redhat.com>
29216 Martin Sebor <msebor@redhat.com>
29218 PR tree-optimization/80109
29219 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29220 on INTEGRAL_TYPE_P.
29222 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29223 Segher Boessenkool <segher@kernel.crashing.org>
29226 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29227 check reg_used_between_p between insn and one of succ or succ2
29228 depending on if succ is artificial insn not inserted into insn
29231 2017-03-21 Martin Liska <mliska@suse.cz>
29233 PR gcov-profile/80081
29234 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29235 * doc/gcc.texi: Include gcov-dump stuff.
29236 * doc/gcov-dump.texi: New file.
29238 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29240 PR rtl-optimization/79150
29241 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29242 conditional jump, if the jump is the last insn of the loop.
29244 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29245 Richard Biener <rguenth@suse.de>
29247 PR tree-optimization/79908
29248 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29249 been cast away, use force_gimple_operand to construct the side
29252 2017-03-21 Martin Liska <mliska@suse.cz>
29254 PR libfortran/79956
29255 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29258 2017-03-21 Brad Spengler <spender@grsecurity.net>
29261 * plugin.c (htab_hash_plugin): New function.
29262 (add_new_plugin): Use it and adjust.
29263 (parse_plugin_arg_opt): Adjust.
29264 (init_one_plugin): Likewise.
29266 2017-03-21 Richard Biener <rguenther@suse.de>
29268 PR tree-optimization/80032
29269 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29270 if set force the cleanup to happen unconditionally.
29271 (gimplify_target_expr): Push inserted clobbers with force_uncond
29272 to avoid them being removed by control-dependent DCE.
29274 2017-03-21 Richard Biener <rguenther@suse.de>
29276 PR tree-optimization/80122
29277 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29278 va_arg_pack_len when the inlined call stmt requires pack
29280 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29282 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29285 * tsan.c (instrument_builtin_call): If the memory model argument
29286 is not a constant, assume it is valid.
29289 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29292 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29294 PR rtl-optimization/79910
29295 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29296 if its dest is used by an insn before I2 (other than the combined
29297 insns themselves, which are properly handled already).
29299 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29302 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29304 * combine.c (record_used_regs): New static function.
29305 (try_combine): Handle situations where there is an additional
29306 instruction between I2 and I3 which needs to have a LOG_LINK
29310 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29312 * combine.c (try_combine): Delete redundant i1 test. Call
29313 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29315 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29318 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29319 alternatives 13/14.
29321 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29323 PR tree-optimization/80054
29324 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29325 the optimization if a PHI or any of its arguments is not dominated
29326 by the candidate's basis. Use gphi* rather than gimple* as
29328 (replace_profitable_candidates): Clean up a gimple* variable that
29329 should be a gphi* variable.
29331 2017-03-20 Martin Sebor <msebor@redhat.com>
29334 * doc/extend.texi (attribute constructor): Document present limitation.
29336 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29339 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29340 __POWER9_VECTOR__ #ifdef control, change template definition to
29341 use Power9-specific built-in function.
29342 (vec_any_eq): Likewise.
29343 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29344 to control outcomes from this test.
29345 (vector_ae_<mode>p): For VEC_F modes, likewise.
29347 2017-03-20 Ian Lance Taylor <iant@google.com>
29349 * config/i386/i386.c (ix86_function_regparm): Save an extra
29350 register for -fsplit-stack with DECL_STATIC_CHAIN.
29352 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29355 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29356 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29358 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29360 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29362 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29365 2017-03-20 Marek Polacek <polacek@redhat.com>
29368 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29370 2017-03-20 Richard Biener <rguenther@suse.de>
29372 PR tree-optimization/80113
29373 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29374 allocate extra SSA name for PHI def.
29375 (add_close_phis_to_outer_loops): Likewise.
29376 (add_close_phis_to_merge_points): Likewise.
29377 (copy_loop_close_phi_args): Likewise.
29378 (copy_cond_phi_nodes): Likewise.
29380 2017-03-20 Martin Liska <mliska@suse.cz>
29382 PR middle-end/79753
29383 * tree-chkp.c (chkp_build_returned_bound): Do not build
29384 returned bounds for a LHS that's not a BOUNDED_P type.
29386 2017-03-20 Martin Liska <mliska@suse.cz>
29390 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29391 COMPLEX_CST and VECTOR_CST.
29393 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29396 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29397 target operand. A new splitter adds the clobber statement in case
29398 the target operand is dead anyway.
29400 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29402 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29403 to age-old versions of binutils and glibc.
29405 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29407 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29409 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29411 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29413 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29415 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29416 requirement for binutils 2.13.
29418 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29420 * combine.c (try_combine): Delete redundant i1 test. Call
29421 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29423 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29425 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29426 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29428 <riscv64-*-elf>: Re-arrange section
29429 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29430 <riscv32-*-linux>: Likewise.
29431 <riscv64-*-elf>: Likewise
29432 <riscv64-*-linux>: Likewise.
29434 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29437 * aarch64.opt(verbose-cost-dump): Fix typo.
29439 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29442 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29443 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29445 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29447 * reload.c (find_reloads): When reloading a nonoffsettable address,
29448 use RELOAD_OTHER for it and its address reloads.
29450 PR rtl-optimization/79910
29451 * combine.c (record_used_regs): New static function.
29452 (try_combine): Handle situations where there is an additional
29453 instruction between I2 and I3 which needs to have a LOG_LINK
29456 2017-03-17 Jeff Law <law@redhat.com>
29458 PR tree-optimization/71437
29459 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29460 conditional in the hash table first.
29461 (vrp_dom_walker::before_dom_children): Extract condition from
29462 ASSERT_EXPR. Record condition, its inverion and any implied
29463 conditions as well.
29465 2017-03-17 Marek Polacek <polacek@redhat.com>
29466 Markus Trippelsdorf <markus@trippelsdorf.de>
29468 PR tree-optimization/80079
29469 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29472 2017-03-17 Richard Biener <rguenther@suse.de>
29474 PR middle-end/80075
29475 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29476 Properly verify the LHS before the RHS possibly claims to be
29478 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29481 2017-03-17 Martin Jambor <mjambor@suse.cz>
29483 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29484 (List of -O2 options): Likewise.
29485 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29488 2017-03-17 Tom de Vries <tom@codesourcery.com>
29490 * gcov-dump.c (print_usage): Print bug_report_url.
29492 2017-03-17 Richard Biener <rguenther@suse.de>
29494 PR middle-end/80050
29495 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29496 (parser::peek): Likewise.
29498 2017-03-17 Richard Biener <rguenther@suse.de>
29500 PR tree-optimization/80048
29501 * sese.c (free_sese_info): Properly release rename_map and
29502 copied_bb_map elements.
29504 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29506 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29507 Add linked-list forward and backlinks. Insert on
29508 construction, remove on destruction.
29509 (class pass_store_merging): Add m_stores_head field.
29510 (pass_store_merging::terminate_and_process_all_chains):
29511 Iterate over m_stores_head list.
29512 (pass_store_merging::terminate_all_aliasing_chains):
29514 (pass_store_merging::execute): Check for debug stmts first.
29515 Push new chains onto the m_stores_head stack.
29517 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29520 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29521 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29522 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29524 2017-03-16 Jeff Law <law@redhat.com>
29526 PR tree-optimization/71437
29527 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29528 member function. Implementation moved into after_dom_children
29529 member function and into the threader's thread_outgoing_edges
29531 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29532 some code into new thread_outgoing_edges.
29533 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29534 definition. Simplify marker handling (do it here). Assume we always
29535 have the available expression and the const/copies tables.
29536 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29538 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29539 * tree-vrp.c (equiv_stack): No longer file scoped.
29540 (vrp_dom_walker): New class.
29541 (vrp_dom_walker::before_dom_children): New member function.
29542 (vrp_dom_walker::after_dom_children): Likewise.
29543 (identify_jump_threads): Setup domwalker. Use it rather than
29544 walking edges in a random order by hand. Simplify setup/finalization.
29545 (finalize_jump_threads): Remove.
29546 (vrp_finalize): Do not call identify_jump_threads here.
29547 (execute_vrp): Do it here instead and call thread_through_all_blocks
29550 PR tree-optimization/71437
29551 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29553 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29555 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29556 (dom_opt_dom_walker::thread_across_edge): Remove
29557 handle_dominating_asserts argument. All callers changed.
29558 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29559 changes. Remove calls to lhs_of_dominating_assert. Other
29560 uses of handle_dominating_asserts turn into unconditional code
29561 (simplify_control_stmt_condition_1): Likewise.
29562 (simplify_control_stmt_condition): Likewise.
29563 (thread_through_normal_block, thread_across_edge): Likewise.
29564 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29565 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29566 object if it is not an SSA_NAME.
29567 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29568 before calling into the VRP specific simplifiers.
29569 (identify_jump_threads): Remove handle_dominating_asserts
29572 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29575 * tree-diagnostic.c (default_tree_printer): No longer static.
29576 * tree-diagnostic.h (default_tree_printer): New prototype.
29578 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29580 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29581 Change ins into fmov.
29583 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29585 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29586 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29587 Use h_con constraint for operand 1.
29588 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29589 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29591 2017-03-15 Jeff Law <law@redhat.com>
29593 PR tree-optimization/71437
29594 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29595 (record_temporary_equivalences): Use it.
29597 PR tree-optimization/71437
29598 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29599 tree-ssa-scopedtables.
29600 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29601 (record_conditions, record_cond, vuse_eq): Likewise.
29602 (record_edge_info): Adjust to API tweak of record_conditions.
29603 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29604 (record_temporary_equivalences, optimize_stmt): Likewise.
29605 (eliminate_redundant_computations): Likewise.
29606 (record_equivalences_from_stmt): Likewise.
29607 * tree-ssa-scopedtables.c: Include options.h and params.h.
29608 (vuse_eq): New function, moved from tree-ssa-dom.c
29609 (build_and_record_new_cond): Likewise.
29610 (record_conditions): Likewise. Accept vector of conditions rather
29611 than edge_equivalence structure for first argument.
29612 for the first argument.
29613 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29614 from tree-ssa-dom.c.
29615 (avail_exprs_stack::record_cond): Likewise.
29616 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29617 from tree-ssa-dom.c.
29618 (avail_exprs_stack): Add new member functions lookup_avail_expr
29620 (record_conditions): Declare.
29622 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29625 * lra-constraints.c (process_alt_operands): Increase reject for
29626 reloading an input/output operand.
29628 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29631 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29632 insns to convert from signed/unsigned char/short to IEEE 128-bit
29634 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29636 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29639 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29640 subreg of inner mode for values already in registers.
29642 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29644 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29645 iteration reg is used after the loop.
29647 2017-03-14 Martin Sebor <msebor@redhat.com>
29649 PR tree-optimization/79800
29650 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29651 precision in negative-positive range.
29652 (format_floating): Call non-const overload with adjusted precision.
29654 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29657 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29660 2017-03-14 Martin Sebor <msebor@redhat.com>
29662 PR middle-end/80020
29663 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29664 * builtins.def (aligned_alloc): Use it.
29667 * Makefile.in (GTFILES): Add calls.c.
29668 * calls.c: Include "gt-calls.h".
29670 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29672 PR rtl-optimization/79728
29673 * regs.h (struct target_regs): New field
29674 x_contains_allocatable_regs_of_mode.
29675 (contains_allocatable_regs_of_mode): New macro.
29676 * reginfo.c (init_reg_sets_1): Initialize it, and change
29677 contains_reg_of_mode so it includes global regs as well.
29678 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29679 rather than contains_regs_of_mode.
29681 2017-03-14 Martin Liska <mliska@suse.cz>
29683 * doc/invoke.texi: Document options that can't be combined with
29684 -fcheck-pointer-bounds.
29686 2017-03-14 Martin Liska <mliska@suse.cz>
29688 PR middle-end/79831
29689 * doc/invoke.texi (-Wchkp): Document the option.
29691 2017-03-14 Martin Liska <mliska@suse.cz>
29693 * Makefile.in: Install gcov-dump.
29695 2017-03-14 Martin Liska <mliska@suse.cz>
29697 * multiple_target.c (expand_target_clones): Bail out for
29698 an invalid attribute.
29700 2017-03-14 Richard Biener <rguenther@suse.de>
29702 * alias.c (struct alias_set_entry): Pack properly.
29703 * cfgloop.h (struct loop): Likewise.
29704 * cse.c (struct set): Likewise.
29705 * ipa-utils.c (struct searchc_env): Likewise.
29706 * loop-invariant.c (struct invariant): Likewise.
29707 * lra-remat.c (struct cand): Likewise.
29708 * recog.c (struct change_t): Likewise.
29709 * rtl.h (struct address_info): Likewise.
29710 * symbol-summary.h (function_summary): Likewise.
29711 * tree-loop-distribution.c (struct partition): Likewise.
29712 * tree-object-size.c (struct object_size_info): Likewise.
29713 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29714 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29715 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29716 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29717 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29718 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29719 (struct _stmt_vec_info): Likewise.
29721 2017-03-14 Martin Liska <mliska@suse.cz>
29724 * multiple_target.c (create_dispatcher_calls): Check that
29725 a target can create a function dispatcher.
29727 2017-03-14 Martin Liska <mliska@suse.cz>
29730 * multiple_target.c (expand_target_clones): Drop local.local
29731 flag for default implementation.
29733 2017-03-14 Richard Biener <rguenther@suse.de>
29735 PR tree-optimization/80030
29736 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29738 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29740 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29741 gcc_fallthrough() instead of __attribute__((fallthrough));
29743 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29745 * doc/gcc.texi: Remove "up" link to (DIR).
29746 * doc/gccint.texi: Ditto.
29748 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29750 * doc/install.texi (Specific) <avr>: Remove reference to
29753 2017-03-13 Jeff Law <law@redhat.com>
29755 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29756 attribute rather than comments.
29758 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29759 match_scratch operand is highest.
29761 2017-03-13 Martin Liska <mliska@suse.cz>
29763 PR middle-end/78339
29764 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29765 is a CHKP clone, use original declaration.
29767 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29769 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29770 (arc_conditional_register_usage): Use a different allocation order
29771 when optimizing for size.
29772 * common/config/arc/arc-common.c (arc_option_optimization_table):
29773 Section anchors default on when optimizing for size.
29775 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29777 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29779 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29781 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29782 * config/arc/arc.md (cpu_facility): Add cd variant.
29783 (*movqi_insn): Add code density variant.
29784 (*movhi_insn): Likewise.
29785 (*movqi_insn): Likewise.
29786 (*addsi3_mixed): Likewise.
29787 (subsi3_insn): Likewise.
29789 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29791 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29793 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29795 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29796 expressions with MINUS and UNARY ops.
29798 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29801 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29803 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29804 between vec_select and vector argument.
29805 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29806 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29807 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29808 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29809 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29810 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29812 2017-03-13 Richard Biener <rguenther@suse.de>
29815 * params.def (vect-max-peeling-for-alignment): Fix typo.
29817 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29819 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29820 issue that only occurred with binutils below 2.18.
29822 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29824 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29825 refer to binutils 2.11/2.12 minimum.
29827 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29829 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29830 ftp.kernel.org and simplify binutils requirement.
29832 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29834 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29836 (Optimize Options): Ditto. Also remove redundancy.
29838 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29840 PR translation/79848
29841 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29843 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29844 to G_ to avoid double translation.
29846 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29848 PR translation/79923
29849 * auto-profile.c (get_combined_location): Convert leading
29850 character of diagnostics to lower case and remove trailing period.
29851 (read_profile): Likewise for various diagnostics.
29852 * config/arm/arm.c (arm_option_override): Remove trailing period
29853 from various diagnostics.
29854 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29855 (msp430_expand_delay_cycles): Likewise.
29857 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29860 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29861 full command-line argument, rather than just "str".
29862 (aarch64_validate_march): Likewise.
29863 (aarch64_validate_mtune): Likewise.
29865 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29867 PR rtl-optimization/78911
29868 * lra-assigns.c (must_not_spill_p): New function.
29869 (spill_for): Use it.
29871 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29873 PR tree-optimization/79981
29874 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29875 ATOMIC_COMPARE_EXCHANGE ifn result.
29876 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29877 IFN_ATOMIC_COMPARE_EXCHANGE.
29879 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29882 * opts.c (parse_sanitizer_options): Add missing question mark to
29883 "did you mean" message.
29885 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29887 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29889 (VMULEUH_UNS): Likewise.
29890 (VMULOUB_UNS): Likewise.
29891 (VMULOUH_UNS): Likewise.
29892 * config/rs6000/rs6000.c (builtin_function_type): Remove
29893 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29895 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29898 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29899 x with result of extra_parsing_for_operand_code_0.
29900 (function_reader::extra_parsing_for_operand_code_0): Convert
29901 return type from void to rtx, returning x. When reading
29902 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29903 larger size containing struct block_symbol.
29905 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29907 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29908 -mfloat128-hardware without -m64.
29910 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29913 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29914 entries to the case statement that marks unsigned arguments to
29915 overloaded functions.
29917 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29919 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29920 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29922 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29925 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29926 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29928 2017-03-10 Martin Liska <mliska@suse.cz>
29932 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29933 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29936 2017-03-10 Olivier Hainque <hainque@adacore.com>
29938 * tree-switch-conversion (array_value_type): Start by resetting
29939 candidate type to it's main variant.
29941 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29943 PR rtl-optimization/79909
29944 * combine.c (try_combine): Use simplify_replace_rtx on individual
29945 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29946 of the whole CALL_INSN_FUNCTION_USAGE.
29948 PR tree-optimization/79972
29949 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29950 get_range_info on SSA_NAMEs. Formatting fixes.
29952 2017-03-10 Richard Biener <rguenther@suse.de>
29953 Jakub Jelinek <jakub@redhat.com>
29955 PR tree-optimization/77975
29956 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29957 edge to be constant.
29958 (get_val_for): For constant x return it. Formatting fix.
29959 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29960 would use the same bases as the current one.
29962 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29964 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29965 instead of vec_select for V1TImode.
29966 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29968 (VSX_LE_128): Add V1TI to this mode iterator.
29969 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29970 (*vsx_le_perm_store_<mode>): Likewise.
29971 (pre-reload splitter for VSX stores): Likewise.
29972 (post-reload splitter for VSX stores): Likewise.
29973 (*vsx_xxpermdi2_le_<mode>): Likewise.
29974 (*vsx_lxvd2x2_le_<mode>): Likewise.
29975 (*vsx_stxvd2x2_le_<mode>): Likewise.
29977 2017-03-09 Michael Eager <eager@eagercon.com>
29979 Correct failures with --enable-checking=yes,rtl.
29981 * config/microblaze/microblaze.c (microblaze_expand_shift):
29982 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29983 test for const0_rtx.
29984 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29985 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29987 2017-03-09 Richard Biener <rguenther@suse.de>
29989 PR tree-optimization/79977
29990 * graphite-scop-detection.c (scop_detection::merge_sese):
29991 Handle the case of extra exits to blocks dominating the entry.
29993 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29995 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29998 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30000 PR rtl-optimization/79949
30001 * lra-constraints.c (process_alt_operands): Check memory when
30002 trying to predict a cycle. Print about the overall increase.
30004 2017-03-09 Richard Biener <rguenther@suse.de>
30006 PR middle-end/79971
30007 * gimple-expr.c (useless_type_conversion_p): Preserve
30008 TYPE_SATURATING for fixed-point types.
30010 2017-03-09 Richard Biener <rguenther@suse.de>
30013 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30014 alignment of BLKmode params.
30016 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30019 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30020 (VALL_NO_V2Q): Likewise.
30022 * config/aarch64/aarch64-simd.md
30023 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30025 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30026 VALL_NO_V2Q mode iterator.
30027 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30029 2017-03-09 Martin Liska <mliska@suse.cz>
30031 PR tree-optimization/79631
30032 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30033 tree_int_cst_sign_bit just for INTEGER constants.
30035 2017-03-09 Martin Liska <mliska@suse.cz>
30039 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30042 2017-03-09 Marek Polacek <polacek@redhat.com>
30045 * tree.c (inchash::add_expr): Handle TREE_VEC.
30047 2017-03-09 Martin Liska <mliska@suse.cz>
30050 (chkp_narrow_size_and_offset): New function.
30051 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30052 (void chkp_parse_bit_field_ref): New function.
30053 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30054 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30056 2017-03-09 Martin Liska <mliska@suse.cz>
30059 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30060 (chkp_find_bounds_1): Remove gcc_unreachable.
30062 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30065 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30066 BUILT_IN_SYNC*, determine the access type from the size suffix and
30067 always build a MEM_REF with that type. Handle forgotten
30068 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30071 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30072 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30073 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30074 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30075 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30076 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30077 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30078 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30079 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30080 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30081 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30082 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30083 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30084 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30085 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30086 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30087 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30088 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30089 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30090 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30091 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30092 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30093 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30094 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30095 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30096 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30097 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30098 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30099 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30100 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30101 definitions outside of __OPTIMIZE__ guarded section.
30104 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30105 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30106 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30107 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30110 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30112 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30113 ("vfenez<mode>"): Add missing constraints.
30115 2017-03-08 Martin Sebor <msebor@redhat.com>
30118 * config/nds32/nds32.c (nds32_option_override):
30119 Fix misspelled diagnostic.
30121 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30124 * gimplify.c (gimplify_omp_for): Replace index var in outer
30125 taskloop statement with an artificial variable and add
30126 OMP_CLAUSE_PRIVATE clause for it.
30128 2017-03-08 Richard Biener <rguenther@suse.de>
30130 PR tree-optimization/79955
30131 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30132 for accesses that are completely outside of the variable.
30134 2017-03-08 Andrew Haley <aph@redhat.com>
30136 PR tree-optimization/79943
30137 * tree-ssa-loop-split.c (compute_new_first_bound): When
30138 calculating the new upper bound, (END-BEG) should be added, not
30141 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30143 * config/avr/avr.md (setmemhi): Make sure match_dup
30144 operand number comes before match_scratch.
30146 2017-03-08 Richard Biener <rguenther@suse.de>
30148 PR tree-optimization/79920
30149 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30150 with ncopies == 1 to ...
30151 (vect_transform_slp_perm_load): ... here. Properly compute
30152 all element loads by iterating VF times over the group. Do
30153 not handle ncopies (computed in a broken way) in
30154 vect_create_mask_and_perm.
30156 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30159 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30160 is a uniform vector, use uniform_vector_p return value instead of
30161 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30163 2017-03-07 Marek Polacek <polacek@redhat.com>
30165 PR middle-end/79809
30166 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30167 (alloca_call_type): Likewise.
30169 2017-03-07 Martin Liska <mliska@suse.cz>
30171 * gcov.c (process_args): Put comment to correct location.
30173 2017-03-07 Martin Liska <mliska@suse.cz>
30175 PR middle-end/68270
30176 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30177 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30178 (chkp_narrow_bounds_for_field): Likewise.
30179 (chkp_parse_array_and_component_ref): Pass one more argument to
30182 2017-03-07 Richard Biener <rguenther@suse.de>
30184 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30187 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30189 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30190 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30192 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30195 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30196 to end of description.
30197 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30199 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30201 PR rtl-optimization/79901
30202 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30204 (*avx512f_<code><mode>3<mask_name>): ... this.
30205 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30206 iterator instead of VI8_AVX2_AVX512BW.
30208 PR rtl-optimization/79901
30209 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30210 min/max expander, expand it using expand_vec_cond_expr.
30213 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30216 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30219 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30220 to void * for PCH reasons.
30221 * dwarf2out.c (output_loc_operands, output_die): Cast
30222 v.val_vec.array to unsigned char *.
30224 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30227 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30230 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30232 PR rtl-optimization/79571
30233 * lra-constraints.c (process_alt_operands): Calculate static
30234 reject and subtract it from overall when only addresses will be
30237 2017-03-06 Julia Koval <julia.koval@intel.com>
30240 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30241 incoming stack boundary to 128 for 64-bit targets.
30243 2017-03-06 Richard Biener <rguenther@suse.de>
30245 PR tree-optimization/79894
30246 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30247 to NULL after folding it.
30249 2017-03-06 Richard Biener <rguenther@suse.de>
30251 PR tree-optimization/79824
30252 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30253 check disabling peeling for gaps.
30255 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30257 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30258 attributes): Document gettimeofday.
30260 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30262 * config/s390/s390.c (s390_option_override_internal): Set
30263 PARAM_MIN_VECT_LOOP_BOUND
30265 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30267 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30268 * config/s390/s390.md: Likewise.
30270 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30273 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30274 (<avx2_avx512>_perm<mode>): Rename to ...
30275 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30277 (<avx512>_perm<mode>_mask): Rename to ...
30278 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30280 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30281 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30282 instead of VI8F_256_512.
30283 (avx512f_perm<mode>): New define_expand.
30284 (avx512f_perm<mode>_mask): Likewise.
30285 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30286 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30288 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30290 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30291 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30293 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30295 2017-03-06 Martin Liska <mliska@suse.cz>
30298 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30299 when having a SSA NAME w/o VAR_DECL assigned to it.
30301 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30303 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30304 msa_dpsub_<su>_d): Fix MODE for vec_select.
30306 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30308 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30310 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30312 2017-03-06 Richard Biener <rguenther@suse.de>
30314 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30315 * plugin.c (register_plugin_info): Likewise.
30316 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30318 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30320 * config/i386/sse.md (sse_storehps, sse_storelps,
30321 avx_<castmode><avxsizesuffix>_<castmode>,
30322 avx512f_<castmode><avxsizesuffix>_<castmode>,
30323 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30324 in condition that at least one operand is not a MEM.
30326 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30328 PR middle-end/79805
30329 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30330 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30332 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30333 gimple_call_nothrow_p flag based on whether original builtin can throw.
30334 If it can, emit following stmts on the fallthrough edge.
30335 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30336 don't create new bb if inserting just debug stmts on the edge, try to
30337 insert them on the fallthru bb or just reset debug stmts.
30339 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30342 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30343 restore recog_data (including the operand rtxes inside it) around
30344 the call to get_insn_template.
30346 2017-03-03 Martin Sebor <msebor@redhat.com>
30348 PR tree-optimization/79699
30349 * context.c (context::~context): Free MPFR caches to avoid
30350 a memory leak on program exit.
30352 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30354 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30355 Use wide_int::ulow () instead of .elt (0).
30357 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30359 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30360 (*pushxf): Limit oF constraint to 32bit targets and add oC
30361 constraint for 64bit targets.
30362 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30363 (*pushdf): Change rmF constraint to rmC.
30365 2017-03-03 Martin Liska <mliska@suse.cz>
30367 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30368 Remove unused variable.
30370 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30373 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30374 is a memory operand, increase num_memory.
30375 (ix86_expand_args_builtin): Likewise.
30377 2017-03-03 Jan Hubicka <jh@suse.cz>
30380 * ipa-devirt.c (maybe_record_node): Properly handle
30381 __cxa_pure_virtual visibility.
30383 2017-03-03 Martin Liska <mliska@suse.cz>
30385 PR tree-optimization/79803
30386 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30388 (pass_loop_prefetch::execute): Disabled optimization if an
30389 assumption about L1 cache size is not met.
30391 2017-03-03 Martin Liska <mliska@suse.cz>
30393 PR rtl-optimization/79574
30394 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30395 (hash_scan_set): Likewise.
30396 (dump_hash_table): Likewise.
30397 (hoist_code): Likewise.
30399 2017-03-03 Richard Biener <rguenther@suse.de>
30401 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30402 in place of uhigh ().
30403 (fixed_convert_from_real): Likewise.
30405 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30408 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30410 2017-03-03 Richard Biener <rguenther@suse.de>
30412 PR middle-end/79818
30413 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30414 TYPE_OVERFLOW_UNDEFINED check.
30416 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30418 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30420 (vector_ae_<mode>_p): Likewise.
30421 (vector_nez_<mode>_p): Likewise.
30422 (vector_ne_v2di_p): Likewise.
30423 (vector_ae_v2di_p): Likewise.
30424 (vector_ne_<mode>_p): Likewise.
30425 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30427 (vsx_tsqrt<mode>2_fe): Likewise.
30429 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30432 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30434 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30436 PR rtl-optimization/79780
30437 * cprop.c (one_cprop_pass): When second and further conditional trap
30438 in a single basic block is turned into an unconditional trap, turn it
30439 into a deleted note to avoid RTL verification failures.
30441 2017-03-02 Richard Biener <rguenther@suse.de>
30443 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30445 2017-03-02 Richard Biener <rguenther@suse.de>
30447 PR tree-optimization/79345
30449 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30450 param and abort the walk, returning -1 if it is hit.
30451 (walk_aliased_vdefs): Take a limit param and pass it on.
30452 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30453 defaulting to 0 and return a signed int.
30454 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30455 (check_defs): New helper.
30456 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30457 about uninitialized memory.
30458 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30459 bogus uninitialized warning.
30460 (fixed_convert_from_real): Likewise.
30462 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30464 PR tree-optimization/66768
30465 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30466 iv_use if base object can't be determined.
30468 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30470 PR tree-optimization/79345
30471 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30472 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30473 (get_pattern_stats): Initialize it.
30474 * genemit.c (gen_expand): Verify match_scratch numbers come after
30475 match_operand/match_dup numbers.
30476 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30477 match_scratch numbers.
30478 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30480 * config/s390/s390.md (trunctdsd2): Likewise.
30482 2017-03-02 Richard Biener <rguenther@suse.de>
30484 * wide-int.h (wide_int_storage::operator=): Implement in terms
30487 2017-03-02 Richard Biener <rguenther@suse.de>
30489 PR tree-optimization/79777
30490 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30491 the to insert expression to sth existing.
30493 2017-03-01 Martin Sebor <msebor@redhat.com>
30495 PR middle-end/79692
30496 * gimple-ssa-sprintf.c
30497 (directive::known_width_and_precision): New function.
30498 (format_integer): Use it.
30499 (get_mpfr_format_length): Consider the full range of precision
30500 when computing %g output with the # flag. Set the likely byte
30501 count to 3 rather than 1 when precision is indeterminate.
30502 (format_floating): Correct the lower bound of precision.
30504 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30506 * doc/invoke.texi: Document default code model for 64-bit Linux.
30508 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30511 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30512 udiv rather than div since input pattern is unsigned.
30514 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30516 * config/i386/i386.c (print_reg): Warn for values of
30517 unsupported size in integer register.
30519 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30522 * config/rs6000/predicates.md (current_file_function_operand): Do
30523 not allow self calls to be local if the function is replaceable.
30525 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30528 * config/rs6000/altivec.h (vec_ctz and others): Change the
30529 preprocessor macro that controls conditional compilation from
30530 _ARCH_PWR9 to __POWER9_VECTOR__.
30531 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30532 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30533 control (instead of _ARCH_PWR9 control) so that template
30534 definition uses power9-specific function.
30535 (vec_any_eq): Likewise.
30536 (vec_all_ne): Change macro definition to use a power9-specific
30537 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30538 _ARCH_PWR9 control).
30539 (vec_any_eq) Likewise.
30540 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30541 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30542 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30543 support for xvcmpnedp instruction.
30544 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30545 macro expansion so that Power9 implementation of vec_all_ne does
30546 not use the AltiVec predicate framework.
30547 (VCMPNEH_P): Likewise.
30548 (VCMPNEW_P): Likewise.
30549 (VCMPNED_P): Likewise.
30550 (VCMPNEFP_P): Likewise.
30551 (VCMPNEDP_P): Likewise.
30552 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30553 implementation of vec_any_eq to not use AltiVec predicate
30555 (VCMPAEH_P): Likewise.
30556 (VCMPAEW_P): Likewise.
30557 (VCMPAED_P): Likewise.
30558 (VCMPAEFP_P): Likewise.
30559 (VCMPAEDP_P): Likewise.
30560 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30561 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30562 not use the AltiVec predicate framework.
30563 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30564 of vec_any_eq to not use AltiVec predicate framework.
30565 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30566 support for predefined __POWER9_VECTOR__ macro to indicate that
30567 Power9 instruction selection is enabled.
30568 (altivec_overloaded_builtins): Remove extraneous
30569 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30570 function argument types RS6000_BTI_bool_V16QI and
30571 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30572 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30573 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30574 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30575 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30576 Power9 for implementations of vec_cmpne. Change the signature for
30577 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30578 (representing vec_all_ne) to remove the previously described first
30579 argument of type RS6000_BTI_INTSI, as this was an artifact of
30580 reliance on the AltiVec predicate framework, which is no longer
30581 used in the implementation of these functions. Add
30582 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30583 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30584 since, unlike the AltiVec predicate framework implementation, we
30585 do not share function descriptors between vec_alle and vec_anyeq.
30586 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30587 set of modes that receive special treatment even when
30588 TARGET_P9_VECTOR is true. The special treatment emits code that
30589 does not depend on Power9 instructions.
30590 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30591 define_expand to not rely on AltiVec predicate framework.
30592 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30594 (vector_ne_v2di_p): Change this define_expand to not rely on
30595 AltiVec predicate framework.
30596 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30598 (vector_ne_<mode>_p): Change this define_expand to not rely on
30599 AltiVec predicate framework.
30600 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30602 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30603 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30604 define_insn pattern.
30605 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30606 define_insn pattern because the xvcmpne<VSs>. instruction is not
30608 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30609 instruction is not supported.
30611 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30613 * config/nvptx/nvptx.c: Include intl.h.
30615 2017-03-01 Martin Jambor <mjambor@suse.cz>
30618 * ipa-prop.h (ipa_bits): Removed field known.
30619 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30620 to pointers. Adjusted their comments to warn about their sharing.
30621 (ipcp_transformation_summary): Change bits to a vector of pointers.
30622 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30623 (ipa_get_ipa_bits_for_value): Declare.
30624 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30625 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30626 (ipa_bits_hash_table): Likewise.
30627 (ipa_vr_ggc_hash_traits): Likewise.
30628 (ipa_vr_hash_table): Likewise.
30629 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30630 being pointers and vr_known being removed.
30631 (ipa_set_jf_unknown): Likewise.
30632 (ipa_get_ipa_bits_for_value): New function.
30633 (ipa_set_jfunc_bits): Likewise.
30634 (ipa_get_value_range): New overloaded functions.
30635 (ipa_set_jfunc_vr): Likewise.
30636 (ipa_compute_jump_functions_for_edge): Use the above functions to
30637 construct bits and vr parts of jump functions.
30638 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30639 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30641 (ipcp_grow_transformations_if_necessary): Also allocate
30642 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30644 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30645 them. Fix too long lines.
30646 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30647 vr_known being removed.
30648 (ipa_read_jump_function): Use new setter functions to construct bits
30649 and vr parts of jump functions or set them to NULL.
30650 (write_ipcp_transformation_info): Adjust for bits being pointers.
30651 (read_ipcp_transformation_info): Likewise.
30652 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30654 Include gt-ipa-prop.h.
30655 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30657 (ipcp_store_bits_results): Likewise.
30658 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30659 Do not write to existing jump functions but use a temporary instead.
30661 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30664 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30665 attempt to use its first operand as BIT_FIELD_REF base.
30667 2017-03-01 Richard Biener <rguenther@suse.de>
30669 PR middle-end/79721
30670 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30671 interpolating formula in wrapping arithmetic.
30672 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30674 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30676 PR tree-optimization/79734
30677 * tree-vect-generic.c (expand_vector_condition): Optimize
30678 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30679 Handle VEC_COND_EXPR where comparison has different inner width from
30680 type's inner width.
30682 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30684 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30685 markup, and similar issues. Remove @opindex entries for things
30686 that aren't options. Add missing -mmpy-option entries.
30688 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30690 PR tree-optimization/79737
30691 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30692 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30693 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30694 instead of byte_size. Formatting fix.
30695 (shift_bytes_in_array_right): Formatting fix.
30697 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30700 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30701 condition on optimize for the leaf function test.
30703 2017-02-28 Martin Liska <mliska@suse.cz>
30706 * read-rtl-function.c (function_reader::handle_unknown_directive):
30707 Bail out when one uses -flto.
30709 2017-02-28 Martin Liska <mliska@suse.cz>
30711 * common.opt: Replace space with tabular for options of <number>
30713 * config/i386/i386.opt: Show <number> value for
30714 -mlarge-data-threshold.
30715 * opts.c (print_filtered_help): Do not display number in hexadecimal
30718 2017-02-28 Martin Liska <mliska@suse.cz>
30720 * common.opt: Fix --help=option -Q for options which are of
30723 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30725 * config/i386/i386.c (print_reg): Error out for values
30726 of 8-bit size in invalid integer register.
30728 2017-02-28 Martin Sebor <msebor@redhat.com>
30730 PR tree-optimization/79691
30731 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30733 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30736 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30737 gcc_unreachable with output_operand_lossage.
30739 2017-02-28 Richard Biener <rguenther@suse.de>
30741 PR tree-optimization/79740
30742 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30744 (visit_nary_op): Insert the nary into the hashtable if we
30745 pattern-matched sth.
30746 * tree-ssa-pre.c (eliminate_insert): Robustify.
30748 2017-02-28 Richard Biener <rguenther@suse.de>
30750 PR middle-end/79731
30751 * fold-const.c (decode_field_reference): Reject out-of-bound
30754 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30756 * config/i386/i386.c: Include intl.h.
30757 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30758 instead of just cond ? "..." : "...".
30759 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30760 * coverage.c (read_counts_file): Likewise.
30761 * omp-offload.c: Include intl.h.
30762 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30763 of just cond ? "..." : "...".
30764 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30765 of just cond ? "..." : "...".
30767 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30770 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30772 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30773 'tune for' CPU name.
30774 * config/arm/arm-cpu-data.h: Regenerated.
30776 2017-02-28 Richard Biener <rguenther@suse.de>
30778 PR tree-optimization/79732
30779 * tree-inline.c (expand_call_inline): Do not shadow var.
30781 2017-02-28 Richard Biener <rguenther@suse.de>
30783 PR tree-optimization/79723
30784 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30785 address-space properly.
30787 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30789 * doc/optinfo.texi (Optimization groups): Fix option used for
30791 * doc/invoke.texi (-fopt-info): Document "omp".
30792 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30793 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30794 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30795 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30796 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30798 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30800 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30801 OPTGROUP_OMP with "omp".
30803 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30806 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30807 for arithmetic shift of unsigned V2DI.
30809 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30811 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30812 arc/linux.h headers.
30813 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30814 (LINK_SPEC): Likewise.
30815 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30816 (EXTRA_SPECS): Likewise.
30817 (STARTFILE_SPEC): Likewise.
30818 (ENDFILE_SPEC): Likewise.
30819 (LIB_SPEC): Likewise.
30820 (TARGET_SDATA_DEFAULT): Likewise.
30821 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30822 (MULTILIB_DEFAULTS): Likewise.
30823 (DWARF2_UNWIND_INFO): Likewise.
30824 * config/arc/big.h: New file.
30825 * config/arc/elf.h: Likewise.
30826 * config/arc/linux.h: Likewise.
30827 * config/arc/t-uClibc: Remove.
30829 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30831 PR tree-optimization/77536
30832 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30833 (tree_transform_and_unroll_loop): Use above function to compute the
30834 estimated niter of unrolled loop and use it when scaling profile.
30835 Also use count info rather than frequency if it's non-zero.
30836 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30837 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30838 (vect_transform_loop): Call above function.
30840 2017-02-27 Richard Biener <rguenther@suse.de>
30842 PR tree-optimization/45397
30843 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30844 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30845 (visit_nary_op): Add pattern matching for CSEing sign-changed
30846 or truncated operations with wider ones.
30848 2017-02-27 Richard Biener <rguenther@suse.de>
30850 PR tree-optimization/79690
30851 * tree-vect-stmts.c (vectorizable_store): Use vector type
30852 built from the DR with address-space.
30854 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30856 * doc/invoke.texi (Optimize Options): Refine the description
30857 of asan-use-after-return.
30859 2017-02-25 Alan Modra <amodra@gmail.com>
30861 PR rtl-optimization/79584
30862 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30863 base, not ad->base_term, the reg within base. Remove assertion
30864 that ad->base == ad->base_term. Replace gen_int_mode using
30865 bogus mode with const0_rtx.
30867 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30869 PR middle-end/79396
30870 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30871 FMA_EXPR like tcc_binary or tcc_unary.
30873 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30876 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30878 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30879 (output_loc_operands): Handle DW_OP_call_ref and
30880 DW_OP_GNU_variable_value.
30881 (struct variable_value_struct): New type.
30882 (struct variable_value_hasher): Likewise.
30883 (variable_value_hash): New variable.
30884 (string_types): Remove.
30885 (copy_loc_descr): New function.
30886 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30887 (prepend_loc_descr_to_each): New function.
30888 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30889 instead of add_loc_descr_to_each if the first argument is single
30890 location list and the second has multiple.
30891 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30892 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30893 when looking for variable value which doesn't have other location info.
30894 (loc_list_from_tree): Formatting fix.
30895 (gen_array_type_die): Simplify DW_AT_string_length handling.
30896 (adjust_string_types): Remove.
30897 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30898 string_types. Call resolve_variable_values.
30899 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30900 (resolve_addr_in_expr): Likewise. Add A argument.
30901 (copy_deref_exprloc): Remove deref argument. Adjust for the
30902 original expression being DW_OP_GNU_variable_value with optionally
30903 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30904 optionally after it.
30905 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30906 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30907 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30908 (variable_value_hasher::hash, variable_value_hasher::equal): New
30910 (resolve_variable_value_in_expr, resolve_variable_value,
30911 resolve_variable_values, note_variable_value_in_expr,
30912 note_variable_value): New functions.
30913 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30916 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30919 * opts.h (handle_generated_option): Add GENERATED_P argument.
30920 * opts-common.c (handle_option): Adjust function comment.
30921 (handle_generated_option): Add GENERATED_P argument, pass it to
30923 (control_warning_option): Pass false to handle_generated_option
30925 * opts.c (maybe_default_option): Pass true to handle_generated_option
30927 * optc-gen.awk: Likewise.
30929 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30931 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30932 a REG, look at the REG it is a SUBREG of.
30933 (splitter for cmpeqsi_t): Ditto.
30935 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30937 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30938 the special USEs with the pattern of the insn, not the insn itself.
30940 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30943 * doc/invoke.texi: Document -mload-store-pairs.
30945 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30946 Sandra Loosemore <sandra@codesourcery.com>
30948 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30949 argument isn't a CONST_INT.
30950 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30951 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30952 (nios2_validate_compare): Bypass alternate compare logic if *op2
30953 is not a CONST_INT.
30954 (ldstwm_operation_p): Return false if first_base is not a REG or
30955 if first_offset is not a CONST_INT.
30957 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30959 * config/cris/cris.md: Use correct operand in a define_peephole2.
30961 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30963 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30965 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30967 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30968 this_insn if it is an INSN or JUMP_INSN.
30969 (force_offsettable): Look at base, not at addr.
30970 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30971 on things that aren't necessarily CONST_INTs.
30973 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30975 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30976 -mfpmath=sse is the default also for x86-32 targets with SSE2
30977 instruction set when @option{-ffast-math} is enabled
30979 2017-02-24 Jeff Law <law@redhat.com>
30981 PR rtl-optimizatoin/79286
30982 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30985 2017-02-24 Richard Biener <rguenther@suse.de>
30987 PR tree-optimization/79389
30988 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30991 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30993 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30994 function comment to reflect reality.
30995 (loop_exits_before_overflow): Fix typo in function description.
30997 2017-02-24 Richard Biener <rguenther@suse.de>
30999 PR tree-optimization/79389
31000 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31001 properly that a threading opportunity exists. Detect conditional
31002 copy/constant propagation opportunities.
31004 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31006 * config/visium/visium.md (type): Add trap.
31007 (b): New mode attribute.
31008 (*btst): Rename into...
31009 (*btst<mode>): ...this and adjust.
31010 (*cbranchsi4_btst_insn): Rename into...
31011 (*cbranch<mode>4_btst_insn): ...this and adjust.
31012 (trap): New define_insn.
31014 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31016 PR tree-optimization/79389
31017 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31018 (noce_reversed_cond_code): New function.
31019 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31020 reversed_comparison_code. Formatting fix.
31021 (noce_try_store_flag): Test rev_cond != NULL in addition to
31022 reversed_comparison_code.
31023 (noce_try_store_flag_constants): Likewise.
31024 (noce_try_store_flag_mask): Likewise.
31025 (noce_try_addcc): Use rev_cond if non-NULL instead of
31026 reversed_comparison_code.
31027 (noce_try_cmove_arith): Likewise. Formatting fixes.
31028 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31029 (noce_find_if_block): Initialize rev_cond.
31030 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31031 instead of false as last argument never attempt to reverse it
31034 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31036 PR tree-optimization/79663
31037 * tree-predcom.c (combine_chains): Process refs in reverse order
31038 only for ZERO length chains, and add explaining comment.
31040 2017-02-23 Jeff Law <law@redhat.com>
31042 PR tree-optimization/79578
31043 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31044 in call to operand_equal_p.
31046 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31049 * config/i386/cpuid.h: Fix another undefined behavior.
31051 2017-02-23 Richard Biener <rguenther@suse.de>
31053 PR tree-optimization/79683
31054 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31055 vector types for data-refs.
31057 2017-02-23 Martin Liska <mliska@suse.cz>
31059 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31061 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31063 PR middle-end/79665
31064 * internal-fn.c (get_range_pos_neg): Moved to ...
31065 * tree.c (get_range_pos_neg): ... here. No longer static.
31066 * tree.h (get_range_pos_neg): New prototype.
31067 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31068 are known to be in between 0 and signed maximum inclusive, try to
31069 expand both unsigned and signed divmod and use the cheaper one from
31072 2017-02-22 Jeff Law <law@redhat.com>
31074 PR tree-optimization/79578
31075 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31076 to compare base operands.
31078 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31081 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31082 gpc_reg_operand instead of fpr_reg_operand.
31084 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31086 * config/mips/mips.c (mips_return_in_memory): Force FP
31087 vector types to be returned in memory for o32 ABI.
31089 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31091 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31092 instead of DW_TAG_member for static data member declarations and don't
31093 set no_linkage_name for static inline data members.
31094 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31097 2017-02-22 Martin Liska <mliska@suse.cz>
31099 * doc/invoke.texi: Replace inequality signs with square brackets
31102 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31105 * lra-constraints.c (simplify_operand_subreg): Handle
31106 WORD_REGISTER_OPERATIONS targets.
31108 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31111 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31112 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31113 elimination by swapping fld*.
31115 2017-02-22 Richard Biener <rguenther@suse.de>
31117 PR tree-optimization/79673
31118 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31119 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31120 irrelevant address-space qualifiers and avoiding a
31121 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31123 2017-02-22 Richard Biener <rguenther@suse.de>
31125 PR tree-optimization/79666
31126 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31127 to not symbolically negate if that may introduce undefined
31130 2017-02-22 Martin Liska <mliska@suse.cz>
31133 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31134 * data-streamer-out.c (streamer_write_gcov_count_stream):
31136 * value-prof.c (stream_out_histogram_value): Make assert more
31137 precise based on type of counter.
31139 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31142 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31143 Use nonimmediate_operand instead of memory_operand for operand 1.
31144 (float-extend standard_x87sse_constant_load splitter): Ditto.
31146 2017-02-21 Jeff Law <law@redhat.com>
31148 PR tree-optimization/79621
31149 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31150 blocks with edges to themselves.
31152 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31155 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31156 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31157 Use gimple_call_builtin_p.
31160 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31161 on temporarily removed DEBUG_INSNs.
31163 PR tree-optimization/79649
31164 * tree-loop-distribution.c (classify_partition): Give up on
31165 non-generic address space loads/stores.
31167 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31169 * doc/loop.texi (Loop manipulation): Remove nonexistent
31170 tree_ssa_loop_version from the documentation.
31171 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31173 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31176 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31177 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31178 * config/rs6000/rs6000.c: Include except.h.
31179 (rs6000_expand_split_stack_prologue): Call
31180 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31182 2017-02-21 Martin Jambor <mjambor@suse.cz>
31185 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31186 have been analyzed.
31188 2017-02-21 Martin Jambor <mjambor@suse.cz>
31190 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31191 for backward compatibility only.
31192 * doc/invoke.texi (Option Summary): Remove all references to
31193 -fipa-cp-alignment.
31195 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31199 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31201 * lra-constraints.c (curr_insn_transform): Handle
31202 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31204 2017-02-21 Martin Liska <mliska@suse.cz>
31206 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31208 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31210 PR translation/79638
31211 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31213 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31216 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31217 (arm_function_ok_for_sibcall): Return false for an indirect call by
31218 descriptor if all the argument registers are used.
31219 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31220 alignment of the function.
31222 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31224 PR tree-optimization/61441
31225 * simplify-rtx.c (simplify_const_unary_operation): For
31226 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31227 the sNaN unmodified.
31229 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31231 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31232 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31233 instead of SYSTEM_HEADER_DIR.
31235 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31236 Martin Liška <mliska@suse.cz>
31238 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31239 Fix typos and grammar, use active voice, and clarify.
31241 2017-02-20 Marek Polacek <polacek@redhat.com>
31243 PR middle-end/79537
31244 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31247 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31249 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31252 * config/i386/i386.c (ix86_expand_builtin): Handle
31253 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31254 ix86_builtins_isa[fcode].isa as a requirement of those
31255 flags and any other flag in the bitmask.
31256 (ix86_init_mmx_sse_builtins): Use 0 instead of
31257 ~OPTION_MASK_ISA_64BIT as mask.
31258 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31259 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31260 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31261 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31263 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31266 * lra-constraints.c (split_reg): Check requested split mode
31267 is supported by the register.
31269 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31271 * lra-constraints.c (simplify_operand_subreg): Remove early
31274 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31277 * lra-constraints.c (curr_insn_transform): Tighten condition
31278 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31280 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31283 * lra-constraints.c (curr_insn_transform): Handle
31284 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31286 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31289 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31291 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31293 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31296 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31299 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31301 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31304 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31306 * final.c (last_columnnum, override_columnnum): New variables.
31307 (final_start_function): Set last_columnnum, pass it to begin_prologue
31308 hook and pass 0 to dwarf2out_begin_prologue.
31309 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31310 to source_line debug hook.
31311 (notice_source_line): Compute last_columnnum and for debug_column_info
31312 return true on column changes.
31313 * debug.h (struct gcc_debug_hooks): Add column argument to
31314 source_line and begin_prologue hooks.
31315 (debug_nothing_int_charstar_int_bool): Remove prototype.
31316 (debug_nothing_int_int_charstar,
31317 debug_nothing_int_int_charstar_int_bool): New prototypes.
31318 (dwarf2out_begin_prologue): Add column argument.
31319 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31320 begin_prologue hooks.
31321 (debug_nothing_int_charstar_int_bool): Remove.
31322 (debug_nothing_int_int_charstar,
31323 debug_nothing_int_int_charstar_int_bool): New functions.
31324 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31325 through to dwarf2out_source_line.
31326 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31327 (dwarf2out_source_line): Add column argument, emit it if requested.
31328 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31330 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31331 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31332 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31333 through to dwarf2out_begin_prologue.
31334 (vmsdbgout_source_line): Add column argument, pass it through to
31335 dwarf2out_source_line.
31336 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31337 dbxout_source_line caller.
31338 (dbxout_source_line): Add column argument.
31340 * common.opt (gno-column-info, gcolumn-info): New options.
31341 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31342 (check_die): Also test for multiple DW_AT_decl_column attributes.
31343 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31344 DW_AT_decl_column if requested.
31345 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31347 (gen_variable_die): Likewise.
31348 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31349 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31352 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31353 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31354 (ix86_handle_option): Handle OPT_m3dnowa.
31355 * doc/invoke.texi (-m3dnowa): Document.
31356 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31357 -m3dnowa instead of -m3dnow -march=athlon.
31360 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31361 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31363 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31366 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31367 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31368 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31369 generator for vsx_xxpermdi_<mode>_be.
31370 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31371 force big-endian semantics.
31372 (vsx_xxpermdi_<mode>_be): New define_expand with same
31373 implementation as previous version of vsx_xxpermdi_<mode>.
31375 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31377 PR tree-optimization/79327
31378 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31379 variable, its initialization and use.
31381 2017-02-17 Julia Koval <julia.koval@intel.com>
31383 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31384 (OPTION_MASK_ISA_PKU_UNSET): New.
31385 (ix86_handle_option): Handle -mrdpid.
31386 * config/i386/cpuid.h (bit_RDPID): New.
31387 * config/i386/driver-i386.c (host_detect_local_cpu):
31388 Detect RDPID feature.
31389 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31390 * config/i386/i386-c.c (ix86_target_macros_internal):
31392 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31393 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31394 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31395 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31396 * config/i386/i386.md (define_insn "rdpid"): New.
31397 * config/i386/i386.opt Add -mrdpid.
31398 * config/i386/immintrin.h (_rdpid_u32): New.
31400 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31402 PR rtl-optimization/79541
31403 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31404 instead of transforming it into USE.
31406 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31408 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31409 If HONOR_SNANS (SFmode) force the input to a register.
31410 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31411 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31412 an frsp or similar insn.
31414 2017-02-17 Martin Liska <mliska@suse.cz>
31416 PR rtl-optimization/79577
31417 * params.def (selsched-max-sched-times): Increase minimum to 1.
31419 2017-02-17 Martin Liska <mliska@suse.cz>
31421 PR rtl-optimization/79574
31422 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31424 2017-02-17 Martin Liska <mliska@suse.cz>
31426 PR tree-optimization/79529
31427 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31428 ssa_defined_default_def_p to handle cases which are implicitly
31430 * tree-ssa.c (ssa_defined_default_def_p): New function.
31431 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31432 which are implicitly defined.
31433 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31435 2017-02-17 Richard Biener <rguenther@suse.de>
31437 PR middle-end/79576
31438 * params.def (max-ssa-name-query-depth): Limit to 10.
31440 2017-02-17 Richard Biener <rguenther@suse.de>
31442 PR tree-optimization/79552
31443 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31446 2017-02-17 Richard Biener <rguenther@suse.de>
31449 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31451 2017-02-17 Marek Polacek <polacek@redhat.com>
31453 PR middle-end/79536
31454 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31455 (fold_negate_expr): New wrapper.
31457 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31459 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31460 Correct terminology and de-emphasize pre-standard behavior.
31462 2017-02-16 Alan Modra <amodra@gmail.com>
31464 PR rtl-optimization/79286
31465 * ira.c (def_dominates_uses): New function.
31466 (update_equiv_regs): Don't create an equivalence for insns that
31467 may trap where the register def does not dominate the use.
31469 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31471 PR rtl-optimization/78127
31472 * lra.c (lra): Call lra_eliminate before finish the loop after
31475 2017-02-16 Richard Biener <rguenther@suse.de>
31477 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31479 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31480 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31481 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31482 (isl_val_int_from_wi): New function.
31483 (extract_affine_gmp): Rename to ...
31484 (extract_affine_wi): ... this, take a widest_int.
31485 (extract_affine_int): Just wrap extract_affine_wi.
31486 (add_param_constraints): Use isl_val_int_from_wi.
31487 (add_loop_constraints): Likewise, and extract_affine_wi.
31489 2017-02-15 Jeff Law <law@redhat.com>
31491 PR middle-end/79521
31492 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31493 ira_init_register_move_cost_if_necessary.
31495 2017-02-15 Martin Sebor <msebor@redhat.com>
31497 PR middle-end/32003
31498 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31499 removed in a prior commit.
31501 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31503 PR tree-optimization/79347
31504 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31505 counters during peeling.
31507 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31509 * Makefile.in (site.exp): Remove "set ISLVER".
31511 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31514 * real.c (real_from_integer): Call real_convert even for decimal.
31516 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31519 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31521 2017-02-14 Andrew Pinski <apinski@cavium.com>
31523 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31524 cores and change the partno/implementer to be correct.
31525 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31526 the 'B" as the implementer.
31527 * config/aarch64/aarch64-tune.md: Regenerate.
31529 2017-02-14 Carl Love <cel@us.ibm.com>
31531 * config/rs6000/rs6000.c: Add case statement entry to make the
31532 xvcvuxdsp built-in argument unsigned.
31533 * config/rs6000/vsx.md: Fix the source and return operand types so they
31534 match the instruction definitions from the ISA document. Fix typo
31535 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31538 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31541 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31542 member early_clobber_alts.
31543 * lra-lives.c (reg_early_clobber_p): New.
31544 (process_bb_lives): Use it.
31545 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31546 (debug_operand_data): Initialize early_clobber_alts.
31547 (setup_operand_alternative): Set up early_clobber_alts.
31548 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31549 alternatives to new_insn_reg.
31550 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31552 (lra_update_insn_regno_info): Pass the new arg.
31554 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31556 PR middle-end/79505
31557 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31558 (new_oacc_loop_raw): Don't clear already cleared fields.
31561 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31562 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31563 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31565 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31568 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31570 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31573 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31574 the extra instruction to the right place to store 128-bit constant
31577 2017-02-14 Martin Sebor <msebor@redhat.com>
31579 PR middle-end/79448
31580 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31581 warning for strings of unknown length.
31583 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31585 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31587 2017-02-14 Jeff Law <law@redhat.com>
31590 * ira-costs.c (scan_one_insn): Initialize register move costs
31591 for pseudos seen in USE/CLOBBER insns.
31593 PR tree-optimization/79095
31594 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31595 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31596 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31597 if the operands are known to be not equal, then the resulting range
31599 (intersect_ranges): If the new range is ~[0,0] and the old range is
31600 wide, then prefer ~[0,0].
31601 * tree-vrp.c (overflow_comparison_p_1): New function.
31602 (overflow_comparison_p): New function.
31603 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31604 if NAME is used in an overflow test.
31605 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31606 overflow check that can be expressed as an equality test, then adjust
31607 ops to be that equality test.
31609 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31611 * config/s390/s390-builtin-types.def: Remove flags argument.
31612 * config/s390/s390.c (s390_init_builtins): Likewise.
31614 2017-02-14 Martin Liska <mliska@suse.cz>
31616 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31617 vector. Fix trailing white spaces.
31619 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31621 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31624 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31626 PR rtl-optimization/68664
31627 * config/arm/arm.c (arm_sched_can_speculate_insn):
31628 New function. Declare prototype.
31629 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31631 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31633 PR rtl-optimization/68664
31634 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31636 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31638 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31640 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31641 max skip bytes for function, loop and jump.
31643 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31645 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31646 ABS_EXPR for gimple dump.
31648 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31651 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31653 PR tree-optimization/79408
31654 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31655 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31656 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31657 also if rhs1 is INTEGER_CST.
31659 2017-02-14 Richard Biener <rguenther@suse.de>
31661 PR middle-end/79432
31662 * tree-into-ssa.c (insert_phi_nodes): When the function can
31663 have abnormal edges rewrite SSA names with broken use-def
31664 dominance out of SSA and register them for PHI insertion.
31666 2017-02-13 Martin Sebor <msebor@redhat.com>
31668 PR middle-end/79496
31669 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31670 clearing info.nowrite flag when snprintf size argument is a range.
31672 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31674 * cprop.c (cprop_jump): Add missing space in string literal.
31675 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31676 (get_constraint_for_component_ref): Likewise.
31677 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31678 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31679 * lra-constraints.c (process_alt_operands): Likewise.
31680 * ipa-inline.c (inline_small_functions): Likewise.
31681 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31682 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31683 * trans-mem.c (diagnose_tm_1_op): Likewise.
31684 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31685 (grid_parallel_clauses_gridifiable): Likewise.
31687 * config/nvptx/mkoffload.c (process): Add space in between
31690 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31691 "MOD4_SSE_REGS" and "ALL_REGS".
31693 * spellcheck.c (test_data): Add , in between "foo" and "food".
31695 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31698 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31699 boundary crossing check and subsequent code generation agree.
31701 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31703 * config/aarch64/aarch64.c (has_memory_op): Delete.
31704 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31707 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31709 PR rtl-optimization/79388
31710 PR rtl-optimization/79450
31711 * combine.c (distribute_notes): When removing TEM_INSN for which
31712 corresponding dest has last value recorded, invalidate that last
31715 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31717 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31718 of explicit '@'. Add missing assembly comment marker on branch costs
31721 2017-02-13 Nathan Sidwell <nathan@acm.org>
31723 * gengtype-lex.l (<in_struct>): Add '/'.
31725 2017-02-13 Martin Liska <mliska@suse.cz>
31728 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31730 2017-02-13 Richard Biener <rguenther@suse.de>
31732 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31734 * configure: Re-generate.
31735 * config.in: Likewise.
31736 * graphite-dependences.c: Simplify as if
31737 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31738 * graphite-isl-ast-to-gimple.c: Likewise.
31739 * graphite-optimize-isl.c: Likewise.
31740 * graphite-poly.c: Likewise.
31741 * graphite-sese-to-poly.c: Likewise.
31742 * graphite.h: Likewise.
31743 * toplev.c: Include isl/version.h and use isl_version () for
31744 printing the ISL version.
31745 * doc/install.texi: Update ISL requirement.
31747 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31749 * doc/standards.texi (Standards): Update reference to
31752 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31754 * doc/extend.texi (Named Address Spaces): sourceware.org now
31756 * doc/install.texi (Binaries): Ditto.
31759 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31761 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31762 standard terminology "stringize"/"stringizing" throughout.
31763 * doc/cppinternals.texi: Likewise.
31765 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31767 * doc/extend.texi: Fix some spelling mistakes and typos.
31768 * doc/invoke.texi: Likewise.
31770 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31773 * params.def (inline-min-speedup) Change from 10 to 8.
31775 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31777 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31780 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31783 * ipa-inline-analysis.c (get_minimal_bb): New function.
31784 (record_modified): Use it.
31785 (remap_edge_change_prob): Handle also ancestor functions.
31787 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31789 * doc/contrib.texi (Contributors): Remove broken link into
31790 the Mauve CVS repository.
31792 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31794 PR middle-end/79454
31795 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31796 result computation whenever lhs doesn't have vector mode, not
31797 just when it has BLKmode.
31799 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31801 * doc/makefile.texi (profiledbootstrap): Refer to the
31802 installation instructions only in textual form.
31804 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31807 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31809 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31811 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31812 (Specific): Update mingw-w64 reference.
31814 (Specific): Remove broken link to Renesas RX processor.
31816 2017-02-10 Richard Biener <rguenther@suse.de>
31818 * toplev.c (process_options): Do not mention obsolete graphite
31819 options when printing sorry message about missing graphite support.
31820 Mention -floop-nest-optimize.
31822 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31824 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31825 (vtst_p16): Likewise.
31826 (vtstq_p8): Likewise.
31827 (vtstq_p16): Likewise.
31829 (vtstq_p64): Likewise.
31830 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31831 (vset_lane_p64): New.
31832 (vsetq_lane_p64): New.
31834 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31836 PR tree-optimization/79411
31837 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31838 stmt operands are SSA_NAMEs used in abnormal phis.
31839 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31842 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31845 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31848 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31850 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31852 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31854 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31858 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31859 not arbitrary TREE_CONSTANT.
31862 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31863 "omp declare target link" attribute unless is_global_var.
31864 * omp-offload.c (find_link_var_op): Likewise.
31866 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31867 Chung-Lin Tang <cltang@codesourcery.com>
31869 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31871 (gimplify_adjust_omp_clauses): Don't delete TILE.
31872 (gimplify_omp_for): Deal with TILE.
31873 * internal-fn.c (expand_GOACC_TILE): New function.
31874 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31876 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31877 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31879 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31880 avoid DIV for outermost collapse var.
31881 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31882 Remove out of date comments, fix whitespace.
31883 * omp-general.c (omp_extract_for_data): Deal with tiling.
31884 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31885 adjust OLF_DIM_BASE value.
31886 (struct omp_for_data): Add tiling field.
31887 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31888 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31889 for auto loops. Remove default auto determining, moved to
31890 oacc_loop_fixed_partitions.
31891 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31892 stmts, add e_mask field.
31893 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31894 (oacc_thread_numbers): Use oacc_dim_call.
31895 (oacc_xform_tile): New.
31896 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31897 (finish_oacc_loop): Adjust for ifns vector.
31898 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31899 add case for GOACC_TILE fns.
31900 (oacc_loop_xform_loop): Delete.
31901 (oacc_loop_process): Iterate over call list directly, and add
31902 handling for GOACC_TILE fns.
31903 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31905 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31906 vector partitioning to outer loops. Assign 2 partitions to loops
31907 when available. Add TILE handling.
31908 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31909 (execite_oacc_device_lower): Process GOACC_TILE fns,
31910 ignore unknown specs.
31911 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31912 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31913 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31915 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31917 * configure.ac (ACX_BUGURL): Update.
31918 * configure: Regenerate.
31920 2017-02-09 Richard Biener <rguenther@suse.de>
31922 PR tree-optimization/69823
31923 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31924 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31926 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31928 * config/arc/arc-c.def: Add __NPS400__ definition.
31929 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31930 (TARGET_NPS400): Define.
31932 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31934 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31936 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31937 pointer, arch_info.
31938 (arc_cpu_types): Fill the arch_info field with a pointer into the
31939 arc_arch_types table.
31940 (arc_selected_cpu): Declare.
31941 * config/arc/arc.c (arc_selected_cpu): Make global.
31942 (arc_selected_arch): Delete.
31943 (arc_base_cpu): Delete.
31944 (arc_override_options): Remove references to deleted variables,
31945 update access to arch information.
31946 (ARC_OPT): Update access to arch information.
31947 (ARC_OPTX): Likewise.
31948 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31949 (TARGET_ARC600): Update access to arch information.
31950 (TARGET_ARC601): Likewise.
31951 (TARGET_ARC700): Likewise.
31952 (TARGET_EM): Likewise.
31953 (TARGET_HS): Likewise.
31954 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31957 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31960 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31961 condition/operands for integer GE/LE/GEU/LEU operations.
31963 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31965 PR translation/79397
31966 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31969 2017-02-08 Martin Jambor <mjambor@suse.cz>
31972 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31973 whether allocation happened.
31974 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31975 nothing was allocated.
31977 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31979 PR tree-optimization/79408
31980 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31981 constant, but SSA_NAME with a known integer range, use the minimum
31982 of that range instead of op1 to determine if modulo can be replaced
31983 with its first operand.
31985 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31987 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31989 2017-02-08 Richard Biener <rguenther@suse.de>
31991 PR tree-optimization/71824
31992 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31993 Check all loops contained in the merged region.
31995 2017-02-07 Andrew Pinski <apinski@cavium.com>
31997 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31999 2017-02-07 Andrew Pinski <apinski@cavium.com>
32001 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32002 (thunderxt88): Likewise.
32003 (thunderxt81): Disable LSE and change v8.1 to v8.
32004 (thunderxt83): Likewise.
32006 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32007 Richard Biener <rguenther@suse.de>
32009 PR middle-end/79399
32010 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32011 type from int to size_t.
32012 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32014 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32016 PR rtl-optimization/79386
32017 * cprop.c (bypass_conditional_jumps): Initialize
32018 bypass_last_basic_block already before splitting bbs after
32019 unconditional traps...
32020 (bypass_conditional_jumps): ... rather than here.
32023 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32024 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32025 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32026 fix -masm=intel patterns.
32028 2017-02-07 Richard Biener <rguenther@suse.de>
32030 PR tree-optimization/79256
32031 PR middle-end/79278
32032 * builtins.c (get_object_alignment_2): Use min_align_of_type
32033 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32034 and ADJUST_FIELD_ALIGN.
32036 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32038 * doc/tm.texi: Regenerate.
32039 * stor-layout.c (layout_decl): Adjust.
32040 (update_alignment_for_field): Likewise.
32041 (place_field): Likewise.
32042 (min_align_of_type): Likewise.
32043 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32044 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32045 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32046 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32047 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32048 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32049 * config/i386/i386.c (x86_field_alignment): Likewise.
32050 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32051 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32052 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32053 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32054 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32055 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32059 2017-01-30 Richard Biener <rguenther@suse.de>
32061 PR tree-optimization/79256
32062 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32063 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32066 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32068 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32069 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32070 builtins to SImode and emit a zero-extend, if necessary.
32072 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32074 * docs/invoke.texi (RISC-V Options): Alphabetize.
32076 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32078 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32081 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32083 * config/riscv/riscv.c: New file.
32084 * common/config/riscv/riscv-common.c: Likewise.
32085 * config.gcc: Likewise.
32086 * config/riscv/constraints.md: Likewise.
32087 * config/riscv/elf.h: Likewise.
32088 * config/riscv/generic.md: Likewise.
32089 * config/riscv/linux.h: Likewise.
32090 * config/riscv/multilib-generator: Likewise.
32091 * config/riscv/peephole.md: Likewise.
32092 * config/riscv/pic.md: Likewise.
32093 * config/riscv/predicates.md: Likewise.
32094 * config/riscv/riscv-builtins.c: Likewise.
32095 * config/riscv/riscv-c.c: Likewise.
32096 * config/riscv/riscv-ftypes.def: Likewise.
32097 * config/riscv/riscv-modes.def: Likewise.
32098 * config/riscv/riscv-opts.h: Likewise.
32099 * config/riscv/riscv-protos.h: Likewise.
32100 * config/riscv/riscv.h: Likewise.
32101 * config/riscv/riscv.md: Likewise.
32102 * config/riscv/riscv.opt: Likewise.
32103 * config/riscv/sync.md: Likewise.
32104 * config/riscv/t-elf-multilib: Likewise.
32105 * config/riscv/t-linux: Likewise.
32106 * config/riscv/t-linux-multilib: Likewise.
32107 * config/riscv/t-riscv: Likewise.
32108 * configure.ac: Likewise.
32109 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32110 Waterman as RISC-V maintainers.
32111 * doc/install.texi: Add RISC-V entries.
32112 * doc/invoke.texi: Add RISC-V options section.
32113 * doc/md.texi: Add RISC-V constraints section.
32114 * configure: Regenerated.
32116 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32119 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32120 false values to be constant vectors with all 0 or all 1 bits set.
32121 (vcondu<mode><mode>): Likewise.
32122 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32124 (fpmask_comparison_operator): Update comment.
32125 (vecint_comparison_operator): New predicate.
32126 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32127 vector conditionals when the true and false values are constant
32128 vectors with all 0 bits or all 1 bits set.
32130 2017-02-06 Martin Sebor <msebor@redhat.com>
32132 PR tree-optimization/79376
32133 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32135 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32137 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32138 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32139 to simplify split condition.
32141 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32143 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32144 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32147 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32149 PR rtl-optimization/68664
32150 * target.def (can_speculate_insn): New hook.
32151 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32152 * doc/tm.texi: Regenerate.
32153 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32154 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32155 (rs6000_sched_can_speculate_insn): New function.
32157 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32159 PR tree-optimization/79284
32160 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32161 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32162 vectorizable_mask_load_store, vectorizable_operation,
32163 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32164 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32165 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32166 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32167 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32168 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32169 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32170 is_gimple_assign (stmt). Replace another such test with
32171 is_gimple_assign (stmt).
32173 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32176 * config/avr/avr.c (rtl-iter.h): Include it.
32177 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32178 (avr_legitimate_combined_insn): ...and implementation.
32180 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32182 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32183 * config/s390/s390.c (s390_const_operand_ok)
32184 (s390_canonicalize_comparison, s390_extract_part)
32185 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32186 (s390_contiguous_bitmask_p, s390_rtx_costs)
32187 (legitimize_pic_address): Likewise.
32188 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32189 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32190 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32191 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32192 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32194 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32196 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32197 REGNO($0) == REGNO($1).
32199 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32201 * config/s390/linux.h(SIZE_TYPE): Add comment.
32203 2017-02-06 Julian Brown <julian@codesourcery.com>
32204 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32205 Virendra Pathak <virendra.pathak@broadcom.com>
32207 * config/aarch64/aarch64-cores.def: Change the scheduler
32209 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32210 * config/aarch64/thunderx2t99.md: New file.
32212 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32214 * doc/standards.texi (Go Language): Update link to language
32217 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32219 * tree-eh.c (lower_resx): Sanitize profile.
32220 (cleanup_empty_eh_move_lp): Likewise.
32222 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32225 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32227 * cfgloopmanip.h (loop_version): Update prototype.
32228 * modulo-sched.c (sms_schedule): Update call of loop_version.
32229 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32230 * tree-parloops.c (gen_parallel_loop): Likewise.
32231 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32232 * tree-ssa-loop-split.c (split_loop): Likewise.
32233 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32234 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32236 2017-02-05 Martin Liska <mliska@suse.cz>
32239 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32241 (print_operand_address): Initialize a struct to zero.
32243 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32245 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32246 garbage collector only in textual form.
32248 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32250 * doc/extend.texi (x86 specific memory model extensions for
32251 transactional memory): Simplify a phrase.
32253 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32256 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32257 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32258 (atomic_storedi_1): Likewise.
32260 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32262 PR tree-optimization/79338
32263 * tree-parloops.c (gather_scalar_reductions): Don't call
32264 vect_analyze_loop_form for loop->inner before destroying loop's
32267 2017-02-03 Martin Sebor <msebor@redhat.com>
32269 PR tree-optimization/79327
32270 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32271 when precision has resulted in leading zeros.
32272 (format_integer): Adjust the likely counter to assume an unknown
32273 argument that may be zero is non-zero.
32275 2017-02-03 Jason Merrill <jason@redhat.com>
32278 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32279 avoid copying non-taken branch.
32281 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32283 PR tree-optimization/79340
32284 * tree-vect-loop.c (vectorizable_reduction): Release
32285 vec_defs elements after safe_splicing them into other vectors.
32288 PR tree-optimization/79327
32289 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32290 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32292 (format_integer): Use wide_int_to_tree instead of build_int_cst
32293 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32294 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32295 of shortest and longest sequence.
32297 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32299 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32300 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32302 2017-02-03 Walter Lee <walt@tilera.com>
32305 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32306 after initial stackframe link reg save.
32307 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32309 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32312 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32313 wu for stxssp alternative.
32315 2017-02-03 Martin Sebor <msebor@redhat.com>
32317 PR tree-optimization/79352
32318 * gimple-fold.c (get_range_strlen): Add argument.
32319 (get_range_strlen): Change return type to bool.
32320 (get_maxval_strlen): Pass in a dummy argument.
32321 * gimple-fold.h (get_range_strlen): Change return type to bool.
32322 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32323 * tree.h (array_at_struct_end_p): Add argument.
32324 * tree.c (array_at_struct_end_p): Handle it.
32326 2017-02-03 Martin Liska <mliska@suse.cz>
32329 * multiple_target.c (create_dispatcher_calls): Redirect edge
32330 from a caller of a dispatcher.
32331 (expand_target_clones): Make the clones local.
32332 (ipa_target_clone): Do both target clones and resolvers.
32333 (ipa_dispatcher_calls): Remove the pass.
32334 (pass_dispatcher_calls::gate): Likewise.
32335 (make_pass_dispatcher_calls): Likewise.
32336 * passes.def (pass_target_clone): Put as very first IPA early
32339 2017-02-03 Martin Liska <mliska@suse.cz>
32341 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32342 in case of a function with ifunc attribute.
32344 2017-02-03 Martin Liska <mliska@suse.cz>
32346 * cgraph.c (cgraph_node::dump): Dump function version info.
32347 * symtab.c (symtab_node::dump_base): Add missing new line.
32349 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32351 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32352 (ifcombine_ifandif): Use it.
32354 2017-02-03 Martin Liska <mliska@suse.cz>
32356 * doc/invoke.texi: Document default value for
32357 use-after-scope-direct-emission-threshold.
32359 2017-02-03 Martin Liska <mliska@suse.cz>
32361 PR tree-optimization/79339
32362 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32363 (format_floating): Likewise.
32365 2017-02-03 Martin Liska <mliska@suse.cz>
32368 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32370 (ipa_node_params_t::remove): Likewise.
32371 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32372 initialization from removed ipa_node_params_t::insert.
32373 (ipa_node_params::~ipa_node_params): Move from removed
32374 ipa_node_params_t::release.
32375 * symbol-summary.h (symbol_summary::m_released): New member.
32376 Do not release a summary twice. Do not allow to call finalizer
32377 for types of a summary that live in GGC memory.
32379 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32381 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32384 2017-02-02 Martin Sebor <msebor@redhat.com>
32386 PR middle-end/79275
32387 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32388 (format_string): Tighten up the range of output for non-constant
32389 strings and correct the expected range for wide non-constant strings.
32391 2017-02-02 Martin Sebor <msebor@redhat.com>
32393 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32395 PR middle-end/32003
32396 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32398 (-fdump-tree-@var): Add to index and document how to come up
32399 with pass-specific option and dump file names.
32400 (-fdump-passes): Clarify where to look for output.
32402 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32404 PR middle-end/77445
32405 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32406 statistics of the analyzed path; allow threading for speed when
32407 any of BBs along the path are optimized for speed.
32409 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32411 PR middle-end/78468
32412 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32413 settings of the virtual registers.
32416 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32418 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32419 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32422 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32424 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32425 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32427 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32429 * config/s390/s390.md: Add missing comments with the expanded
32431 * config/s390/vector.md: Likewise.
32432 * config/s390/vx-builtins.md: Likewise.
32434 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32437 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32438 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32439 conditions on a single line.
32441 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32443 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32444 __S390_VX__ to __VX__.
32446 2017-02-01 Andrew Pinski <apinski@cavium.com>
32448 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32449 stmt_info to record_stmt_cost.
32450 (vect_get_known_peeling_cost): Pass stmt_info if known to
32452 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32453 cpu_vector_cost field into
32454 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32455 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32456 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32457 splitting of scalar_stmt_cost and vec_stmt_cost.
32458 (thunderx_vector_cost): Likewise.
32459 (cortexa57_vector_cost): LIkewise.
32460 (exynosm1_vector_cost): Likewise.
32461 (xgene1_vector_cost): Likewise.
32462 (thunderx2t99_vector_cost): Improve after the splitting of the two
32464 (aarch64_builtin_vectorization_cost): Update for the splitting of
32465 scalar_stmt_cost and vec_stmt_cost.
32467 2017-02-01 Torvald Riegel <triegel@redhat.com>
32468 Richard Henderson <rth@redhat.com>
32470 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32471 conditional on existance of a fast atomic load.
32472 * optabs-query.c (can_atomic_load_p): New function.
32473 * optabs-query.h (can_atomic_load_p): Declare it.
32474 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32475 no fast atomic load is available for the particular size of access.
32476 (expand_atomic_compare_and_swap): Likewise.
32477 (expand_atomic_load): Likewise.
32478 (expand_atomic_store): Likewise.
32479 (expand_atomic_fetch_op): Likewise.
32480 * testsuite/lib/target-supports.exp
32481 (check_effective_target_sync_int_128): Remove x86 because it provides
32482 no fast atomic load.
32483 (check_effective_target_sync_int_128_runtime): Likewise.
32485 2017-02-01 Richard Biener <rguenther@suse.de>
32487 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32488 (graphite_transform_loops): Provide opt-info for optimized nests.
32489 * tree-parloop.c (parallelize_loops): Provide opt-info for
32490 parallelized loops.
32492 2017-02-01 Richard Biener <rguenther@suse.de>
32494 PR middle-end/79315
32495 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32496 was not set before.
32498 2017-02-01 Richard Biener <rguenther@suse.de>
32500 PR tree-optimization/71824
32501 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32502 Verify the loops are valid in the merged SESE region.
32503 (scop_detection::can_represent_loop_1): Check analyzing the
32504 evolution of the number of iterations in the region succeeds.
32506 2017-01-31 Ian Lance Taylor <iant@golang.org>
32508 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32509 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32511 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32513 PR preprocessor/79210
32514 * input.c (get_substring_ranges_for_loc): Replace line_width
32515 assertion with error-handling.
32517 2017-01-31 Richard Biener <rguenther@suse.de>
32519 PR tree-optimization/77318
32520 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32521 (create_pw_aff_from_tree): Take loop parameter.
32522 (add_condition_to_pbb): Pass loop of the condition to
32523 create_pw_aff_from_tree.
32525 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32527 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32528 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32530 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32534 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32536 (convert_int_to_float128): Likewise.
32537 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32538 (convert_int_to_float128): Likewise.
32539 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32540 (UNSPEC_IEEE128_CONVERT): Likewise.
32541 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32542 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32543 Use local variables for IBM extended format.
32544 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32545 (fix_trunc<mode>si2_fprs): Likewise.
32546 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32547 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32548 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32549 to know that we can now have integers of all sizes in vector
32551 (fix<uns>_<mode>di2_hw): Likewise.
32552 (float<uns>_<mode>si2_hw): Likewise.
32553 (fix_<mode>si2_hw): Likewise.
32554 (fixuns_<mode>si2_hw): Likewise.
32555 (float<uns>_<mode>di2_hw): Likewise.
32556 (float_<mode>di2_hw): Likewise.
32557 (float_<mode>si2_hw): Likewise.
32558 (floatuns_<mode>di2_hw): Likewise.
32559 (floatuns_<mode>si2_hw): Likewise.
32560 (xscvqp<su>wz_<mode>): Delete, no longer used.
32561 (xscvqp<su>dz_<mode>): Likewise.
32562 (xscv<su>dqp_<mode>): Likewise.
32563 (ieee128_mfvsrd_64bit): Likewise.
32564 (ieee128_mfvsrd_32bit): Likewise.
32565 (ieee128_mfvsrwz): Likewise.
32566 (ieee128_mtvsrw): Likewise.
32567 (ieee128_mtvsrd_64bit): Likewise.
32568 (ieee128_mtvsrd_32bit): Likewise.
32570 2017-01-31 Martin Liska <mliska@suse.cz>
32573 * ipa-prop.c (ipa_free_all_node_params): Call release method
32574 instead of ~sumbol_summary to not to trigger double times
32577 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32579 PR tree-optimization/71691
32580 * bitmap.h (class auto_bitmap): New.
32581 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32582 is_maybe_undefined instead of ssa_undefined_value_p.
32584 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32586 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32587 __S390_ARCH_LEVEL__ to __ARCH__.
32589 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32591 PR tree-optimization/79267
32592 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32593 if should_remove_lhs_p is true.
32595 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32598 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32599 (add_alignment_attribute): New.
32600 (base_type_die): Add alignment attribute.
32601 (subrange_type_die): Likewise.
32602 (modified_type_die): Likewise.
32603 (gen_array_type_die): Likewise.
32604 (gen_descr_array_type_die: Likewise.
32605 (gen_enumeration_type_die): Likewise.
32606 (gen_subprogram_die): Likewise.
32607 (gen_variable_die): Likewise.
32608 (gen_field_die): Likewise.
32609 (gen_ptr_to_mbr_type_die): Likewise.
32610 (gen_struct_or_union_type_die): Likewise.
32611 (gen_subroutine_type_die): Likewise.
32612 (gen_typedef_die): Likewise.
32613 (base_type_cmp): Compare alignment attribute.
32615 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32618 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32619 (setb_unsigned) New pattern for setb with CCUNS.
32620 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32621 subfc./subfe sequence to avoid overflow problems. Generate a
32622 shorter sequence with cmpld/setb for power9.
32623 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32624 for generating subfc. instruction.
32625 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32626 now uses this instruction.
32628 2017-01-30 Ian Lance Taylor <iant@google.com>
32631 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32632 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32634 2017-01-30 Martin Sebor <msebor@redhat.com>
32636 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32637 Move constant to the right of a relational operator.
32638 (get_mpfr_format_length, format_character, format_string): Ditto.
32639 (should_warn_p, maybe_warn): Same.
32641 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32643 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32646 * asan.c (get_translation_unit_decl): Remove function.
32647 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32649 2017-01-30 Martin Liska <mliska@suse.cz>
32651 PR gcov-profile/79259
32652 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32653 -fprofile-generate.
32655 2017-01-30 Martin Liska <mliska@suse.cz>
32658 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32659 Initialize variables with NULL value.
32661 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32664 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32666 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32668 2017-01-30 Richard Biener <rguenther@suse.de>
32670 PR tree-optimization/79276
32671 * tree-vrp.c (process_assert_insertions): Properly adjust common
32672 when removing a duplicate.
32674 2017-01-30 Richard Biener <rguenther@suse.de>
32676 PR tree-optimization/79256
32677 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32678 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32680 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32682 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32685 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32686 ("*r<noxa>sbg_<mode>_sll_bitmask")
32687 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32688 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32689 Use contiguous_bitmask_nowrap_operand.
32691 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32694 * config/rs6000/altivec.h (vec_xl): Revise #define.
32695 (vec_xst): Likewise.
32697 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32699 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32701 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32703 PR rtl-optimization/79194
32704 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32705 traps before call to bypass_conditional_jumps.
32707 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32709 PR tree-optimization/71374
32710 * lra-constraints.c (check_conflict_input_operands): New.
32711 (match_reload): Use it.
32713 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32716 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32717 account to calculate conflict_set.
32719 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32721 PR rtl-optimization/78559
32722 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32723 other_insn in combine.
32725 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32727 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32728 uint16_type_node for BT_UINT16.
32730 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32732 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32733 "RTL Tests" to menu.
32734 (GIMPLE Tests): New node.
32735 (RTL Tests): New node.
32737 2017-01-27 Richard Biener <rguenther@suse.de>
32739 PR tree-optimization/79245
32740 * tree-loop-distribution.c (distribute_loop): Apply cost
32741 modeling also to detected patterns.
32743 2017-01-27 Richard Biener <rguenther@suse.de>
32745 PR tree-optimization/71433
32746 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32747 (compare_assert_loc): New function.
32748 (process_assert_insertions): Sort and optimize assert locations
32749 to remove duplicates and push down identical assertions on
32750 edges to their destination block.
32752 2017-01-27 Richard Biener <rguenther@suse.de>
32754 PR tree-optimization/79244
32755 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32756 out SSA names even if abnormal.
32758 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32760 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32761 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32762 instead of MPFR_RNDN.
32764 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32767 * arm.c (arm_option_override): Don't call build_target_option_node
32768 until after doing all option overrides.
32769 (arm_valid_target_attribute_tree): Likewise.
32771 2017-01-27 Martin Liska <mliska@suse.cz>
32773 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32774 for {cd}tors and C++ {cd}tors.
32776 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32778 * config/s390/s390.md ("*setmem_long_and")
32779 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32781 2017-01-26 Martin Sebor <msebor@redhat.com>
32783 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32786 2017-01-26 Martin Sebor <msebor@redhat.com>
32788 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32789 HAVE_DFmode before using XFmode or DFmode.
32790 (parse_directive): Avoid using the z length modifier to avoid
32791 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32793 PR middle-end/78703
32794 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32795 to accept adjustment as an array.
32796 (get_int_range): New function.
32797 (struct directive): Make width and prec arrays.
32798 (directive::set_width, directive::set_precision): Call get_int_range.
32799 (format_integer, format_floating): Handle width and precision ranges.
32800 (format_string, parse_directive): Same.
32802 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32805 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32806 comdat_type_p set, just clone them, but keep the children in the
32810 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32811 which have direct callers with -fvar-tracking-assignments enabled
32813 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32814 inside of type units.
32816 2017-01-26 Martin Sebor <msebor@redhat.com>
32818 PR middle-end/78703
32819 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32821 (struct format_result): Replace number_chars, number_chars_min,
32822 and number_chars_max with a single member of struct result_range.
32824 (format_result::operator+=): Adjust.
32825 (struct fmtresult): Remove bounded. Handle likely and unlikely
32827 (fmtresult::adjust_for_width_or_precision): New function.
32828 (fmtresult:type_max_digits): New function.
32829 (bytes_remaining): Handle likely and unlikely counters.
32830 (min_bytes_remaining): Remove.
32831 (format_percent): Simplify.
32832 (format_integer, format_floating): Set likely and unlikely counters.
32833 (get_string_length, format_character, format_string): Same.
32834 (format_plain, should_warn_p): New function.
32835 (maybe_warn): Call should_warn_p. Update diagnostic messages
32836 and handle those for all directives, including plain strings.
32837 (format_directive): Handle likely and unlikely counters.
32838 Remove unnecessary quoting from diagnostics. Add an informational
32840 (add_bytes): Remove.
32841 (pass_sprintf_length::compute_format_length): Simplify.
32842 (try_substitute_return_value): Handle likely and unlikely counters.
32844 2017-01-26 Carl Love <cel@us.ibm.com>
32846 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32847 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32849 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32852 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32853 endianess for subregs into account.
32854 * lra-constraints.c (lra_constraints): Do risky transformations
32855 always on the first iteration.
32856 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32857 last_call_used_reg_set.
32858 (process_bb_lives): Define and use last_call_used_reg_set.
32859 * lra.c (lra): Always continue after lra_constraints on the first
32862 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32864 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32866 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32868 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32870 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32871 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32872 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32873 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32874 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32875 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32876 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32877 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32878 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32880 2017-01-26 Marek Polacek <polacek@redhat.com>
32883 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32884 for the third operand.
32886 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32888 PR middle-end/79236
32889 * omp-low.c (struct omp_context): Add simt_stmt field.
32890 (scan_omp_for): Return omp_context *.
32891 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32892 context to the _simt_ SIMD stmt.
32893 (lower_omp_for): For combined SIMD with sibling _simt_
32894 SIMD, make sure to use the same decls in _looptemp_
32895 clauses as in the sibling.
32897 2017-01-26 David Sherwood <david.sherwood@arm.com>
32899 PR middle-end/79212
32900 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32903 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32906 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32907 emit fld b; fld a; if possible.
32909 * brig-builtins.def: Update copyright years.
32910 * config/arm/arm_acle_builtins.def: Update copyright years.
32912 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32915 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32916 constraint instead of o for the stxsd instruction.
32918 2017-01-25 Carl Love <cel@us.ibm.com>
32920 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32921 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32923 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32925 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32927 2017-01-25 Richard Biener <rguenther@suse.de>
32929 PR tree-optimization/69264
32930 * target.def (vector_alignment_reachable): Improve documentation.
32931 * doc/tm.texi: Regenerate.
32932 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32934 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32935 earlier changes with respect to TYPE_USER_ALIGN.
32936 (vector_alignment_reachable_p): Likewise. Improve dumping.
32938 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32941 * config/arm/arm.md (xordi3): Force constant operand into a register
32944 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32946 * doc/invoke.texi (-fstore-merging): Correct default optimization
32947 levels at which it is enabled.
32948 (-O): Move -fstore-merging from list to...
32951 2017-01-25 Richard Biener <rguenther@suse.de>
32954 * omp-expand.c: Include debug.h.
32955 (expand_omp_taskreg): Make sure to generate early debug before
32956 outlining anything from a function.
32957 (expand_omp_target): Likewise.
32958 (grid_expand_target_grid_body): Likewise.
32960 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32963 * asan.c (get_translation_unit_decl): New function.
32964 (asan_add_global): Extract modules file name from globals
32965 TRANSLATION_UNIT_DECL name.
32967 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32970 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32971 for long calls with APCS frame and VFP.
32973 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32975 * cfg.c (original_copy_tables_initialized_p): New function.
32976 * cfg.h (original_copy_tables_initialized_p): New decl.
32977 * cfgrtl.c (relink_block_chain): Guard the call to
32978 free_original_copy_tables with a call to
32979 original_copy_tables_initialized_p.
32980 * cgraph.h (symtab_node::native_rtl_p): New decl.
32981 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32982 (symtab_node::needed_p): Don't assert for early assembly output
32983 for __RTL functions.
32984 (cgraph_node::finalize_function): Set "force_output" for __RTL
32986 (cgraph_node::analyze): Bail out early for __RTL functions.
32987 (analyze_functions): Update assertion to support __RTL functions.
32988 (cgraph_node::expand): Bail out early for __RTL functions.
32989 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32991 * function.h (struct function): Update comment for field
32993 * gimple-expr.c: Include "tree-pass.h".
32994 (gimple_has_body_p): Return false for __RTL functions.
32995 * Makefile.in (OBJS): Add run-rtl-passes.o.
32996 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32998 (gcc::pass_manager::get_clean_slate): New accessor.
32999 * passes.c: Include "insn-addr.h".
33000 (should_skip_pass_p): Add logging. Update logic for running
33001 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33002 property-provider override so it is only done for gimple passes.
33004 (skip_pass): New function.
33005 (execute_one_pass): Call skip_pass when skipping passes.
33006 * read-md.c (md_reader::read_char): Support filtering
33007 the input to a subset of line numbers.
33008 (md_reader::md_reader): Initialize fields
33009 m_first_line and m_last_line.
33010 (md_reader::read_file_fragment): New function.
33011 * read-md.h (md_reader::read_file_fragment): New decl.
33012 (md_reader::m_first_line): New field.
33013 (md_reader::m_last_line): New field.
33014 * read-rtl-function.c (function_reader::create_function): Only
33015 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33016 curr_properties. Set DECL_INITIAL to a dummy block.
33017 (read_rtl_function_body_from_file_range): New function.
33018 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33020 * run-rtl-passes.c: New file.
33021 * run-rtl-passes.h: New file.
33023 2017-01-24 Jeff Law <law@redhat.com>
33025 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33028 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33030 PR tree-optimization/79159
33031 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33032 (record_nonwrapping_iv): Improve boundary using above function if no
33033 value range information.
33035 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33036 Martin Jambor <mjambor@suse.cz>
33038 * brig-builtins.def: New file.
33039 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33040 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33041 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33042 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33043 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33044 * builtin-types.def (BT_INT8): New.
33045 (BT_INT16): Likewise.
33046 (BT_UINT8): Likewise.
33047 (BT_UINT16): Likewise.
33048 (BT_FN_ULONG): Likewise.
33049 (BT_FN_UINT_INT): Likewise.
33050 (BT_FN_UINT_ULONG): Likewise.
33051 (BT_FN_UINT_LONG): Likewise.
33052 (BT_FN_UINT_PTR): Likewise.
33053 (BT_FN_ULONG_PTR): Likewise.
33054 (BT_FN_INT8_FLOAT): Likewise.
33055 (BT_FN_INT16_FLOAT): Likewise.
33056 (BT_FN_UINT32_FLOAT): Likewise.
33057 (BT_FN_UINT16_FLOAT): Likewise.
33058 (BT_FN_UINT8_FLOAT): Likewise.
33059 (BT_FN_UINT64_FLOAT): Likewise.
33060 (BT_FN_UINT16_UINT32): Likewise.
33061 (BT_FN_UINT32_UINT16): Likewise.
33062 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33063 (BT_FN_INT_PTR_INT): Likewise.
33064 (BT_FN_UINT_PTR_UINT): Likewise.
33065 (BT_FN_LONG_PTR_LONG): Likewise.
33066 (BT_FN_ULONG_PTR_ULONG): Likewise.
33067 (BT_FN_VOID_UINT64_UINT64): Likewise.
33068 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33069 (BT_FN_INT8_INT8_INT8): Likewise.
33070 (BT_FN_INT16_INT16_INT16): Likewise.
33071 (BT_FN_INT_INT_INT): Likewise.
33072 (BT_FN_UINT_FLOAT_UINT): Likewise.
33073 (BT_FN_FLOAT_UINT_UINT): Likewise.
33074 (BT_FN_ULONG_UINT_UINT): Likewise.
33075 (BT_FN_ULONG_UINT_PTR): Likewise.
33076 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33077 (BT_FN_UINT_UINT_UINT): Likewise.
33078 (BT_FN_VOID_UINT_PTR): Likewise.
33079 (BT_FN_UINT_UINT_PTR: Likewise.
33080 (BT_FN_UINT32_UINT64_PTR): Likewise.
33081 (BT_FN_INT_INT_UINT_UINT): Likewise.
33082 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33083 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33084 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33085 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33086 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33087 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33088 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33089 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33090 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33091 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33092 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33093 * doc/frontends.texi: List BRIG FE.
33094 * doc/install.texi (Testing): Add BRIG tesring requirements.
33095 * doc/invoke.texi (Overall Options): Mention BRIG.
33096 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33098 2017-01-24 Richard Biener <rguenther@suse.de>
33100 PR translation/79208
33101 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33103 2017-01-24 Martin Jambor <mjambor@suse.cz>
33106 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33107 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33108 and known_contexts.
33110 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33112 PR middle-end/79123
33113 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33114 casts from signed to unsigned really don't have a range.
33116 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33118 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33119 GMP_RNDx for compatiblity.
33121 2017-01-24 Martin Liska <mliska@suse.cz>
33124 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33125 that would prevent us to call alloca with -1 as argument.
33127 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33129 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33130 Avoid -Wformat-security warning.
33132 2017-01-23 Andrew Pinski <apinski@cavium.com>
33134 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33137 2017-01-23 Martin Sebor <msebor@redhat.com>
33139 PR middle-end/78703
33140 * gimple-ssa-sprintf.c (warn_level): New global.
33141 (format_integer): Use it here and throughout the rest of the file.
33142 Use the same switch to compute sign as base.
33143 (maybe_warn): New function.
33144 (format_directive): Factor out warnings into maybe_warn.
33145 Add debugging output. Use warn_level.
33146 (add_bytes): Use warn_level.
33147 (pass_sprintf_length::compute_format_length): Add debugging output.
33148 (try_substitute_return_value): Same.
33149 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33151 PR middle-end/78703
33152 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33153 (struct fmtresult, format_integer, format_floating): Adjust.
33154 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33155 (get_string_length, format_string,format_directive): Same.
33156 (pass_sprintf_length::compute_format_length): Same.
33157 (try_substitute_return_value): Simplify slightly.
33159 PR middle-end/78703
33160 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33161 (fmtresult::operator+=): Outlined.
33162 (struct fmtresult): Add ctors.
33163 (struct conversion_spec): Rename...
33164 (struct directive): ...to this. Add and remove data members.
33165 (directive::set_width, directive::set_precision): New functions.
33166 (format_percent): Use fmtresult ctor.
33167 (get_width_and_precision): Remove.
33168 (format_integer): Make naming changes. Avoid computing width and
33170 (format_floating): Same. Adjust indentation.
33171 (format_character, format_none): New functions.
33172 (format_string): Moved character handling to format_character.
33173 (format_directive): Remove arguments, change return type.
33174 (parse_directive): New function.
33175 (pass_sprintf_length::compute_format_length): Move directive
33176 parsing to parse_directive.
33178 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33180 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33181 (assign_assembler_name_if_needed): ... this.
33182 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33183 (assign_assembler_name_if_needed): ... this.
33184 (free_lang_data_in_cgraph): Adjust callers.
33185 * cgraphunit.c (cgraph_node::analyze): Likewise.
33186 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33189 2017-01-23 Richard Biener <rguenther@suse.de>
33191 PR tree-optimization/79088
33192 PR tree-optimization/79188
33193 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33194 resetting loop bounds after last path deletion. Reset loop
33195 bounds of the target loop, make code match the comments.
33196 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33197 Make sure loops need no fixups.
33199 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33201 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33202 exponent support with double type for first argument.
33203 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33204 type returned by __builtin_vec_extract_sig,
33205 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33206 functions from "vector int" to "vector unsigned int" or from
33207 "vector long long int" to "vector unsigned long long int".
33208 Changed type returned by __builtin_vec_extract_exp,
33209 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33210 functions from "vector int" to "vector unsigned int" or from
33211 "vector long long int" to "vector unsigned long long int".
33212 Changed return type of __builtin_vec_test_data_class,
33213 __builtin_vec_test_data_class_sp, and
33214 __builtin_vec_test_data_class_dp from "vector int" to
33215 "vector bool int" or from "vector long long int" to "vector bool
33216 long long int" and changed second argument type from "unsigned
33217 int" to "int". Added new overloaded function forms "vector float
33218 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33219 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33220 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33221 double, vector unsigned long long int)" and "vector double
33222 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33223 long int)". Changed return type of
33224 __builtin_scalar_test_data_class and
33225 __builtin_scalar_test_data_class_sp and
33226 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33227 int" and changed second argument from "unsigned int" to "int".
33228 Changed type returned by __builtin_scalar_test_neg,
33229 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33230 from "int" to "bool int". Added new overloaded function form
33231 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33232 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33233 exponent double-precision with floating point first argument.
33234 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33235 documentation of scalar_test_data_class, scalar_test_neg,
33236 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33237 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33238 vec_test_data_class built-in functions to reflect refinements in
33239 their type signatures.
33241 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33243 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33245 (aarch64_elf_asm_destructor): Likewise.
33247 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33249 PR rtl-optimization/78634
33250 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33251 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33252 * ifcvt.c (noce_try_cmove): Add missing cost check.
33254 PR rtl-optimization/71724
33255 * combine.c (if_then_else_cond): Look for situations where it is
33256 beneficial to undo the work of one of the recursive calls.
33258 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33260 PR tree-optimization/70754
33261 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33262 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33263 combined stmt before it if not NULL.
33264 (combine_chains): Process refs reversely and compute dominance point
33267 2017-01-23 Martin Liska <mliska@suse.cz>
33269 PR tree-optimization/79196
33270 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33271 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33273 (strlen_optimize_stmt): Call the renamed function.
33275 2017-01-23 Michael Matz <matz@suse.de>
33277 PR tree-optimization/78384
33278 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33280 2017-01-23 Richard Biener <rguenther@suse.de>
33282 PR tree-optimization/79186
33283 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33284 both incoming edges before moving an assert.
33286 2017-01-23 Martin Jambor <mjambor@suse.cz>
33288 * ipa-prop.c (load_from_param_1): Removed.
33289 (load_from_unmodified_param): Bits from load_from_param_1 put back
33291 (load_from_param): Removed.
33292 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33293 with stmt. Reverted back to use of load_from_unmodified_param.
33295 2017-01-23 Martin Jambor <mjambor@suse.cz>
33298 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33299 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33300 field a pointer to garbage collected vector, mark lattices and
33301 ipcp_orig_node with GTY((skip)).
33302 (ipa_get_param_count): Adjust to descriptors being a pointer.
33303 (ipa_get_param): Likewise.
33304 (ipa_get_type): Likewise.
33305 (ipa_get_param_move_cost): Likewise.
33306 (ipa_set_param_used): Likewise.
33307 (ipa_get_controlled_uses): Likewise.
33308 (ipa_set_controlled_uses): Likewise.
33309 (ipa_is_param_used): Likewise.
33310 (ipa_node_params_t): Move into garbage collector. New methods insert
33312 (ipa_node_params_sum): Annotate wth GTY(()).
33313 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33315 (ipa_load_from_parm_agg): Adjust declaration.
33316 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33317 * ipa-profile.c (ipa_profile): Likewise.
33318 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33319 (ipa_populate_param_decls): Make descriptors parameter garbage
33321 (ipa_dump_param): Adjust to descriptors being a pointer.
33322 (ipa_alloc_node_params): Likewise.
33323 (ipa_initialize_node_params): Likewise.
33324 (load_from_param_1): Make descriptors parameter garbage collected.
33325 (load_from_unmodified_param): Likewise.
33326 (load_from_param): Likewise.
33327 (ipa_load_from_parm_agg): Likewise.
33328 (ipa_node_params::~ipa_node_params): Removed.
33329 (ipa_free_all_node_params): Remove call to delete operator.
33330 (ipa_node_params_t::insert): New.
33331 (ipa_node_params_t::remove): Likewise.
33332 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33333 copy known_csts and known_contexts vectors.
33334 (ipa_read_node_info): Adjust to descriptors being a pointer.
33335 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33337 (ipcp_transform_function): Make descriptors variable garbage
33340 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33342 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33343 * config/i386/avx512dqintrin.h: Ditto.
33344 * config/i386/avx512fintrin.h: Ditto.
33345 * config/i386/i386.c: Handle new builtins.
33346 * config/i386/i386-builtin.def: Add new builtins.
33347 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33348 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33350 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33351 Martin Liska <mliska@suse.cz>
33353 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33354 * asan.c (asan_expand_poison_ifn): Support stores and use
33355 appropriate ASAN report function.
33356 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33357 * internal-fn.def (ASAN_POISON_USE): Declare.
33358 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33359 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33360 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33361 ASAN_POISON calls w/o LHS.
33362 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33363 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33364 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33365 * gimplify.c (asan_poison_variables): Add attribute
33366 use_after_scope_memory to variables that really needs to live
33368 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33369 having the attribute.
33371 2017-01-23 Martin Liska <mliska@suse.cz>
33373 * asan.c (create_asan_shadow_var): New function.
33374 (asan_expand_poison_ifn): Likewise.
33375 * asan.h (asan_expand_poison_ifn): New declaration.
33376 * internal-fn.c (expand_ASAN_POISON): Likewise.
33377 * internal-fn.def (ASAN_POISON): New builtin.
33378 * sanopt.c (pass_sanopt::execute): Expand
33379 asan_expand_poison_ifn.
33380 * tree-inline.c (copy_decl_for_dup_finish): Make function
33382 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33383 * tree-ssa.c (is_asan_mark_p): New function.
33384 (execute_update_addresses_taken): Rewrite local variables
33385 (identified just by use-after-scope as addressable) into SSA.
33387 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33389 * doc/install.texi (Specific): opensource.apple.com uses https
33390 now. Remove trailing slash.
33392 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33394 * README.Portability: Remove note on an Irix compatibility issue.
33396 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33398 * gcov.c (INCLUDE_ALGORITHM): Define.
33399 (INCLUDE_VECTOR): Define.
33400 No longer include <vector> and <algorithm> directly.
33402 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33404 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33406 * doc/invoke.texi (Code Gen Options): Ditto.
33408 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33411 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33413 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33415 rtl-optimization/79125
33416 * cprop.c (local_cprop_pass): Handle cases where we make an
33417 unconditional trap.
33419 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33423 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33424 read from, for big endian.
33426 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33428 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33429 register pauth builtins for LP64 only.
33431 2017-01-20 Marek Polacek <polacek@redhat.com>
33434 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33437 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33439 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33441 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33442 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33443 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33445 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33448 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33449 in big-endian mode when they are not a single duplicated value.
33451 2017-01-20 Richard Biener <rguenther@suse.de>
33453 * BASE-VER: Bump to 7.0.1.
33455 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33457 * omp-low.c (omplow_simd_context): New struct. Use it...
33458 (lower_rec_simd_input_clauses): ...here and...
33459 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33460 references to idx, lane, max_vf, is_simt.
33462 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33464 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33467 2017-01-20 Martin Jambor <mjambor@suse.cz>
33469 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33470 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33472 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33473 (GTFILES): Rename hsa.c to hsa-common.c.
33474 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33475 * hsa-dump.c: Likewise.
33476 * hsa-gen.c: Likewise.
33477 * hsa-regalloc.c: Likewise.
33478 * ipa-hsa.c: Likewise.
33479 * omp-expand.c: Likewise.
33480 * omp-low.c: Likewise.
33481 * toplev.c: Likewise.
33483 2017-01-20 Marek Polacek <polacek@redhat.com>
33486 * doc/invoke.texi: Document -Wduplicated-branches.
33487 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33488 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33489 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33490 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33491 return 0 only when not OEP_LEXICOGRAPHIC.
33492 (fold_build_cleanup_point_expr): Use the expression
33493 location when building CLEANUP_POINT_EXPR.
33494 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33495 * tree.c (add_expr): Handle error_mark_node.
33497 2017-01-20 Martin Liska <mliska@suse.cz>
33500 * tree-profile.c (init_ic_make_global_vars): Do not call
33502 (gimple_init_gcov_profiler): Likewise.
33504 2017-01-20 Martin Liska <mliska@suse.cz>
33507 * cgraph.h (maybe_create_reference): Remove argument and
33509 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33511 * ipa-cp.c (create_specialized_node): Likewise.
33512 * symtab.c (symtab_node::maybe_create_reference): Handle
33513 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33515 2017-01-20 Martin Liska <mliska@suse.cz>
33517 * read-rtl-function.c (function_reader::create_function): Use
33518 build_decl instread of build_decl_stat.
33520 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33522 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33523 * config/i386/avx512dqintrin.h: Ditto.
33524 * config/i386/avx512fintrin.h: Ditto.
33525 * config/i386/i386-builtin-types.def: Add new types.
33526 * config/i386/i386.c: Handle new types.
33527 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33528 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33529 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33530 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33531 (__builtin_ia32_kshiftridi): New.
33532 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33534 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33538 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33539 define to rs6000_init_stack_protect_guard.
33540 (rs6000_init_stack_protect_guard): New function.
33542 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33543 Yunqiang Su <yunqiang.su@imgtec.com>
33545 * config.gcc (supported_defaults): Add madd4.
33546 (with_madd4): Add validation.
33547 (all_defaults): Add madd4.
33548 * config/mips/mips.opt (mmadd4): New option.
33549 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33551 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33553 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33554 (ISA_HAS_FUSED_MADD4): Likewise.
33555 * doc/invoke.texi (-mmadd4): Document the new option.
33556 * doc/install.texi (--with-madd4): Document the new option.
33558 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33560 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33561 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33562 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33563 (aarch64_init_pauth_hint_builtins): New.
33564 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33565 (aarch64_expand_builtin): Expand new builtins.
33567 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33569 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33570 * combine-stack-adj.c (no_unhandled_cfa): Handle
33571 REG_CFA_TOGGLE_RA_MANGLE.
33572 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33573 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33574 info for return address signing.
33575 (aarch64_expand_epilogue): Likewise.
33577 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33579 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33580 * config/aarch64/aarch64-protos.h
33581 (aarch64_return_address_signing_enabled): New declaration.
33582 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33584 (aarch64_expand_prologue): Sign return address before it's pushed onto
33586 (aarch64_expand_epilogue): Authenticate return address fetched from
33588 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33589 (aarch64_attributes): New function attributes for "sign-return-address".
33590 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33591 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33592 ("*do_return"): Generate combined instructions according to key index.
33593 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33594 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33596 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33597 * config/aarch64/aarch64.opt (msign-return-address=): New.
33598 * doc/extend.texi (AArch64 Function Attributes): Documents
33599 "sign-return-address=".
33600 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33602 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33604 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33605 overall option summary.
33607 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33609 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33610 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33611 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33612 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33614 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33616 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33617 -mpower9-minmax by default for -mcpu=power9.
33618 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33619 128-bit floating point.
33621 2017-01-20 Alan Modra <amodra@gmail.com>
33623 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33624 optimizing for size.
33626 2017-01-20 Alan Modra <amodra@gmail.com>
33629 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33630 for strcmp and strncmp from corresponding builtin decl.
33632 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33634 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33635 instead of i386/rtems-64.h.
33636 * config/i386/rtems-64.h: Remove.
33638 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33642 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33644 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33646 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33648 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33649 Change int to HOST_WIDE_INT.
33650 * config/aarch64/aarch64-protos.h
33651 (aarch64_simd_gen_const_vector_dup): Likewise.
33652 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33654 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33656 * langhooks-def.h (lhd_type_for_size): New decl.
33657 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33658 * langhooks.c (lhd_type_for_size): New function, taken from
33661 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33663 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33664 define_bypass for CR latency.
33665 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33666 (power9-alu2): Add define_bypass for CR latency.
33668 (power9-mul): Update insn latency.
33669 (power9-mul-compare): Update insn latency, bypass latency and remove
33672 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33674 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33676 * config/aarch64/aarch64.md
33677 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33678 aarch64_nopcrelative_literal_loads.
33679 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33681 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33683 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33684 TARGET_LOONGSON_3A.
33685 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33687 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33690 * config.gcc (supported_defaults): Add lxc1-sxc1.
33691 (with_lxc1_sxc1): Add validation.
33692 (all_defaults): Add lxc1-sxc1.
33693 * config/mips/mips.opt (mlxc1-sxc1): New option.
33694 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33696 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33697 __mips_no_lxc1_sxc1.
33698 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33699 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33700 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33702 2017-01-19 Richard Biener <rguenther@suse.de>
33704 PR tree-optimization/72488
33705 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33706 sure to restore SSA info.
33707 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33709 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33711 PR rtl-optimization/79121
33712 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33713 of the inner type when shifting an extended value.
33715 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33718 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33719 interposable aliases.
33721 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33724 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33725 Use the evmergelohi instruction.
33726 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33727 (mov_sitf_e500_subreg8_2_be): Likewise.
33728 (mov_sitf_e500_subreg12_2_le): Likewise.
33729 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33730 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33731 (mov_sitf_e500_subreg8_2_le): Likewise.
33732 (mov_sitf_e500_subreg12_2_be): Likewise.
33734 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33736 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33737 attribute from vecsimple to vecperm.
33738 (altivec_vbpermq2): Likewise.
33740 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33743 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33745 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33746 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33747 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33748 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33749 case where N arg is SIZE_MAX.
33750 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33751 (cmpstrsi): Add pattern.
33753 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33755 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33756 __builtin_vec_revb builtins.
33757 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33758 built-in functions to support generation of the ISA 3.0 XXBR<x>
33759 vector byte reverse instructions.
33760 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33761 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33762 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33763 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33764 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33765 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33766 (P9V_BUILTIN_VEC_REVB): Likewise.
33767 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33768 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33769 (p9_xxbrq_v16qi): Likewise.
33770 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33771 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33772 (p9_xxbrh_v8hi): Likewise.
33773 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33774 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33775 vec_revb built-in functions.
33777 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33779 PR rtl-optimization/78952
33780 * config/i386/i386.md (any_extract): New code iterator.
33781 (*insvqi_2): Use any_extract for source operand.
33782 (*insvqi_3): Use any_shiftrt for source operand.
33784 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33786 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33788 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33790 2017-01-18 Matthias Klose <doko@ubuntu.com>
33792 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33794 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33796 * config/rs6000/altivec.h (vec_bperm): Change #define.
33797 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33798 (altivec_vbpermq2): New define_insn.
33799 (altivec_vbpermd): Likewise.
33800 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33801 function interface.
33802 (VBPERMD): Likewise.
33803 (VBPERM): New polymorphic function interface.
33804 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33805 Add entries for P9V_BUILTIN_VEC_VBPERM.
33806 * doc/extend.texi: Add interfaces for vec_bperm.
33808 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33810 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33811 first letter of error messages.
33812 (s390_resolve_overloaded_builtin): Likewise.
33813 * config/s390/s390.c (s390_expand_builtin): Likewise.
33814 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33815 (s390_valid_target_attribute_inner_p): Likewise.
33816 * config/s390/s390.md ("tabort"): Likewise.
33818 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33820 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33821 (ISA_AVOID_DIV_HILO): New macro.
33822 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33823 (ISA_HAS_DDIV): Likewise.
33825 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33827 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33829 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33831 * doc/invoke.texi (fabi-version): Spelling fix.
33833 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33836 * doc/invoke.texi (fabi-version): Mention mangling fix for
33839 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33842 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33844 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33847 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33848 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33850 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33852 (stack_protect_test): Ditto.
33853 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33854 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33856 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33857 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33858 -mstack-protector-guard-offset=.
33859 (RS/6000 and PowerPC Options): Ditto.
33861 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33863 * config/i386/i386.h (MASK_CLASS_P): New define.
33864 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33865 there are no registers from different register sets also when
33866 mask registers are used. Update function comment.
33867 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33868 to (*k/*r) and (*k/*km) alternatives.
33870 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33872 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33873 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33874 (EH_RETURN_HANDLER_RTX): New define.
33875 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33876 Force frame pointer in EH return functions.
33877 (aarch64_expand_epilogue): Add barrier for eh_return.
33878 (aarch64_final_eh_return_addr): Remove.
33879 (aarch64_eh_return_handler_rtx): New function.
33880 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33882 (aarch64_eh_return_handler_rtx): New prototype.
33884 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33886 * config/rs6000/altivec.h (vec_rlmi): New #define.
33887 (vec_vrlnm): Likewise.
33888 (vec_rlnm): Likewise.
33889 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33890 (UNSPEC_VRLNM): Likewise.
33891 (VIlong): New mode iterator.
33892 (altivec_vrl<VI_char>mi): New define_insn.
33893 (altivec_vrl<VI_char>nm): Likewise.
33894 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33896 (VRLDNM): Likewise.
33897 (RLNM): New polymorphic function entry.
33898 (VRLWMI): New monomorphic function entry.
33899 (VRLDMI): Likewise.
33900 (RLMI): New polymorphic function entry.
33901 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33902 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33903 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33906 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33909 * dwarf2out.c (field_byte_offset): Restore the
33910 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33911 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33912 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33915 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33918 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33920 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33923 * dwarf2out.c (add_data_member_location_attribute): For constant
33924 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33925 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33926 DW_AT_byte_size attributes.
33928 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33930 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33931 after forcing to constant memory when the code model is medium.
33933 2017-01-17 Julia Koval <julia.koval@intel.com>
33936 * config/i386/avx512fintrin.h
33937 (_mm512_i32gather_ps): Change __addr type to void const*.
33938 (_mm512_mask_i32gather_ps): Ditto.
33939 (_mm512_i32gather_pd): Ditto.
33940 (_mm512_mask_i32gather_pd): Ditto.
33941 (_mm512_i64gather_ps): Ditto.
33942 (_mm512_mask_i64gather_ps): Ditto.
33943 (_mm512_i64gather_pd): Ditto.
33944 (_mm512_mask_i64gather_pd): Ditto.
33945 (_mm512_i32gather_epi32): Ditto.
33946 (_mm512_mask_i32gather_epi32): Ditto.
33947 (_mm512_i32gather_epi64): Ditto.
33948 (_mm512_mask_i32gather_epi64): Ditto.
33949 (_mm512_i64gather_epi32): Ditto.
33950 (_mm512_mask_i64gather_epi32): Ditto.
33951 (_mm512_i64gather_epi64): Ditto.
33952 (_mm512_mask_i64gather_epi64): Ditto.
33953 (_mm512_i32scatter_ps): Change __addr type to void*.
33954 (_mm512_mask_i32scatter_ps): Ditto.
33955 (_mm512_i32scatter_pd): Ditto.
33956 (_mm512_mask_i32scatter_pd): Ditto.
33957 (_mm512_i64scatter_ps): Ditto.
33958 (_mm512_mask_i64scatter_ps): Ditto.
33959 (_mm512_i64scatter_pd): Ditto.
33960 (_mm512_mask_i64scatter_pd): Ditto.
33961 (_mm512_i32scatter_epi32): Ditto.
33962 (_mm512_mask_i32scatter_epi32): Ditto.
33963 (_mm512_i32scatter_epi64): Ditto.
33964 (_mm512_mask_i32scatter_epi64): Ditto.
33965 (_mm512_i64scatter_epi32): Ditto.
33966 (_mm512_mask_i64scatter_epi32): Ditto.
33967 (_mm512_i64scatter_epi64): Ditto.
33968 (_mm512_mask_i64scatter_epi64): Ditto.
33969 * config/i386/avx512pfintrin.h
33970 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33971 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33972 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33973 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33974 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33975 (_mm512_prefetch_i32scatter_ps): Ditto.
33976 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33977 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33978 (_mm512_prefetch_i64scatter_pd): Ditto.
33979 (_mm512_prefetch_i64scatter_ps): Ditto.
33980 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33981 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33982 * config/i386/avx512vlintrin.h
33983 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33984 (_mm_mmask_i32gather_ps): Ditto.
33985 (_mm256_mmask_i32gather_pd): Ditto.
33986 (_mm_mmask_i32gather_pd): Ditto.
33987 (_mm256_mmask_i64gather_ps): Ditto.
33988 (_mm_mmask_i64gather_ps): Ditto.
33989 (_mm256_mmask_i64gather_pd): Ditto.
33990 (_mm_mmask_i64gather_pd): Ditto.
33991 (_mm256_mmask_i32gather_epi32): Ditto.
33992 (_mm_mmask_i32gather_epi32): Ditto.
33993 (_mm256_mmask_i32gather_epi64): Ditto.
33994 (_mm_mmask_i32gather_epi64): Ditto.
33995 (_mm256_mmask_i64gather_epi32): Ditto.
33996 (_mm_mmask_i64gather_epi32): Ditto.
33997 (_mm256_mmask_i64gather_epi64): Ditto.
33998 (_mm_mmask_i64gather_epi64): Ditto.
33999 (_mm256_i32scatter_ps): Change __addr type to void*.
34000 (_mm256_mask_i32scatter_ps): Ditto.
34001 (_mm_i32scatter_ps): Ditto.
34002 (_mm_mask_i32scatter_ps): Ditto.
34003 (_mm256_i32scatter_pd): Ditto.
34004 (_mm256_mask_i32scatter_pd): Ditto.
34005 (_mm_i32scatter_pd): Ditto.
34006 (_mm_mask_i32scatter_pd): Ditto.
34007 (_mm256_i64scatter_ps): Ditto.
34008 (_mm256_mask_i64scatter_ps): Ditto.
34009 (_mm_i64scatter_ps): Ditto.
34010 (_mm_mask_i64scatter_ps): Ditto.
34011 (_mm256_i64scatter_pd): Ditto.
34012 (_mm256_mask_i64scatter_pd): Ditto.
34013 (_mm_i64scatter_pd): Ditto.
34014 (_mm_mask_i64scatter_pd): Ditto.
34015 (_mm256_i32scatter_epi32): Ditto.
34016 (_mm256_mask_i32scatter_epi32): Ditto.
34017 (_mm_i32scatter_epi32): Ditto.
34018 (_mm_mask_i32scatter_epi32): Ditto.
34019 (_mm256_i32scatter_epi64): Ditto.
34020 (_mm256_mask_i32scatter_epi64): Ditto.
34021 (_mm_i32scatter_epi64): Ditto.
34022 (_mm_mask_i32scatter_epi64): Ditto.
34023 (_mm256_i64scatter_epi32): Ditto.
34024 (_mm256_mask_i64scatter_epi32): Ditto.
34025 (_mm_i64scatter_epi32): Ditto.
34026 (_mm_mask_i64scatter_epi32): Ditto.
34027 (_mm256_i64scatter_epi64): Ditto.
34028 (_mm256_mask_i64scatter_epi64): Ditto.
34029 (_mm_i64scatter_epi64): Ditto.
34030 (_mm_mask_i64scatter_epi64): Ditto.
34031 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34032 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34033 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34034 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34035 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34036 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34037 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34038 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34039 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34040 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34041 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34042 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34043 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34044 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34045 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34046 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34047 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34048 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34049 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34050 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34051 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34052 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34053 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34054 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34055 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34056 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34057 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34058 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34059 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34060 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34061 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34062 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34063 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34064 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34065 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34066 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34067 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34068 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34069 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34070 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34071 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34072 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34073 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34074 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34075 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34076 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34077 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34078 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34079 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34080 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34081 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34082 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34083 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34084 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34085 definitions accordingly.
34087 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34088 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34091 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34094 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34097 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34098 hard regs for inner regno.
34100 2017-01-17 Martin Liska <mliska@suse.cz>
34103 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34104 assumption and add comment.
34106 2017-01-17 Nathan Sidwell <nathan@acm.org>
34108 * ipa-visibility.c (localize_node): New function, broken out of ...
34109 (function_and_variable_visibility): ... here. Call it.
34111 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34113 PR middle-end/77445
34114 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34115 correctly set frequency of oudgoing edge.
34116 (duplicate_thread_path): Fix profile updating.
34118 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34121 * configure.ac: Add GCC_BASE_VER.
34122 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34123 version from BASE-VER file.
34124 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34125 (gcc.o): Depend on $(BASEVER).
34126 * common.opt (dumpfullversion): New option.
34127 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34128 * doc/invoke.texi: Document -dumpfullversion.
34129 * doc/install.texi: Document --with-gcc-major-version-only.
34130 * configure: Regenerated.
34132 2017-01-17 Richard Biener <rguenther@suse.de>
34134 PR tree-optimization/71433
34135 * tree-vrp.c (register_new_assert_for): Merge same asserts
34136 on all incoming edges.
34137 (process_assert_insertions_for): Handle insertions at the
34140 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34142 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34143 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34145 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34148 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34151 2017-01-17 Alan Modra <amodra@gmail.com>
34154 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34155 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34156 symbolic stack limit when pic.
34158 2017-01-16 Martin Sebor <msebor@redhat.com>
34160 PR tree-optimization/78608
34161 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34163 2017-01-16 Jeff Law <law@redhat.com>
34166 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34167 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34168 for several include directories that may be relative to sysroot.
34169 * config/i386/x-mingw32 (gplus_includedir): Define.
34170 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34171 (native_system_includedir): Likewise.
34172 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34173 override if TARGET_SYSTEM_ROOT is defined.
34174 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34176 PR tree-optimization/79090
34177 PR tree-optimization/33562
34178 PR tree-optimization/61912
34179 PR tree-optimization/77485
34180 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34181 and computed trims into the dump file.
34183 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34185 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34187 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34190 * gimplify.c (gimplify_init_constructor): If want_value and
34191 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34195 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34196 sequence. Formatting fixes.
34197 (doloop_optimize): Formatting fixes.
34200 * gcc.c (debug_level_greater_than_spec_func): New function.
34201 (static_spec_functions): Add debug-level-gt spec function.
34202 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34204 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34205 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34206 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34207 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34208 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34209 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34211 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34213 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34214 QImode fixups to general and mask registers only.
34216 2017-01-16 Carl Love <cel@us.ibm.com>
34218 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34219 for built-in functions
34220 vector signed char vec_nabs (vector signed char)
34221 vector signed short vec_nabs (vector signed short)
34222 vector signed int vec_nabs (vector signed int)
34223 vector signed long long vec_nabs (vector signed long long)
34224 vector float vec_nabs (vector float)
34225 vector double vec_nabs (vector double)
34226 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34228 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34229 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34230 * doc/extend.texi: Update the documentation file for the new built-in
34233 2017-01-16 Martin Sebor <msebor@redhat.com>
34235 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34238 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34240 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34241 UNSPEC_VSX__XXSPLTD to require special splat handling.
34243 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34246 * system.h: Poison strndup.
34248 2017-01-16 Alan Modra <amodra@gmail.com>
34251 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34254 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34256 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34258 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34260 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34261 call recog here. Assert that INSN_CODE (insn) is non-negative.
34263 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34266 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34268 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34269 in the currently scheduled RTL fragment.
34271 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34273 PR rtl-optimization/78751
34274 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34277 2017-01-14 Jeff Law <law@redhat.com>
34279 PR tree-optimization/79090
34280 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34281 variable length stores.
34282 (compute_trims): Delete dead assignment to *trim_tail.
34283 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34286 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34288 PR rtl-optimization/78626
34289 PR rtl-optimization/78727
34290 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34291 of a block, and split such blocks after everything else is finished.
34293 2017-01-14 Alan Modra <amodra@gmail.com>
34296 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34297 target legitimate_combined_insn.
34298 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34299 (rs6000_legitimate_combined_insn): New function.
34300 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34302 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34303 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34304 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34306 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34308 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34310 2017-01-13 Jeff Law <law@redhat.com>
34312 PR tree-optimization/33562
34313 PR tree-optimization/61912
34314 PR tree-optimization/77485
34315 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34317 (delete_dead_assignment): Likewise.
34318 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34319 statement to delete_dead_call and delete_dead_assignment.
34321 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34324 * substring-locations.c (format_warning_va): Strengthen case 1 so
34325 that both endpoints of the substring must be within the format
34326 range for just the substring to be printed.
34328 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34330 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34331 * config/i386/i386.c (ix86_target_string): Add missing options
34332 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34333 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34334 flags_other and ix86_target_other to flags2_other. Display unknown
34336 (ix86_valid_target_attribute_inner_p): Add missing options and
34337 reorder options by implied ISAs, as in ix86_target_string.
34339 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34341 * hash-table.h (hash_table::too_empty_p): New function.
34342 (hash_table::expand): Use it.
34343 (hash_table::traverse): Likewise.
34344 (hash_table::empty_slot): Use sizeof (value_type) instead of
34345 sizeof (PTR) to convert bytes to elements. Shrink the table
34346 if the current size is excessive for the current number of
34349 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34351 * ira-costs.c (record_reg_classes): Break from the inner loop
34352 early once alt_fail is known to be true. Update outer loop
34353 handling accordingly.
34355 2017-01-13 Jeff Law <law@redhat.com>
34357 * tree-ssa-dse.c (decrement_count): New function.
34358 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34359 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34360 when we know the partially dead statement is a mem* function.
34362 PR tree-optimization/61912
34363 PR tree-optimization/77485
34364 * tree-ssa-dse.c: Include expr.h.
34365 (maybe_trim_constructor_store): New function.
34366 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34368 PR tree-optimization/33562
34369 PR tree-optimization/61912
34370 PR tree-optimization/77485
34371 * doc/invoke.texi: Document new dse-max-object-size param.
34372 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34373 * tree-ssa-dse.c: Include params.h.
34374 (dse_store_status): New enum.
34375 (initialize_ao_ref_for_dse): New, partially extracted from
34377 (valid_ao_ref_for_dse, normalize_ref): New.
34378 (setup_live_bytes_from_ref, compute_trims): Likewise.
34379 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34380 (maybe_trim_partially_dead_store): Likewise.
34381 (maybe_trim_complex_store): Likewise.
34382 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34383 Track what bytes live from the original store. Return tri-state
34384 for dead, partially dead or live.
34385 (dse_dom_walker): Add constructor, destructor and new private members.
34386 (delete_dead_call, delete_dead_assignment): New extracted from
34388 (dse_optimize_stmt): Make a member of dse_dom_walker.
34389 Use initialize_ao_ref_for_dse.
34391 PR tree-optimization/33562
34392 PR tree-optimization/61912
34393 PR tree-optimization/77485
34394 * sbitmap.h (bitmap_count_bits): Prototype.
34395 (bitmap_clear_range, bitmap_set_range): Likewise.
34396 * sbitmap.c (bitmap_clear_range): New function.
34397 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34399 2017-01-13 Martin Liska <mliska@suse.cz>
34402 * function.c (set_cfun): Add new argument force.
34403 * function.h (set_cfun): Likewise.
34404 * ipa-inline-transform.c (inline_call): Use the function when
34405 strict alising from is dropped for function we inline to.
34407 2017-01-13 Richard Biener <rguenther@suse.de>
34409 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34410 for dumping GIMPLE INTEGER_CSTs.
34412 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34414 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34415 to 201112L since C++17.
34417 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34420 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34421 if -fsanitize=kernel-address is present.
34423 2017-01-13 Richard Biener <rguenther@suse.de>
34425 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34426 as _Literal ( type ) number in case usual suffixes do not
34427 preserve all information.
34429 2017-01-13 Richard Biener <rguenther@suse.de>
34431 PR tree-optimization/77283
34432 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34433 and ssa-iterators.h.
34434 (is_feasible_trace): Implement a cost model based on joiner
34437 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34440 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34441 char or short to __float128/_Float128 directly.
34443 2017-01-12 Martin Sebor <msebor@redhat.com>
34445 to -Wformat-overflow.
34446 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34447 (min_bytes_remaining): Same.
34448 (get_string_length): Same.
34449 (format_string): Same.
34450 (format_directive): Same.
34452 (pass_sprintf_length::handle_gimple_call): Same.
34454 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34456 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34457 info.nowrite calls with no lhs that can't throw. Return bool
34458 whether gsi_remove has been called or not.
34459 (pass_sprintf_length::handle_gimple_call): Return bool whether
34460 try_substitute_return_value called gsi_remove. Formatting fix.
34461 (pass_sprintf_length::execute): Don't use gsi_remove if
34462 handle_gimple_call returned true.
34465 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34466 be removed due to side-effects, don't remove following barrier nor
34467 turn the successor edge into fallthru edge.
34469 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34472 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34473 element-reversing loads and stores as not swappable.
34475 2017-01-12 Nathan Sidwell <nathan@acm.org>
34476 Nicolai Stange <nicstange@gmail.com>
34478 * combine.c (try_combine): Don't ignore result of overlap checking
34479 loop. Combine overlap & asm check into single loop.
34481 2017-01-12 Richard Biener <rguenther@suse.de>
34483 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34484 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34486 2017-01-12 Richard Biener <rguenther@suse.de>
34488 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34489 and TS_TARGET_OPTION directly derive from TS_BASE.
34490 * tree-core.h (tree_optimization_option): Derive from tree_base.
34491 (tree_target_option): Likewise.
34493 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34495 * config/i386/i386.c (memory_address_length): Increase len
34496 only when rip_relative_addr_p returns false.
34498 2017-01-11 Julia Koval <julia.koval@intel.com>
34500 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34501 (OPTION_MASK_ISA_SGX_SET): New.
34502 (ix86_handle_option): Handle OPT_msgx.
34503 * config.gcc: Added sgxintrin.h.
34504 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34505 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34506 * config/i386/i386.c (ix86_target_string): Add -msgx.
34508 (ix86_option_override_internal): Handle new options.
34509 (ix86_valid_target_attribute_inner_p): Add sgx.
34510 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34511 * config/i386/i386.opt: Add msgx.
34512 * config/i386/sgxintrin.h: New file.
34513 * config/i386/x86intrin.h: Add sgxintrin.h.
34515 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34518 * fold-const.c (maybe_nonzero_address): Return 1 for function
34520 (tree_single_nonzero_warnv_p): Don't handle function local objects
34524 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34527 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34530 * opts.c: Include "spellcheck.h"
34531 (struct string_fragment): New struct.
34532 (struct edit_distance_traits<const string_fragment &>): New
34534 (get_closest_sanitizer_option): New function.
34535 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34537 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34539 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34541 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34542 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34543 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34544 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34545 for initial die_offset if dwarf_split_debug_info.
34546 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34547 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34549 (output_skeleton_debug_sections): Formatting fix. Use
34550 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34551 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34553 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34555 * config/arm/cortex-a53.md: Add bypasses for
34556 cortex_a53_r2f_cvt.
34557 (cortex_a53_r2f): Only use for transfers.
34558 (cortex_a53_f2r): Likewise.
34559 (cortex_a53_r2f_cvt): Add reservation for conversions.
34560 (cortex_a53_f2r_cvt): Likewise.
34562 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34564 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34565 to all inlined functions, change static to extern.
34567 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34570 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34572 (arm_assemble_integer): Likewise.
34574 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34576 * config.gcc: Use new awk script to check CPU, FPU and architecture
34577 parameters for --with-... options.
34578 * config/arm/parsecpu.awk: New file
34579 * config/arm/arm-cpus.in: New file.
34580 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34582 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34584 * config/arm/t-arm: Update dependency rules.
34585 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34586 of processing .def files.
34587 * config/arm/genopt.sh: Deleted.
34588 * config/arm/gentune.sh: Deleted.
34589 * config/arm/arm-cores.def: Deleted.
34590 * config/arm/arm-arches.def: Deleted.
34591 * config/arm/arm-fpus.def: Deleted.
34592 * config/arm/arm-tune.md: Regenerated.
34593 * config/arm/arm-tables.opt: Regenerated.
34594 * config/arm/arm-cpu.h: New generated file.
34595 * config/arm/arm-cpu-data.h: New generated file.
34596 * config/arm/arm-cpu-cdata.h: New generated file.
34598 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34601 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34603 (input_varpool_node): Unpack dynamically_initialized bit.
34605 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34607 PR rtl-optimization/79032
34608 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34609 the alignment of the adjusted memory reference against that of MODE,
34610 instead of the alignment of the original memory reference.
34612 2017-01-11 Martin Jambor <mjambor@suse.cz>
34614 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34616 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34617 decorated functions.
34619 2017-01-11 Richard Biener <rguenther@suse.de>
34621 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34622 set range/nonnull info for PHI results. Do not set it on
34623 stmts marked for removal.
34625 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34627 * expr.c (store_field): In the bitfield case, fetch the return value
34628 from the registers before applying a single big-endian adjustment.
34629 Always do a final load for a BLKmode value not larger than a word.
34631 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34634 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34635 that we correctly handle column numbers greater than
34636 LINE_MAP_MAX_COLUMN_NUMBER.
34638 2017-01-10 Martin Sebor <msebor@redhat.com>
34640 PR middle-end/78245
34641 * gimple-ssa-sprintf.c (get_destination_size): Call
34642 {init,fini}object_sizes.
34643 * tree-object-size.c (addr_object_size): Adjust.
34644 (pass_through_call): Adjust.
34645 (pass_object_sizes::execute): Adjust.
34646 * tree-object-size.h (fini_object_sizes): Declare.
34648 2017-01-10 Martin Sebor <msebor@redhat.com>
34650 PR tree-optimization/78775
34651 * builtins.c (get_size_range): Move...
34652 * calls.c: ...to here.
34653 (alloc_max_size): Accept zero argument.
34654 (operand_signed_p): Remove.
34655 (maybe_warn_alloc_args_overflow): Call get_size_range.
34656 * calls.h (get_size_range): Declare.
34658 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34660 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34661 from TI's devices.csv file as of September 2016.
34662 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34664 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34666 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34667 * doc/invoke.texi: Likewise.
34668 * doc/md.texi: Likewise.
34669 * doc/objc.texi: Likewise.
34671 2017-01-10 Joshua Conner <joshconner@google.com>
34673 * config/arm/fuchsia-elf.h: New file.
34674 * config/fuchsia.h: New file.
34675 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34676 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34678 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34680 2016-01-10 Richard Biener <rguenther@suse.de>
34682 PR tree-optimization/79034
34683 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34684 Propagate out degenerate PHIs in the joiner.
34686 2017-01-10 Martin Liska <mliska@suse.cz>
34688 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34689 (sort_congruence_classes_by_decl_uid): Likewise.
34690 (sort_congruence_class_groups_by_decl_uid): Likewise.
34691 (sem_item_optimizer::merge_classes): Sort class, groups in these
34692 classes and members in the groups by DECL_UID of declarations.
34693 This would make merge operations stable.
34695 2017-01-10 Martin Liska <mliska@suse.cz>
34697 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34698 usage of m_classes_vec.
34699 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34700 (sem_item_optimizer::get_group_by_hash): Likewise.
34701 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34702 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34703 (sem_item_optimizer::verify_classes): Likewise.
34704 (sem_item_optimizer::process_cong_reduction): Likewise.
34705 (sem_item_optimizer::dump_cong_classes): Likewise.
34706 (sem_item_optimizer::merge_classes): Likewise.
34707 * ipa-icf.h (congruence_class_hash): Rename from
34708 congruence_class_group_hash. Remove declaration of m_classes_vec.
34710 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34712 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34713 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34714 * config.gcc: Add avx512vpopcntdqintrin.h.
34715 * config/i386/avx512vpopcntdqintrin.h: New.
34716 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34717 * config/i386/i386-builtin-types.def: Add new types.
34718 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34719 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34720 __builtin_ia32_vpopcountq_v8di_mask): New.
34721 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34722 __AVX512VPOPCNTDQ__.
34723 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34724 (PTA_AVX512VPOPCNTDQ): Define.
34725 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34726 TARGET_AVX512VPOPCNTDQ_P): Define.
34727 * config/i386/i386.opt: Add mavx512vpopcntdq.
34728 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34729 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34731 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34733 PR middle-end/77484
34734 * predict.def (PRED_CALL): Set to 67.
34736 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34738 * expr.c (store_field): In the bitfield case, if the value comes from
34739 a function call and is of an aggregate type returned in registers, do
34740 not modify the field mode; extract the value in all cases if the mode
34741 is BLKmode and the size is not larger than a word.
34743 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34746 * config/i386/cpuid.h: Fix undefined behavior.
34748 2017-01-04 Jeff Law <law@redhat.com>
34750 PR tree-optimization/79007
34751 PR tree-optimization/67955
34752 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34753 conservative for pt.null when flag_non_call_exceptions is on.
34755 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34757 PR translation/79019
34758 PR translation/79020
34759 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34760 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34761 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34763 * config/avr/avr.opt (maccumulate-args): Likewise.
34764 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34765 * common.opt (freport-bug): Likewise.
34766 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34767 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34768 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34769 translatable string.
34770 * config/i386/i386.c (function_value_32): Likewise.
34771 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34772 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34774 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34775 * common/config/msp430/msp430-common.c (msp430_handle_option):
34777 * symtab.c (symtab_node::verify_base): Likewise.
34778 * opts.c (set_debug_level): Likewise.
34779 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34780 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34781 missing whitespace to translatable strings.
34782 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34783 * config/sh/superh.h: Likewise.
34784 * config/i386/xopintrin.h: Likewise.
34785 * config/i386/znver1.md: Likewise.
34786 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34787 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34788 * double-int.h (struct double_int): Likewise.
34789 * double-int.c (div_and_round_double): Likewise.
34790 * wide-int.cc: Likewise.
34791 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34792 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34793 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34794 (crossjumps_occurred): ... this.
34795 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34798 PR tree-optimization/78899
34799 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34800 returning bool return struct loop *, NULL for failure and the new
34802 (versionable_outer_loop_p): Don't version outer loop if it has
34803 dont_vectorized bit set.
34804 (tree_if_conversion): When versioning outer loop, ensure
34805 tree_if_conversion is performed also on the inner loop of the
34806 non-vectorizable outer loop copy.
34807 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34808 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34809 prevent vectorization of it.
34810 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34811 the outer loop vectorization of the non-scalar version is attempted
34812 before vectorization of the inner loop in scalar version. If
34813 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34814 vectorization of its inner loop.
34815 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34816 has 2 inner loops, rename also on edges from bb whose single pred
34817 is outer_loop->header. Fix typo in function comment.
34819 2017-01-09 Martin Sebor <msebor@redhat.com>
34822 * asan.c (asan_emit_stack_protection): Increase local buffer size
34823 to avoid snprintf truncation warning.
34825 2017-01-09 Andrew Pinski <apinski@cavium.com>
34827 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34828 to reference thunderx2t99 for the tuning structure
34829 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34831 (thunderx2t99_extra_costs): This.
34832 * config/aarch64/aarch64-tune.md: Regenerate.
34833 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34834 (vulcan_addrcost_table): This.
34835 (vulcan_regmove_cost): Rename to ...
34836 (thunderx2t99_regmove_cost): This.
34837 (vulcan_vector_cost): Rename to ...
34838 (thunderx2t99_vector_cost): this.
34839 (vulcan_branch_cost): Rename to ...
34840 (thunderx2t99_branch_cost): This.
34841 (vulcan_tunings): Rename to ...
34842 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34843 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34845 2017-01-09 Martin Jambor <mjambor@suse.cz>
34849 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34850 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34851 (propagate_vr_accross_jump_function): Use the above function for all
34852 value range computations for pass-through jump functions and type
34853 converasion from explicit value range values.
34854 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34855 parameters from TYPE_ARG_TYPES.
34856 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34857 (ipa_write_node_info): Stream type of the actual argument.
34858 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34860 2017-01-09 Martin Liska <mliska@suse.cz>
34863 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34864 (lookup_compiler): Do not show error message with have_E.
34866 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34868 PR tree-optimization/78938
34869 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34870 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34871 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34872 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34875 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34877 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34880 2017-01-09 Richard Biener <rguenther@suse.de>
34882 PR tree-optimization/78997
34883 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34884 name condition properly.
34886 2017-01-09 Richard Biener <rguenther@suse.de>
34889 * dwarf2out.c (is_cxx): New overload with context.
34890 (is_naming_typedef_decl): Use it.
34892 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34894 * invoke.texi (Option Summary): Correct spacing in option lists
34895 and add line breaks to fix over-long lines.
34897 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34899 PR middle-end/17660
34901 * extend.texi (Common Variable Attributes): Add xref to GCC
34902 Internals manual to explain mode attribute keywords.
34904 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34907 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34908 and Preprocessor Options.
34909 (Options for Linking): Document -pthread here....
34910 (RS/6000 and PowerPC Options): ...not here.
34911 (Solaris 2 Options): ...or here.
34912 * doc/cppopts.texi: Document -pthread.
34914 2017-01-08 Martin Sebor <msebor@redhat.com>
34916 PR middle-end/77708
34917 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34918 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34919 New member functions.
34920 (format_directive): Used them.
34922 (pass_sprintf_length::handle_gimple_call): Same.
34923 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34924 to avoid truncation for any argument.
34925 (extract_affine_mul): Same.
34926 * tree.c (get_file_function_name): Same.
34928 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34930 PR middle-end/77484
34931 * predict.def (PRED_INDIR_CALL): Set to 86.
34933 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34935 PR preprocessor/54124
34936 * doc/cppopts.texi: Reformat -d subtable to list the full name
34937 of the options. Add cross-reference to the docs for the general
34938 compiler -d options.
34939 * doc/invoke.texi (Developer Options): Add cross-reference to the
34940 preprocessor-specific -d option documentation.
34942 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34944 PR preprocessor/13498
34945 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34946 redudant material, and reflect new command-line options.
34947 (System Headers): Likewise.
34949 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34951 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34952 -isystem, and -idirafter. Copy-edit.
34953 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34954 default for -ftrack-macro-expansion. Delete obsolete and
34955 badly-formatted implementation details about -fdebug-cpp output.
34956 * doc/cppwarnopts.texi: Copy-edit.
34958 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34961 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34962 that the transition from a max line width >= 1<<10 to narrower
34963 lines works correctly.
34965 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34967 * doc/options.texi (PerFunction): New.
34968 * opt-functions.awk (switch_flags): Map both Optimization and
34969 PerFunction to CL_OPTIMIZATION.
34970 * opth-gen.awk: Test for PerFunction flag along with
34972 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34973 it only when the latter is present. Skip those that don't in
34974 the hash function generator.
34975 * common.opt (fvar-tracking): Mark as PerFunction instead of
34977 (fvar-tracking-assignments): Likewise.
34978 (fvar-tracking-assignments-toggle): Likewise.
34979 (fvar-tracking-uninit): Likewise.
34981 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34983 PR translation/79018
34984 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34987 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34990 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34991 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34992 TARGET_LONG_JUMP_TABLE_OFFSETS.
34993 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34994 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34995 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34996 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34997 * config/m68k/m68k.md (tablejump expander): Likewise.
34998 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34999 TARGET_LONG_JUMP_TABLE_OFFSETS.
35000 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35001 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35003 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35004 David Holsgrove <david.holsgrove@xilinx.com>
35006 * common/config/microblaze/microblaze-common.c
35007 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35008 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35010 * config/microblaze/microblaze.c (microblaze_must_save_register)
35011 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35013 (microblaze_eh_return): New function.
35014 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35015 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35016 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35017 * config/microblaze/microblaze.md (eh_return): New pattern.
35019 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35021 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35022 GCC_DIAGNOSTIC_STRINGIFY): Define.
35024 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35026 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35028 * config/arm/arm.md (<mcrr>): New.
35030 * config/arm/arm.c (arm_arch5te): New.
35031 (arm_option_override): Set arm_arch5te.
35032 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35034 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35035 (arm_mcrr_qualifiers): ... this. New.
35036 (MRRC_QUALIFIERS): Define to...
35037 (arm_mrrc_qualifiers): ... this. New.
35038 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35040 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35041 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35042 (MRRCI, mrrc, MRRC): New.
35043 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35044 VUNSPEC_MRRC2): New.
35046 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35048 * config/arm/arm.md (<mcr>): New.
35050 * config/arm/arm.c (arm_coproc_builtin_available): Add
35051 support for mcr, mrc, mcr2 and mrc2.
35052 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35053 (arm_mcr_qualifiers): ... this. New.
35054 (MRC_QUALIFIERS): Define to ...
35055 (arm_mrc_qualifiers): ... this. New.
35056 (MCR_QUALIFIERS): Define to ...
35057 (arm_mcr_qualifiers): ... this. New.
35058 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35060 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35061 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35062 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35063 VUNSPEC_MRC2): New.
35065 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35067 * config/arm/arm.md (*ldc): New.
35071 * config/arm/arm.c (arm_coproc_builtin_available): Add
35072 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35073 (arm_coproc_ldc_stc_legitimate_address): New.
35074 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35075 'qualifier_const_pointer'.
35076 (LDC_QUALIFIERS): Define to...
35077 (arm_ldc_qualifiers): ... this. New.
35078 (STC_QUALIFIERS): Define to...
35079 (arm_stc_qualifiers): ... this. New.
35080 * config/arm/arm-protos.h
35081 (arm_coproc_ldc_stc_legitimate_address): New.
35082 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35083 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35084 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35085 stc2, stcl, stc2l): New.
35086 * config/arm/constraints.md (Uz): New.
35087 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35088 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35089 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35090 VUNSPEC_STC2L): New.
35092 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35094 * config/arm/arm.md (<cdp>): New.
35095 * config/arm/arm.c (neon_const_bounds): Rename this ...
35096 (arm_const_bounds): ... this.
35097 (arm_coproc_builtin_available): New.
35098 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35099 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35100 (CDP_QUALIFIERS): Define to...
35101 (arm_cdp_qualifiers): ... this. New.
35103 (arm_expand_builtin_args): Add case for 6 arguments.
35104 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35105 (arm_const_bounds): ... this.
35106 (arm_coproc_builtin_available): New.
35107 * config/arm/arm_acle.h (__arm_cdp): New.
35109 * config/arm/arm_acle_builtins.def (cdp): New.
35111 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35112 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35113 'arm_const_bounds'.
35114 * config/arm/types.md (coproc): New.
35115 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35116 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35117 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35118 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35120 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35122 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35123 (UBINOP_QUALIFIERS): New.
35125 (acle_builtin_data): New. Change comment.
35126 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35127 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35128 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35129 arm_acle_builtins.def.
35130 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35131 (arm_init_acle_builtins): New.
35132 (CRC32_BUILTIN): Remove.
35133 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35134 crc32cb, crc32ch and crc32cw.
35135 (arm_init_crc32_builtins): Remove.
35136 (arm_init_builtins): Use arm_init_acle_builtins rather
35137 than arm_init_crc32_builtins.
35138 (arm_expand_acle_builtin): New.
35139 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35140 * config/arm/arm_acle_builtins.def: New.
35142 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35144 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35145 (arm_builtin_datum): ... this.
35146 (arm_init_neon_builtin): Rename to ...
35147 (arm_init_builtin): ... this. Add a new parameters PREFIX
35148 and USE_SIG_IN_NAME.
35149 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35150 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35151 'arm_builtin_datum'.
35152 (arm_init_vfp_builtins): Likewise.
35153 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35154 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35155 (arm_expand_neon_args): Rename to ...
35156 (arm_expand_builtin_args): ... this. Rename builtin_arg
35157 enum values and differentiate between ARG_BUILTIN_MEMORY
35158 and ARG_BUILTIN_NEON_MEMORY.
35159 (arm_expand_neon_builtin_1): Rename to ...
35160 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35161 values, arm_expand_builtin_args and add bool parameter NEON.
35162 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35163 (arm_expand_vfp_builtin): Likewise.
35164 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35166 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35168 PR middle-end/77484
35169 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35170 * predict.c (tree_estimate_probability_bb): Reverse direction of
35171 polymorphic call predictor.
35173 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35175 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35176 (determine_pass_name_match): ...this new function and...
35177 (should_skip_pass_p): ...this new function.
35179 2017-01-06 Nathan Sidwell <nathan@acm.org>
35181 * ipa-visibility.c (function_and_variable_visibility): Reformat
35182 comments and long lines. Remove extrneous if.
35183 * symtab.c (symtab_node::make_decl_local): Fix code format.
35184 (symtab_node::set_section_for_node): Fix comment typo.
35186 2017-01-06 Martin Liska <mliska@suse.cz>
35189 * lra-constraints.c: Rename invariant to lra_invariant.
35190 * predict.c (set_even_probabilities): Initialize e to NULL.
35192 2017-01-05 Martin Sebor <msebor@redhat.com>
35194 PR tree-optimization/78910
35195 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35196 (format_integer): Correct off-by-one error in the handling
35197 of precision with negative numbers in signed conversions..
35199 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35201 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35203 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35205 PR tree-optimization/71016
35206 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35207 factor_out_conditional_conversion. Formatting fix.
35208 (factor_out_conditional_conversion): Add cond_stmt argument.
35209 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35210 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35213 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35215 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35216 read-rtl-function.o, and selftest-rtl.o.
35217 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35218 (selftest::aarch64_test_loading_full_dump): New function.
35219 (selftest::aarch64_run_selftests): New function.
35220 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35221 selftest::aarch64_run_selftests.
35222 * config/i386/i386.c
35223 (selftest::ix86_test_loading_dump_fragment_1): New function.
35224 (selftest::ix86_test_loading_call_insn): New function.
35225 (selftest::ix86_test_loading_full_dump): New function.
35226 (selftest::ix86_test_loading_unspec): New function.
35227 (selftest::ix86_run_selftests): Call the new functions.
35228 * emit-rtl.c (maybe_set_max_label_num): New function.
35229 * emit-rtl.h (maybe_set_max_label_num): New decl.
35230 * function.c (instantiate_decls): Guard call to
35231 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35232 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35234 * gensupport.c (gen_reader::gen_reader): Pass "false"
35235 for new "compact" param of rtx_reader.
35236 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35237 rather than an empty string for NULL strings.
35238 * read-md.c: Potentially include config.h rather than bconfig.h.
35239 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35240 (have_error): New global, copied from errors.c.
35241 (md_reader::read_name): Rename to...
35242 (md_reader::read_name_1): ...this, adding "out_loc" param,
35243 and converting "missing name or number" to returning false, rather
35245 (md_reader::read_name): Reimplement in terms of read_name_1.
35246 (md_reader::read_name_or_nil): New function.
35247 (md_reader::read_string): Handle "(nil)" by returning NULL.
35248 (md_reader::md_reader): Add new param "compact".
35249 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35250 (md_reader::read_file): New method.
35251 * read-md.h (md_reader::md_reader): Add new param "compact".
35252 (md_reader::read_file): New method.
35253 (md_reader::is_compact): New accessor.
35254 (md_reader::read_name): Convert return type from void to file_location.
35255 (md_reader::read_name_or_nil): New decl.
35256 (md_reader::read_name_1): New decl.
35257 (md_reader::m_compact): New field.
35258 (noop_reader::noop_reader): Pass "false" for new "compact" param
35260 (rtx_reader::rtx_reader): Add new "compact" param.
35261 (rtx_reader::read_rtx_operand): Make virtual and convert return
35262 type from void to rtx.
35263 (rtx_reader::read_until): New decl.
35264 (rtx_reader::handle_any_trailing_information): New virtual function.
35265 (rtx_reader::postprocess): New virtual function.
35266 (rtx_reader::finalize_string): New virtual function.
35267 (rtx_reader::m_in_call_function_usage): New field.
35268 (rtx_reader::m_reuse_rtx_by_id): New field.
35269 * read-rtl-function.c: New file.
35270 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35271 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35272 (selftest::verify_three_block_rtl_cfg): New decl.
35273 * read-rtl-function.h: New file.
35274 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35275 For host, include function.h, memmodel.h, and emit-rtl.h.
35276 (one_time_initialization): New function.
35277 (struct compact_insn_name): New struct.
35278 (compact_insn_names): New array.
35279 (find_code): Handle insn codes in compact dumps.
35280 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35281 (bind_subst_iter_and_attr): Likewise.
35282 (add_condition_to_string): Likewise.
35283 (add_condition_to_rtx): Likewise.
35284 (apply_attribute_uses): Likewise.
35285 (add_current_iterators): Likewise.
35286 (apply_iterators): Likewise.
35287 (initialize_iterators): Guard usage of apply_subst_iterator with
35288 #ifdef GENERATOR_FILE.
35289 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35290 (md_reader::read_mapping): Likewise.
35291 (add_define_attr_for_define_subst): Likewise.
35292 (add_define_subst_attr): Likewise.
35293 (read_subst_mapping): Likewise.
35294 (check_code_iterator): Likewise.
35295 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35297 (one_time_initialization): New function.
35298 (rtx_reader::read_until): New method.
35299 (read_flags): New function.
35300 (parse_reg_note_name): New function.
35301 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35302 Handle reuse_rtx ids.
35303 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35304 Add parsing support for RTL dumps, mirroring the special-cases in
35305 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35306 values, and calling handle_any_trailing_information.
35307 (rtx_reader::read_rtx_operand): Convert return type from void
35308 to rtx, returning return_rtx. Handle case 'e'. Call
35309 finalize_string on XSTR and XTMPL fields.
35310 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35311 "(nil)" values were omitted. Call the postprocess vfunc on the
35313 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35314 class ctor. Initialize m_in_call_function_usage. Call
35315 one_time_initialization.
35316 * rtl-tests.c (selftest::test_uncond_jump): Call
35317 set_new_first_and_last_insn.
35318 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35319 * selftest-rtl.c: New file.
35320 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35321 (selftest::get_insn_by_uid): New decl.
35322 * selftest-run-tests.c (selftest::run_tests): Call
35323 read_rtl_function_c_tests.
35324 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35325 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35328 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35330 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35331 operands in a special way. Assert that pos+len <= mode precision.
35333 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35335 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35336 3 argument Alias with unlimited for the negative form.
35337 (fno-vect-cost-model): Removed.
35339 2017-01-05 Martin Liska <mliska@suse.cz>
35341 * hsa-gen.c (gen_hsa_divmod): New function.
35342 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35344 2017-01-05 Martin Liska <mliska@suse.cz>
35347 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35350 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35352 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35353 small constant length operands.
35355 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35357 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35358 between loop iterations.
35360 2017-01-05 Martin Liska <mliska@suse.cz>
35363 * gimplify.c (gimplify_decl_expr): Compare to
35364 asan_poisoned_variables instread of checking flags.
35365 (gimplify_target_expr): Likewise.
35366 (gimplify_expr): Likewise.
35367 (gimplify_function_tree): Conditionally initialize
35368 asan_poisoned_variables.
35370 2017-01-04 Jeff Law <law@redhat.com>
35372 PR tree-optimizatin/78812
35373 * rtl.h (contains_mem_rtx_p): Prototype.
35374 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35375 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35376 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35377 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35378 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35380 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35382 * input.c (assert_char_at_range): Default-initialize actual_range.
35384 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35386 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35387 to match the caller.
35389 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35391 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35392 insns after final jump in test to emit dummy move.
35394 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35396 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35397 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35399 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35401 * multiple_target.c (create_dispatcher_calls): Init e_next.
35402 * tree-ssa-loop-split.c (split_loop): Init border.
35403 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35406 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35411 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35412 (altivec_register_operand): Do not return true if the operand
35413 contains a SUBREG mixing SImode and SFmode.
35414 (vsx_register_operand): Likewise.
35415 (vsx_reg_sfsubreg_ok): New predicate.
35416 (vfloat_operand): Do not return true if the operand contains a
35417 SUBREG mixing SImode and SFmode.
35418 (vint_operand): Likewise.
35419 (vlogical_operand): Likewise.
35420 (gpc_reg_operand): Likewise.
35421 (int_reg_operand): Likewise.
35422 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35423 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35424 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35426 (rs6000_emit_move_si_sf_subreg): New helper function.
35427 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35428 fixup SUBREGs involving SImode and SFmode.
35429 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35430 numbers for the new peephole2 optimization.
35431 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35432 the GLIBC math library that do AND/IOR/XOR operations on single
35433 precision floating point.
35434 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35435 target macros to say whether we need to avoid SUBREGs mixing
35437 (TARGET_ALLOW_SF_SUBREG): Likewise.
35438 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35439 (UNSPEC_SI_FROM_SF): Likewise.
35440 (iorxor): Change spacing.
35441 (and_ior_xor): New iterator for AND, IOR, and XOR.
35442 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35443 (movdi_from_sf_zero_ext): Likewise.
35444 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35445 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35446 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35447 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35448 (fms<mode>4): Likewise.
35449 (fnma<mode>4): Likewise.
35450 (fnms<mode>4): Likewise.
35451 (nfma<mode>4): Likewise.
35452 (nfms<mode>4): Likewise.
35454 2017-01-04 Marek Polacek <polacek@redhat.com>
35457 * doc/invoke.texi: Document -Wpointer-compare.
35459 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35461 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35464 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35465 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35468 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35471 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35472 documentation of the powerpc_popcntb_ok attribute.
35473 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35474 code to issue warning messages if a requested CPU configuration is
35475 not supported by the binary (assembler and loader) toolchain.
35476 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35477 made to define a built-in function that has been disabled.
35478 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35479 made to define a built-in function that has been disabled.
35480 (altivec_init_builtins): Add comment explaining why definition
35481 of the DST built-in functions is not preceded by an assertion
35482 check. Add assertions to prevent ICE if attempts are made to
35483 define an altivec predicate or an abs* built-in function that has
35485 (htm_init_builtins): Add comment explaining why definition of the
35486 htm built-in functions is not preceded by an assertion check.
35488 2017-01-04 Jeff Law <law@redhat.com>
35490 PR tree-optimizatin/67955
35491 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35492 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35493 the points-to solution does not include pt_null. Use DECL_PT_UID
35496 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35498 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35499 Use gen_int_mode instead of gen_lopwart for const_int operands.
35501 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35503 PR tree-optimization/71563
35504 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35505 out of range value - has low bits known to be zero.
35507 2017-01-04 Alan Modra <amodra@gmail.com>
35509 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35510 * configure: Regenerate.
35511 * config.in: Regenerate.
35513 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35516 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35517 a substring of the message, but strcmp with the whole message. Ifdef
35518 ENABLE_NLS, translate the message first using dgettext.
35520 2017-01-03 Jeff Law <law@redhat.com>
35522 PR tree-optimizatin/78856
35523 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35524 (mark_threaded_blocks): Remove code to truncate thread paths that
35525 cross multiple loop headers. Instead invalidate the cached loop
35526 iteration information and handle case of a thread path walking
35527 into an irreducible region.
35529 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35532 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35533 assertions. Add support for doing the signbit if the IEEE 128-bit
35534 floating point value is in a GPR.
35535 * config/rs6000/rs6000.md (Fsignbit): Delete.
35536 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35537 Update the length attribute if the value is in a GPR.
35538 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35539 the sign or zero extension instruction, since the value is always 0/1.
35540 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35543 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35544 extracting SImode to a GPR register so that we can generate a
35545 store, limit the vector to be in a traditional Altivec register
35546 for the vextuwrx instruction.
35548 2017-01-03 Ian Lance Taylor <iant@google.com>
35550 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35552 2017-01-03 Martin Sebor <msebor@redhat.com>
35554 PR tree-optimization/78696
35555 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35556 precision. Use MPFR for %f for greater fidelity. Correct handling
35558 (pass_sprintf_length::compute_format_length): Set width and precision
35559 specified by asrerisk to void_node for vararg functions.
35560 (try_substitute_return_value): Adjust dump output.
35562 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35564 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35566 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35568 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35569 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35570 -mlra/-mno-lra was passed to the compiler.
35572 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35574 PR rtl-optimization/65618
35575 * emit-rtl.c (try_split): Move initialization of "before" and
35576 "after" to just before the call to emit_insn_after_setloc.
35578 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35580 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35582 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35584 * dwarf2out.c (gen_enumeration_type_die): When
35585 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35587 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35589 PR tree-optimization/78965
35590 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35591 Change first argument from const call_info & to call_info &. For %n
35592 set info.nowrite to false.
35594 PR middle-end/78901
35595 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35596 possibly throwing calls.
35598 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35599 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35600 and fns handling, rather than in a separate case SSA_NAME.
35602 2017-01-02 Jeff Law <law@redhat.com>
35604 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35605 fixes for first_period and second_period variables.
35607 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35610 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35611 (*insvqi_1): New insn pattern.
35612 (*insvqi_1_mem_rex64): Ditto.
35613 (*insvqi_2): Ditto.
35614 (*insvqi_3): Rename from *insvqi.
35616 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35618 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35620 * doc/cfg.texi (Edges): Remove reference to Java.
35621 (Maintaining the CFG): Ditto.
35623 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35625 PR middle-end/77674
35626 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35627 transparent aliases.
35629 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35631 PR middle-end/77484
35632 * predict.def (PRED_CALL): Update hitrate.
35633 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35634 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35635 into direct/indirect/polymorphic variants.
35637 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35639 Update copyright years.
35641 * gcc.c (process_command): Update copyright notice dates.
35642 * gcov-dump.c (print_version): Ditto.
35643 * gcov.c (print_version): Ditto.
35644 * gcov-tool.c (print_version): Ditto.
35645 * gengtype.c (create_file): Ditto.
35646 * doc/cpp.texi: Bump @copying's copyright year.
35647 * doc/cppinternals.texi: Ditto.
35648 * doc/gcc.texi: Ditto.
35649 * doc/gccint.texi: Ditto.
35650 * doc/gcov.texi: Ditto.
35651 * doc/install.texi: Ditto.
35652 * doc/invoke.texi: Ditto.
35654 Copyright (C) 2017 Free Software Foundation, Inc.
35656 Copying and distribution of this file, with or without modification,
35657 are permitted in any medium without royalty provided the copyright
35658 notice and this notice are preserved.