1 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
5 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6 Change s_operand to memory_operand.
8 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
12 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13 Emit instructions rather than returning an expression. Handle TFmode
14 and KFmode by casting to TImode.
15 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16 (rs6000_emit_le_vsx_store): Likewise.
17 * config/rs6000/vsx.md (VSX_TI): New iterator.
18 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19 (*vsx_le_undo_permute_<mode>): Likewise.
20 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21 emit the split sequence.
22 (*vsx_le_perm_store_<mode>): Likewise.
24 2017-07-27 Jakub Jelinek <jakub@redhat.com>
26 PR tree-optimization/81555
27 PR tree-optimization/81556
28 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
29 if true, force CHANGED for the recursive invocation.
30 (reassociate_bb): Remember original length of ops array, pass
31 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
33 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
34 attributes for noipa attribute. For naked attribute use
35 lookup_attribute first before lookup_attribute_spec.
36 * final.c (rest_of_handle_final): Disable IPA RA for functions with
38 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
39 for functions with noipa attribute.
40 (cgraph_externally_visible_p): Return true for functions with noipa
42 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
43 for functions with noipa attribute.
44 * doc/extend.texi: Document noipa function attribute.
45 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
46 also for functions with noipa attribute.
47 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
49 2017-07-26 Andrew Pinski <apinski@cavium.com>
51 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
52 vec_unalign_load_cost and vec_unalign_store_cost.
54 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
56 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
57 -mvsx-small-integer option.
58 (ISA_3_0_MASKS_IEEE): Likewise.
59 (OTHER_VSX_VECTOR_MASKS): Likewise.
60 (POWERPC_MASKS): Likewise.
61 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
62 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
63 code, only testing for DImode being allowed in non-VSX floating
65 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
66 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
68 (rs6000_option_override_internal): Delete -mvsx-small-integer.
69 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
70 TARGET_P8_VECTOR test.
71 (rs6000_secondary_reload_simple_move): Likewise.
72 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
73 since TARGET_P9_VECTOR was already tested.
74 (rs6000_opt_masks): Remove -mvsx-small-integer.
75 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
76 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
78 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
79 test for TARGET_VEXTRACTUB was used, and that uses
81 (p9 extract splitter): Likewise.
82 (vsx_extract_<mode>_di_p9): Likewise.
83 (vsx_extract_<mode>_store_p9): Likewise.
84 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
85 for TARGET_P9_VECTOR was used. Delete code that is now dead with
86 the elimination of TARGET_VSX_SMALL_INTEGER.
87 (vsx_extract_<mode>_p8): Likewise.
88 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
89 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
90 (vsx_set_<mode>_p9): Likewise.
91 (vsx_set_v4sf_p9): Likewise.
92 (vsx_set_v4sf_p9_zero): Likewise.
93 (vsx_insert_extract_v4sf_p9): Likewise.
94 (vsx_insert_extract_v4sf_p9_2): Likewise.
95 * config/rs6000/rs6000.md (sign extend splitter): Change
96 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
97 (floatsi<mode>2_lfiwax_mem): Likewise.
98 (floatunssi<mode>2_lfiwzx_mem): Likewise.
99 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
100 since a test for TARGET_P9_VECTOR was used.
101 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
102 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
103 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
104 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
105 TARGET_P8_VECTOR test.
106 (fix_trunc<mode>si2_stfiwx): Likewise.
107 (fix_trunc<mode>si2_internal): Likewise.
108 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
109 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
111 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
112 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
113 TARGET_P8_VECTOR test.
114 (fixuns_trunc<mode>si2_stfiwx): Likewise.
115 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
116 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
118 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
119 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
120 since a test for TARGET_P9_VECTOR was used.
121 (splitter for loading small constants): Likewise.
123 2017-07-26 Andrew Pinski <apinski@cavium.com>
125 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
128 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
131 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
132 (fp_valid_at): Likewise.
134 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
136 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
137 (qdf24xx_addrcost_table): Likewise.
138 (cortexa57_tunings): Update to use generic_branch_cost.
139 (cortexa72_tunings): Likewise.
140 (cortexa73_tunings): Likewise.
141 (qdf24xx_tunings): Likewise.
143 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
145 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
146 (thunderx2t99_branch_cost): Likewise.
147 (cortexa35_tunings): Update to use generic_branch_cost.
148 (cortexa53_tunings): Likewise.
149 (cortexa57_tunings): Likewise.
150 (cortexa72_tunings): Likewise.
151 (cortexa73_tunings): Likewise.
152 (thunderx2t99_tunings): Likewise.
154 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
156 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
157 (sparc_option_override): Honour MASK_FSMULD.
158 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
159 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
160 * config/sparc/sparc.opt (mfsmuld): New option.
161 * doc/invoke.texi (mfsmuld): Document option.
163 2017-07-26 Marek Polacek <polacek@redhat.com>
166 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
168 2017-07-26 Richard Biener <rguenther@suse.de>
170 * gimple-match-head.c (do_valueize): Return OP if valueize
172 (get_def): New helper to get at the def stmt of a SSA name
174 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
175 do_valueize to get at the def stmt.
176 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
178 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
181 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
183 2017-07-26 Martin Liska <mliska@suse.cz>
186 * function.c (expand_function_start): Make expansion of
187 nonlocal_goto_save_area after parm_birth_insn.
189 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
191 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
192 from all CPU target flags enable members.
194 2017-07-26 Richard Biener <rguenther@suse.de>
196 * genmatch.c (dt_simplify::gen): Make iterator vars const.
197 (decision_tree::gen): Make 'type' const.
198 (write_predicate): Likewise.
200 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
202 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
203 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
204 (rs6000_option_override_internal): Likewise.
205 (rs6000_expand_vector_set): Likewise.
206 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
207 (TARGET_UPPER_REGS_SF): Likewise.
208 (TARGET_UPPER_REGS_DI): Likewise.
209 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
210 (TARGET_DIRECT_MOVE_64BIT): Likewise.
211 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
212 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
213 (Splitters for DI constants in Altivec registers): Likewise.
214 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
215 (vsx_set_v4sf_p9): Likewise.
216 (vsx_set_v4sf_p9_zero): Likewise.
217 (vsx_insert_extract_v4sf_p9): Likewise.
218 (vsx_insert_extract_v4sf_p9_2): Likewise.
220 2017-07-25 Carl Love <cel@us.ibm.com>
222 * doc/extend.texi: Update the built-in documentation file for the
223 existing built-in functions
224 vector signed char vec_cnttz (vector signed char);
225 vector unsigned char vec_cnttz (vector unsigned char);
226 vector signed short vec_cnttz (vector signed short);
227 vector unsigned short vec_cnttz (vector unsigned short);
228 vector signed int vec_cnttz (vector signed int);
229 vector unsigned int vec_cnttz (vector unsigned int);
230 vector signed long long vec_cnttz (vector signed long long);
231 vector unsigned long long vec_cnttz (vector unsigned long long);
233 2017-07-25 Andrew Pinski <apinski@cavium.com>
235 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
236 accesses where the use is for the first operand of a BIT_INSERT.
238 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
241 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
242 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
244 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
246 * config/i386/gstabs.h: Delete.
247 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
249 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
251 * config/i386/i386.c (ix86_decompose_address): Do not check for
252 register RTX when looking at index_reg or base_reg.
253 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
255 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
257 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
258 to update EH info here.
260 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
262 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
264 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
266 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
268 2017-07-25 Torsten Duwe <duwe@suse.de>
270 * common.opt: Introduce -fpatchable-function-entry
271 command line option, and its variables function_entry_patch_area_size
272 and function_entry_patch_area_start.
273 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
274 including a two-value parser.
275 * target.def (print_patchable_function_entry): New target hook.
276 * targhooks.h (default_print_patchable_function_entry): New function.
277 * targhooks.c (default_print_patchable_function_entry): Likewise.
278 * toplev.c (process_options): Switch off IPA-RA if
279 patchable function entries are being generated.
280 * varasm.c (assemble_start_function): Look at the
281 patchable-function-entry command line switch and current
282 function attributes and maybe generate NOP instructions by
283 calling the print_patchable_function_entry hook.
284 * doc/extend.texi: Document patchable_function_entry attribute.
285 * doc/invoke.texi: Document -fpatchable_function_entry
287 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
289 * doc/tm.texi: Re-generate.
291 2017-07-25 Jakub Jelinek <jakub@redhat.com>
294 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
295 TARGET_AVX512DQ rather than TARGET_AVX512BW.
297 2017-07-25 Tamar Christina <tamar.christina@arm.com>
299 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
301 2017-07-25 Richard Biener <rguenther@suse.de>
303 PR tree-optimization/81455
304 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
305 not walk in cycles when looking for guards.
307 2017-07-25 Richard Biener <rguenther@suse.de>
309 PR tree-optimization/81529
310 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
311 when optimizing backedge uses.
313 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
315 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
317 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
318 to dl_section_ref. On AIX, append an expression to subtract
319 the size of the section length to dl_section_ref.
321 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
323 * configure.ac: If any of the config.* scripts fail, exit 1.
324 * configure: Regenerate.
326 2017-07-25 Richard Biener <rguenther@suse.de>
329 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
330 of immediate uses, be more verbose on errors.
332 2017-07-25 Richard Biener <rguenther@suse.de>
334 PR tree-optimization/81510
335 * tree-vect-loop.c (vect_is_simple_reduction): When the
336 reduction stmt is not inside the loop bail out.
338 2017-07-25 Richard Biener <rguenther@suse.de>
340 PR tree-optimization/81303
341 * tree-vect-loop-manip.c (vect_loop_versioning): Build
342 profitability check against LOOP_VINFO_NITERSM1.
344 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
346 * domwalk.c (cmp_bb_postorder): Simplify.
347 (sort_bbs_postorder): New function. Use it...
348 (dom_walker::walk): ...here to optimize common cases.
350 2017-07-25 Martin Liska <mliska@suse.cz>
353 * ipa-visibility.c (function_and_variable_visibility): Make the
354 redirection just on target that supports aliasing.
355 Fix GNU coding style.
357 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
360 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
362 * config/arm/rtems.h: Likewise.
363 * config/bfin/rtems.h: Likewise.
364 * config/i386/rtemself.h: Likewise.
365 * config/lm32/rtems.h: Likewise.
366 * config/m32c/rtems.h: Likewise.
367 * config/m68k/rtemself.h: Likewise.
368 * config/microblaze/rtems.h: Likewise.
369 * config/mips/rtems.h: Likewise.
370 * config/moxie/rtems.h: Likewise.
371 * config/nios2/rtems.h: Likewise.
372 * config/powerpcspe/rtems.h: Likewise.
373 * config/rs6000/rtems.h: Likewise.
374 * config/rtems.h: Likewise.
375 * config/sh/rtems.h: Likewise.
376 * config/sh/rtemself.h: Likewise.
377 * config/sparc/rtemself.h: Likewise.
379 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
382 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
383 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
384 * tree-ssa-structalias.c (alias_get_name): Same.
386 2017-07-25 Bin Cheng <bin.cheng@arm.com>
389 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
390 instructions if no du chain is found.
392 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
394 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
396 2017-07-25 Richard Biener <rguenther@suse.de>
399 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
402 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
404 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
406 (ISA_2_7_MASKS_SERVER): Likewise.
407 (ISA_3_0_MASKS_IEEE): Likewise.
408 (OTHER_P8_VECTOR_MASKS): Likewise.
409 (OTHER_VSX_VECTOR_MASKS): Likewise.
410 (POWERPC_MASKS): Likewise.
411 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
412 duplicate list of options.
413 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
414 explicit -mupper-regs options.
415 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
416 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
417 alias for -mupper-regs-df.
418 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
419 (rs6000_init_hard_regno_mode_ok): Likewise.
420 (rs6000_option_override_internal): Likewise.
421 (rs6000_opt_masks): Likewise.
422 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
423 options in terms of whether -mvsx or -mpower8-vector was used.
424 (TARGET_UPPER_REGS_DI): Likewise.
425 (TARGET_UPPER_REGS_SF): Likewise.
426 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
427 -mupper-regs-* options.
429 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
431 * passes.c (emergency_dump_function): Print some empty lines and a
432 header before the RTL dump.
434 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
436 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
438 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
441 * config/aarch64/aarch64.c (aarch64_classify_symbol):
442 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
444 2017-07-24 Carl Love <cel@us.ibm.com>
446 * config/rs6000/rs6000-c.c: Add support for built-in functions
447 vector float vec_extract_fp32_from_shorth (vector unsigned short);
448 vector float vec_extract_fp32_from_shortl (vector unsigned short);
449 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
450 vec_extract_fp_from_shortl): Add defines for the two builtins.
451 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
452 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
454 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
455 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
456 * doc/extend.texi: Update the built-in documentation file for the
457 new built-in function.
459 2017-07-24 Jakub Jelinek <jakub@redhat.com>
462 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
464 * doc/generic.texi: Likewise.
465 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
466 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
468 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
470 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
471 (aarch64_mls_elt_merge<mode>): Likewise.
473 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
475 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
478 2017-07-23 Michael Collison <michael.collison@arm.com>
480 * config/arm/arm.c (arm_option_override): Deprecate
481 use of -mstructure-size-boundary.
482 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
483 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
485 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
488 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
489 Reduce cost estimate for direct moves.
491 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
494 * config/i386/i386.c (ix86_option_override_internal): Disable
495 BMI, BMI2 and TBM instructions for -m16.
497 2017-07-21 Carl Love <cel@us.ibm.com>
499 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
500 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
501 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
502 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
503 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
504 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
505 VMULOSW): New enum "unspec" values.
506 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
507 altivec_vmulosw): New patterns.
508 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
509 VMULOSW): Add definitions.
511 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
513 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
515 * config/aarch64/aarch64-options-extensions.def (rdma); New.
516 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
517 (AARCH64_FL_V8_1): Renumber.
518 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
519 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
520 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
521 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
522 rdma to feature modifiers list.
524 2017-07-21 Yury Gribov <tetra2005@gmail.com>
527 * ipa-visibility (function_and_variable_visibility): Convert
528 recursive PLT call to direct call if appropriate.
530 2017-07-21 Andrew Pinski <apinski@cavium.com>
532 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
533 operand 1 to see if the types precision matches.
534 * fold-const.c (operand_equal_p): Likewise.
536 2017-07-21 Richard Biener <rguenther@suse.de>
538 PR tree-optimization/81303
539 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
540 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
541 (vect_peeling_hash_get_lowest_cost): Adjust.
542 (vect_enhance_data_refs_alignment): Likewise. Use
543 vect_get_peeling_costs_all_drs to compute the penalty for no
544 peeling to match up costs.
546 2017-07-21 Richard Biener <rguenther@suse.de>
548 PR tree-optimization/81500
549 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
550 we didn't identify a reduction path.
552 2017-07-21 Tom de Vries <tom@codesourcery.com>
553 Cesar Philippidis <cesar@codesourcery.com>
555 PR gcov-profile/81442
556 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
559 2017-07-21 Tom de Vries <tom@codesourcery.com>
562 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
564 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
565 nvptx_override_options_after_change.
567 2017-07-21 Ulrich Drepper <drepper@redhat.com>
569 * dwarf2out.c (output_file_names): Avoid double testing for
572 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
574 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
576 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
578 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
580 (try_crossjump_to_edge): Do not punt on partitioned functions.
582 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
584 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
585 Put all BBs reachable only via paths crossing cold region to cold
587 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
589 2016-07-21 Richard Biener <rguenther@suse.de>
591 PR tree-optimization/81303
592 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
593 into account prologue and epilogue iterations when raising
594 min_profitable_iters to sth at least covering one vector iteration.
596 2017-07-21 Tamar Christina <tamar.christina@arm.com>
598 * config/arm/arm.c (arm_test_cpu_arch_dat):
601 2017-07-20 Nathan Sidwell <nathan@acm.org>
604 * tree.h (TYPE_METHODS): Delete.
605 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
606 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
607 (dbxout_type_methods): Scan TYPE_FIELDS.
608 (dbxout_type): Don't check TYPE_METHODS here.
609 * function.c (use_register_for_decl): Always ignore register for
610 class types when not optimizing.
611 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
612 * tree.c (free_lang_data_in_type): Stitch out member functions and
613 templates from TYPE_FIELDS.
614 (build_distinct_type_copy, verify_type_variant,
615 verify_type): Member fns are on TYPE_FIELDS.
616 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
617 * tree-pretty-print.c (dump_generic_node): Likewise.
619 2017-07-20 Jakub Jelinek <jakub@redhat.com>
622 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
623 V2TImode and V4TImode.
624 (ix86_expand_vector_extract): Likewise.
625 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
626 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
627 (ssescalarmode): Handle V4TImode and V2TImode.
628 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
629 (*vec_extractv2ti, *vec_extractv4ti): New insns.
630 (VEXTRACTI128_MODE): New mode iterator.
631 (splitter for *vec_extractv?ti first element): New.
632 (VEC_INIT_MODE): New mode iterator.
633 (vec_init<mode>): Consolidate 3 expanders into one using
634 VEC_INIT_MODE mode iterator.
636 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
638 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
639 non_spilled_static_chain_regno_p.
641 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
643 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
645 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
647 * bb-reorder.c (connect_traces): Allow copying of blocks within
650 2017-07-20 Richard Biener <rguenther@suse.de>
652 * gimple.h (gimple_phi_result): Add gphi * overload.
653 (gimple_phi_result_ptr): Likewise.
654 (gimple_phi_arg): Likewise. Adjust index assert to only
655 allow actual argument accesses rather than all slots available
657 (gimple_phi_arg_def): Add gphi * overload.
658 * tree-phinodes.c (make_phi_node): Initialize only actual
660 (resize_phi_node): Clear memory not covered by old node,
661 do not initialize excess argument slots.
662 (reserve_phi_args_for_new_edge): Initialize new argument slot
665 2017-07-20 Bin Cheng <bin.cheng@arm.com>
667 PR tree-optimization/81388
669 2016-07-21 Bin Cheng <bin.cheng@arm.com>
671 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
672 by removing computation of may_be_zero.
674 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
675 Tom de Vries <tom@codesourcery.com>
678 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
679 when gimple level profile disagrees with what RTL expander did.
681 2017-07-20 Richard Biener <rguenther@suse.de>
683 PR tree-optimization/61171
684 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
685 (vect_analyze_stmt): Add slp instance parameter.
686 (vectorizable_reduction): Likewise.
687 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
688 (vect_is_simple_reduction): Deal with chains not detected
689 as SLP reduction chain, specifically not properly associated
690 chains containing a mix of plus/minus.
691 (get_reduction_op): Remove.
692 (get_initial_defs_for_reduction): Simplify, pass in whether
693 this is a reduction chain, pass in the SLP node for the PHIs.
694 (vect_create_epilog_for_reduction): Get the SLP instance as
696 (vectorizable_reduction): Get the SLP instance as arg.
697 During analysis remember the SLP node with the PHIs in the
698 instance. Simplify getting at the vectorized reduction PHIs.
699 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
700 through SLP instance.
701 (vect_slp_analyze_operations): Likewise.
702 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
703 (vect_transform_stmt): Likewise.
705 2017-07-20 Tom de Vries <tom@codesourcery.com>
707 PR tree-optimization/81489
708 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
709 read of phi arg location to before loop that modifies phi.
711 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
713 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
716 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
719 * except.c (execute): Fix ordering issue.
721 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
723 PR rtl-optimization/81423
724 * combine.c (make_compound_operation_int): Don't try to optimize
725 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
727 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
729 PR rtl-optimization/81423
730 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
731 with a constant that is -1 in the truncated to mode.
733 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
735 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
736 (determine_unlikely_bbs): ... here.
737 * predict.h (propagate_unlikely_bbs_forward): Declare.
738 * cfgexpand.c (pass_expand::execute): Use it.
739 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
741 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
742 propagate_unlikely_bbs_forward.
744 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
747 * except.c (maybe_add_nop_after_section_switch): New function.
750 2017-07-19 Tom de Vries <tom@codesourcery.com>
752 * gimple.h (gimple_phi_set_arg): Make assert more strict.
754 2017-07-19 Tom de Vries <tom@codesourcery.com>
756 * gimple.h (gimple_phi_arg): Make assert more strict.
758 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
760 * config.gcc (powerpc*-*-*): Add mmintrin.h.
761 * config/rs6000/mmintrin.h: New file.
762 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
764 2017-07-19 Jakub Jelinek <jakub@redhat.com>
766 PR tree-optimization/81346
767 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
769 2017-07-19 Tom de Vries <tom@codesourcery.com>
771 * config/nvptx/nvptx.md (VECIM): Add V2DI.
773 2017-07-19 Tom de Vries <tom@codesourcery.com>
775 * config/nvptx/nvptx-modes.def: Add V2DImode.
776 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
777 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
778 (nvptx_output_mov_insn): Handle lack of mov.b128.
779 (nvptx_print_operand): Handle 'H' and 'L' codes.
780 (nvptx_vector_mode_supported): Allow V2DImode.
781 (nvptx_preferred_simd_mode): New function.
782 (nvptx_data_alignment): New function.
783 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
784 nvptx_preferred_simd_mode.
785 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
787 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
789 2017-07-19 Tom de Vries <tom@codesourcery.com>
791 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
792 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
793 (nvptx_vector_mode_supported): New function. Allow V2SImode.
794 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
795 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
796 (mov<VECIM>_insn): New define_insn.
797 (define_expand "mov<VECIM>): New define_expand.
799 2017-07-19 Tom de Vries <tom@codesourcery.com>
801 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
803 2017-07-19 Jakub Jelinek <jakub@redhat.com>
805 PR tree-optimization/81346
806 * fold-const.h (fold_div_compare, range_check_type): Declare.
807 * fold-const.c (range_check_type): New function.
808 (build_range_check): Use range_check_type.
809 (fold_div_compare): No longer static, rewritten into
810 a match.pd helper function.
811 (fold_comparison): Don't call fold_div_compare here.
812 * match.pd (X / C1 op C2): New optimization using fold_div_compare
815 2017-07-19 Nathan Sidwell <nathan@acm.org>
817 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
818 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
819 * tree.c (find_decls_types_r, verify_type): Use
820 TYPE_{MIN,MAX}_VALUE_RAW.
821 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
822 (hash_tree): Likewise.
823 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
825 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
828 2017-07-18 Tom de Vries <tom@codesourcery.com>
831 * omp-expand.c (expand_omp_for_static_chunk): Handle
832 equal-argument loop exit phi.
834 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
837 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
838 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
840 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
842 (ror,rol -> rorx splitters): Use const_int_operand as
845 2017-06-18 Richard Biener <rguenther@suse.de>
847 PR tree-optimization/81410
848 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
849 the gap in the ! slp_perm SLP case after each group.
851 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
854 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
857 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
860 * predict.c (set_even_probabilities): Cleanup; do not affect
861 probabilities that are already known.
862 (combine_predictions_for_bb): Call even when count is set.
864 2017-07-18 Nathan Sidwell <nathan@acm.org>
866 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
869 2017-07-18 Bin Cheng <bin.cheng@arm.com>
872 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
873 optimization for loop niter analysis.
875 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
878 * config/avr/avr.c (avr_optimize_casesi): Don't use
879 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
881 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
883 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
884 body_cost_vec from _vect_peel_extended_info.
885 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
886 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
889 2017-07-18 Bin Cheng <bin.cheng@arm.com>
891 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
893 2017-07-18 Richard Biener <rguenther@suse.de>
895 PR tree-optimization/80620
896 PR tree-optimization/81403
897 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
898 info when re-using a VN table entry.
900 2017-07-18 Richard Biener <rguenther@suse.de>
902 PR tree-optimization/81418
903 * tree-vect-loop.c (vectorizable_reduction): Properly compute
904 vectype_in. Verify that with lane-reducing reduction operations
905 we have a single def-use cycle.
907 2017-07-17 Carl Love <cel@us.ibm.com>
909 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
911 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
912 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
913 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
914 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
915 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
916 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
917 VMULOSW): New enum "unspec" values.
918 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
919 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
920 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
921 altivec_vmulosw): New patterns.
922 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
923 VMULOSW): Add definitions.
924 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
926 * config/alpha/alpha.c: Include predict.h.
928 2017-07-17 Yury Gribov <tetra2005@gmail.com>
930 * tree-vrp.c (compare_assert_loc): Fix comparison function
931 to return predictable results.
933 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
935 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
938 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
939 * doc/invoke.texi (mexpand-adddi): Update text.
941 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
943 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
944 that also clobbers the CC register. The old expand code is moved
946 (*arc_clzsi2): ... here.
947 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
948 the CC register. The old expand code is moved to ...
949 (arc_ctzsi2): ... here.
951 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
953 * config/arc/arc.opt (mindexed-loads): Use initial value
954 TARGET_INDEXED_LOADS_DEFAULT.
955 (mauto-modify-reg): Use initial value
956 TARGET_AUTO_MODIFY_REG_DEFAULT.
957 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
958 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
959 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
960 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
962 2017-07-17 Martin Liska <mliska@suse.cz>
965 * opts.c (finish_options): Do not allow -fgnu-tm
966 w/ -fsanitize={kernel-,}address. Say sorry.
968 2017-07-17 Bin Cheng <bin.cheng@arm.com>
971 * tree-loop-distribution.c (classify_partition): Only assert on
973 (merge_dep_scc_partitions): Delete prameter. Update function call.
974 (distribute_loop): Remove code handling loop with unknown niters.
975 (pass_loop_distribution::execute): Skip loop with unknown niters.
977 2017-07-17 Bin Cheng <bin.cheng@arm.com>
980 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
981 function sort_partitions_by_post_order.
983 2017-07-17 Bin Cheng <bin.cheng@arm.com>
985 PR tree-optimization/81374
986 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
987 the max index of basic blocks, rather than number of basic blocks.
989 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
991 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
993 (arc_legitimate_pic_operand_p): Likewise.
994 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
996 (arc_needs_pcl_p): Likewise.
997 (arc_legitimate_pc_offset_p): Likewise.
998 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
999 function is also used in constrains.md.
1000 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1001 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1002 PLUS. Only return true/false in known cases, otherwise assert.
1003 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1004 is already called in arc_legitimate_constant_p.
1005 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1007 (LEGITIMATE_PIC_OPERAND_P): Use
1008 arc_raw_symbolic_reference_mentioned_p function.
1009 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1014 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1015 Andrew Burgess <andrew.burgess@embecosm.com>
1017 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1018 (arc_return_address_register): New function.
1019 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1020 (arc_handle_fndecl_attribute): Add naked attribute.
1021 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1022 (TARGET_WARN_FUNC_RETURN): Likewise.
1023 (arc_allocate_stack_slots_for_args): New function.
1024 (arc_warn_func_return): Likewise.
1025 (machine_function): Change type fn_type.
1026 (arc_compute_function_type): Consider new naked function type,
1027 change function return type.
1028 (arc_must_save_register): Adapt to handle new
1029 arc_compute_function_type's return type.
1030 (arc_expand_prologue): Likewise.
1031 (arc_expand_epilogue): Likewise.
1032 (arc_return_address_regs): Delete.
1033 (arc_return_address_register): New function.
1034 (arc_epilogue_uses): Use above function.
1035 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1036 (arc_function_type): Change encoding, add naked type.
1037 (ARC_INTERRUPT_P): Change to handle the new encoding.
1038 (ARC_FAST_INTERRUPT_P): Likewise.
1039 (ARC_NORMAL_P): Define.
1040 (ARC_NAKED_P): Likewise.
1041 (arc_compute_function_type): Delete prototype.
1042 * config/arc/arc.md (in_ret_delay_slot): Use
1043 arc_return_address_register function.
1044 (simple_return): Likewise.
1045 (p_return_i): Likewise.
1047 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1049 PR tree-optimization/81428
1050 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1051 can't be built for those types.
1053 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1055 Remove stuff dead since r239246.
1057 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1058 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1059 (avr_inform_devices): Remove dead stuff.
1061 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1063 * config/arm/arm_neon.h: Fix softp typo.
1065 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1067 PR tree-optimization/81365
1068 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1069 aggregate moves onto bb predecessor edges, make sure there are no
1070 loads that could alias the lhs in between the start of bb and the
1073 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1076 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1077 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1078 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1080 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1082 PR tree-optimization/81396
1083 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1084 (init_symbolic_number): Initialize it to 1.
1085 (perform_symbolic_merge): Add n_ops from both operands into the new
1087 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1088 without base_addr as useless if they need more than one operation.
1089 (bswap_replace): Handle !bswap case for NULL base_addr.
1091 2017-07-17 Tom de Vries <tom@codesourcery.com>
1094 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1097 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1099 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1100 conditional builtin define __FIX_LEON3FT_B2BST.
1102 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1104 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1105 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1108 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1110 PR rtl-optimization/81424
1111 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1112 to remove potential trapping from operands if -fnon-call-exceptions.
1114 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1116 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1117 profile_proability for scalling.
1118 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1120 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1122 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1124 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1126 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1127 fixpoint arithmetics.
1129 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1131 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1132 fixpoint arithmetics.
1134 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1136 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1137 fixpoint arithmetics.
1139 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1141 * profile-count.h (profile_probability::from_reg_br_prob_note,
1142 profile_probability::to_reg_br_prob_note): New functions.
1143 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1144 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1145 * predict.c (probability_reliable_p): Update.
1146 (edge_probability_reliable_p): Update.
1147 (br_prob_note_reliable_p): Update.
1148 (invert_br_probabilities): Update.
1149 (add_reg_br_prob_note): New function.
1150 (combine_predictions_for_insn): Update.
1151 * asan.c (asan_clear_shadow): Update.
1152 * cfgbuild.c (compute_outgoing_frequencies): Update.
1153 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1154 (update_br_prob_note): Update.
1155 (rtl_verify_edges): Update.
1156 (purge_dead_edges): Update.
1157 (fixup_reorder_chain): Update.
1158 * emit-rtl.c (try_split): Update.
1159 * ifcvt.c (cond_exec_process_insns): Update.
1160 (cond_exec_process_if_block): Update.
1161 (dead_or_predicable): Update.
1162 * internal-fn.c (expand_addsub_overflow): Update.
1163 (expand_neg_overflow): Update.
1164 (expand_mul_overflow): Update.
1165 * loop-doloop.c (doloop_modify): Update.
1166 * loop-unroll.c (compare_and_jump_seq): Update.
1167 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1168 * predict.h: Update.
1169 * reorg.c (mostly_true_jump): Update.
1171 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1172 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1173 * config/arc/arc.c: (emit_unlikely_jump): Update.
1174 * config/arm/arm.c: (emit_unlikely_jump): Update.
1175 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1176 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1177 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1178 (ix86_print_operand): Update.
1179 (ix86_split_fp_branch): Update.
1180 (predict_jump): Update.
1181 * config/ia64/ia64.c (ia64_print_operand): Update.
1182 * config/mmix/mmix.c (mmix_print_operand): Update.
1183 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1184 (rs6000_expand_split_stack_prologue): Update.
1185 * config/rs6000/rs6000.c: Update.
1186 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1187 (s390_expand_vec_movstr): Update.
1188 (s390_expand_cs_tdsi): Update.
1189 (s390_expand_split_stack_prologue): Update.
1190 * config/sh/sh.c (sh_print_operand): Update.
1191 (expand_cbranchsi4): Update.
1192 (expand_cbranchdi4): Update.
1193 * config/sparc/sparc.c (output_v9branch): Update.
1194 * config/spu/spu.c (get_branch_target): Update.
1195 (ea_load_store_inline): Update.
1196 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1197 * config/tilepro/tilepro.c: Update.
1199 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1201 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1202 (gimplify_save_expr): Likewise.
1204 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1206 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1208 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1210 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1212 * timevar.def (TV_IPA_FNSUMMARY): Define.
1214 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1216 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1217 to back store errata sensitive sequence from being generated.
1218 (sqrtdf2_fix): Likewise.
1220 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1222 * tree-ssa-threadupdate.c (compute_path_counts,
1223 update_joiner_offpath_counts): Use profile_probability.
1225 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1228 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1230 * config/arm/arm-c.c (arm_cpu_builtins): Define
1231 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1233 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1235 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1236 array entries to represent __ieee128 versions of the
1237 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1238 scalar_extract_sig, and scalar_insert_exp built-in functions.
1239 (altivec_resolve_overloaded_builtin): Add special case handling
1240 for the __builtin_scalar_insert_exp function, as represented by
1241 the P9V_BUILTIN_VEC_VSIEDP constant.
1242 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1243 exponent support for __ieee128 argument.
1244 (VSESQP): Add scalar extract signature support for __ieee128
1246 (VSTDCNQP): Add scalar test negative support for __ieee128
1248 (VSIEQP): Add scalar insert exponent support for __int128 argument
1249 with __ieee128 result.
1250 (VSIEQPF): Add scalar insert exponent support for __ieee128
1251 argument with __ieee128 result.
1252 (VSTDCQP): Add scalar test data class support for __ieee128
1254 (VSTDCNQP): Add overload support for scalar test negative with
1256 (VSTDCQP): Add overload support for scalar test data class
1258 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1260 (UNSPEC_VSX_SIEXPQP): New constant.
1261 (xsxexpqp): New insn for VSX scalar extract exponent quad
1263 (xsxsigqp): New insn for VSX scalar extract significand quad
1265 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1266 precision with floating point argument.
1267 (xststdcqp): New expand for VSX scalar test data class quad
1269 (xststdcnegqp): New expand for VSX scalar test negative quad
1271 (xststdcqp): New insn to match expansions for VSX scalar test data
1272 class quad precision and VSX scalar test negative quad precision.
1273 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1274 special case operand checking to enforce that second operand of
1275 VSX scalar test data class with quad precision argument is a 7-bit
1277 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1278 prototypes and descriptions of __ieee128 versions of
1279 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1280 scalar_test_data_class, and scalar_test_neg built-in functions.
1282 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1284 PR tree-optimization/81162
1285 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1286 replace a negate with an add.
1288 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1290 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1292 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1294 * config/arm/arm-c.c (arm_cpu_builtins): Define
1295 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1297 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1299 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1300 (armv8-r): Set ARM Cortex-R52 as default CPU.
1301 * config/arm/arm-tables.opt: Regenerate.
1302 * config/arm/arm-tune.md: Regenerate.
1303 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1305 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1306 extension for -mcpu=cortex-r52.
1308 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1310 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1311 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1312 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1313 (fp-armv8): Define it as FP_ARMv8 only.
1314 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1315 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1317 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1318 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1319 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1320 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1321 than TARGET_FPU_ARMV8.
1322 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1323 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1324 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1326 * config/arm/neon.md (neon_vrint): Likewise.
1327 (neon_vcvt): Likewise.
1328 (neon_<fmaxmin_op><mode>): Likewise.
1329 (<fmaxmin><mode>3): Likewise.
1330 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1331 * config/arm/predicates.md (arm_cond_move_operator): Check against
1332 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1334 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1336 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1339 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1341 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1342 loop in comment with memset.
1344 2017-07-14 Martin Liska <mliska@suse.cz>
1346 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1347 * dwarf2out.c (is_java): Remove the function.
1348 (output_pubname): Remove usage of the function.
1349 (lower_bound_default): Remove usage of DW_LANG_Java.
1350 (gen_compile_unit_die): Likewise.
1351 * gcc.c: Remove compiler defaults for .java and .zip files.
1352 * gimple-expr.c (remove_suffix): Change as there's no longer
1353 extension than 4-letter one.
1354 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1355 (gimplify_save_expr): Likewise.
1356 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1357 as it's possible even for other languages than Java.
1358 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1359 * lto-opts.c (lto_write_options): Remove reference to Java.
1360 * opts.c (strip_off_ending): Update file extension handling.
1361 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1362 * tree-eh.c (lower_resx): Likewise.
1363 * tree.c (free_lang_data_in_type): Remove dead code.
1364 (find_decls_types_r): Likewise.
1365 (build_common_builtin_nodes): Remove Java from a comment.
1366 (verify_type): Remove dead code.
1367 * varasm.c (assemble_external): Remove Java from a comment.
1369 2017-07-14 Martin Liska <mliska@suse.cz>
1371 * opts.c (finish_options): Add quotes.
1372 (common_handle_option): Likewise.
1374 2017-07-14 Martin Liska <mliska@suse.cz>
1376 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1377 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1379 * dwarf2out.c (lower_bound_default): Do not handle
1381 (gen_compile_unit_die): Likewise.
1382 * gcc.c: Remove default extension binding for GNU Pascal.
1383 * stmt.c: Remove Pascal language from a comment.
1384 * xcoffout.c: Likewise.
1386 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1389 * diagnostic-show-locus.c (fixit_cmp): New function.
1390 (layout::layout): Sort m_fixit_hints.
1391 (column_range::column_range): Assert that the values are valid.
1392 (struct char_span): New struct.
1393 (correction::overwrite): New method.
1394 (struct source_line): New struct.
1395 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1396 calls in terms of classes source_line and char_span, and
1397 correction::overwrite.
1398 (selftest::test_overlapped_fixit_printing_2): New function.
1399 (selftest::diagnostic_show_locus_c_tests): Call it.
1401 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1403 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1404 early if there is no lhs.
1406 2017-07-13 Martin Liska <mliska@suse.cz>
1408 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1409 (gen_reference_type_die): Likewise.
1410 * stor-layout.c: Remove Pascal-related comment.
1412 2017-07-13 Martin Liska <mliska@suse.cz>
1414 * opts.c (finish_options): Add quotes to error messages.
1415 (parse_sanitizer_options): Likewise.
1417 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1419 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1421 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1423 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1425 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1427 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1429 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1431 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1434 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1435 provides the hardware capability bits, define the macro
1436 __BUILTIN_CPU_SUPPORTS__.
1437 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1438 if GLIBC does not provide the hardware capability bits. Add a
1439 gcc_unreachable call if the built-in cpu function is neither
1440 __builtin_cpu_is nor __builtin_cpu_supports.
1441 (rs6000_get_function_versions_dispatcher): Change the warning
1442 that an old GLIBC is used which does not export the capability
1443 bits to be an error.
1444 * doc/extend.texi (target_clones attribute): Document the
1445 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1446 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1447 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1448 the macros defined by GCC if the newer GLIBC is available.
1450 2017-07-12 Jeff Law <law@redhat.com>
1452 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1453 remaining includes slightly.
1454 * config/riscv/riscv-builtins.c: Include profile-count.h.
1456 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1459 * config/avr/avr.c (avr_set_current_function): In diagnostic
1460 messages: Quote keywords and (parts of) identifiers.
1461 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1464 2017-07-12 Carl Love <cel@us.ibm.com>
1466 * config/rs6000/rs6000-c.c: Add support for built-in functions
1467 vector bool char vec_revb (vector bool char);
1468 vector bool short vec_revb (vector short char);
1469 vector bool int vec_revb (vector bool int);
1470 vector bool long long vec_revb (vector bool long long);
1471 * doc/extend.texi: Update the built-in documentation file for the
1472 new built-in functions.
1474 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1476 * config/s390/s390.md: Remove movcc splitter.
1478 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1480 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1481 load/store on condition.
1483 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1486 * config/avr/avr.c (avr_encode_section_info)
1487 [progmem && !TREE_READONLY]: Error if progmem object needs
1490 2017-07-11 Michael Collison <michael.collison@arm.com>
1492 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1495 2017-07-11 Carl Love <cel@us.ibm.com>
1497 * config/rs6000/rs6000-c.c: Add support for builtins
1498 vector unsigned int vec_parity_lsbb (vector signed int);
1499 vector unsigned int vec_parity_lsbb (vector unsigned int);
1500 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1501 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1502 vector unsigned long long vec_parity_lsbb (vector signed long long);
1503 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1504 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1505 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1506 * doc/extend.texi: Update the built-in documentation file for the
1507 new built-in functions.
1509 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1511 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1512 (layout::m_primary_loc): New field.
1513 (layout::layout): Initialize new field. Move location filtering
1514 logic from here to...
1515 (layout::maybe_add_location_range): ...this new method. Add
1516 support for filtering to just the lines already specified by other
1518 (layout::will_show_line_p): New method.
1519 (gcc_rich_location::add_location_if_nearby): New method.
1520 (selftest::test_add_location_if_nearby): New test function.
1521 (selftest::diagnostic_show_locus_c_tests): Call it.
1522 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1525 2017-07-11 Tom de Vries <tom@codesourcery.com>
1527 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1528 (bb_first_real_insn): New function.
1529 (nvptx_single): Add extra initialization of broadcasted condition
1532 2017-07-11 Nathan Sidwell <nathan@acm.org>
1534 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1536 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1538 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1539 Binutils doc as TEXI will mess them up.
1540 * doc/invoke.texi (AVR Options): Same here.
1542 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1544 * config/sparc/sparc.opt (mfix-ut700): New option.
1545 (mfix-gr712rc): Likewise.
1546 (sparc_fix_b2bst): New variable.
1547 * doc/invoke.texi (SPARC options): Document them.
1548 (ARM options): Fix warnings.
1549 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1550 instructions to prevent sequences that can trigger the store-store
1551 errata for certain LEON3FT processors.
1552 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1553 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1554 * config/sparc/sparc.md (fix_b2bst): New attribute.
1555 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1557 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1560 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1562 (*rsqrtsf2_sse): Ditto.
1564 (div<mode>3): Macroize insn from divdf3 and divsf3
1565 using MODEF mode iterator.
1567 2017-07-10 Martin Sebor <msebor@redhat.com>
1569 PR tree-optimization/80397
1570 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1571 instead of testing for equality to INTEGER_TYPE.
1573 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1575 * config.gcc: Remove uclibc from arc target spec.
1577 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1579 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1581 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1584 * lto-wrapper.c (remove_option): New function.
1585 (merge_and_complain): Merge PIC/PIE options more realistically.
1587 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1589 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1593 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1594 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1595 * config.in: Regenerate.
1596 * configure: Regenerate.
1597 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1598 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1599 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1600 (TARGET_GASISR_PROLOGUES): ...target mask.
1601 * common/config/avr/avr-common.c
1602 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1603 Set -mgas-isr-prologues.
1604 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1605 INSERT_PASS_BEFORE for it.
1606 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1607 * config/avr/avr.c (avr_option_override)
1608 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1609 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1610 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1611 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1612 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1614 (make_avr_pass_pre_proep): New function.
1615 (emit_push_sfr) <treg>: Add argument to function and use it
1617 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1618 and set machine->gasisr.yes.
1619 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1620 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1621 __gcc_isr.n_pushed to .L__stack_usage.
1622 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1623 (avr_asm_final_postscan_insn): ...this new static function.
1624 * config/avr/avr.h (machine_function)
1625 <is_no_gccisr, use_L__stack_usage>: New fields.
1626 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1627 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1628 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1629 (gasisr, *gasisr): New expander and insn.
1630 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1631 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1632 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1634 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1636 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1639 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1641 Move jump-tables out of .text again.
1644 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1645 (ASM_OUTPUT_ADDR_VEC): New function.
1646 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1647 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1648 INSN_ADDRESSes as asm comment.
1649 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1650 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1651 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1652 * config/avr/avr.md (*tablejump): Adjust comment.
1653 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1654 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1656 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1657 (avr_output_addr_vec): New proto.
1658 (avr_log_t) <insn_addresses>: New field.
1660 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1663 * config/i386/i386.c (ix86_function_arg_advance): Set
1664 outgoing_args_on_stack to true if there are outgoing arguments
1666 (ix86_function_arg): Likewise.
1667 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1668 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1669 * config/i386/i386.h (machine_function): Add
1670 outgoing_args_on_stack.
1672 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1674 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1676 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1678 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1680 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1681 (REAL_H): Remove $(MACHMODE_H).
1682 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1684 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1685 $(MACHMODE_H) and double-int.h.
1686 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1688 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1691 2017-07-07 Andrew Pinski <apinski@cavium.com>
1693 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1694 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1696 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1698 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1699 Add warning if GCC was not configured to link against a GLIBC that
1700 exports the hardware capability bits.
1701 (make_resolver_func): Make resolver function private and not a
1702 COMDAT function. Create the name with clone_function_name instead
1703 of make_unique_name.
1706 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1707 correct operand in doing the split.
1709 2017-07-07 Carl Love <cel@us.ibm.com>
1711 * config/rs6000/rs6000-c: Add support for built-in function
1712 vector unsigned short vec_pack_to_short_fp32 (vector float,
1714 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1715 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1716 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1717 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1718 (convert_4f32_8i16): Add define_expand.
1719 * doc/extend.texi: Update the built-in documentation file for the
1720 new built-in function.
1722 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1724 * config/sparc/m8.md: New file.
1725 * config/sparc/sparc.md: Include m8.md.
1727 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1729 * config/sparc/sparc.opt: New option -mvis4b.
1730 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
1731 (sparc_option_override): Handle VIS4B.
1732 (enum sparc_builtins): Define
1733 SPARC_BUILTIN_DICTUNPACK{8,16,32},
1734 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
1735 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
1736 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
1737 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
1738 (check_constant_argument): New function.
1739 (sparc_vis_init_builtins): Define builtins
1740 __builtin_vis_dictunpack{8,16,32},
1741 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
1742 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
1743 __builtin_vis_fpcmpde{8,16,32}shl and
1744 __builtin_vis_fpcmpur{8,16,32}shl.
1745 (sparc_expand_builtin): Check that the constant operands to
1746 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
1747 constant and in range.
1748 * config/sparc/sparc-c.c (sparc_target_macros): Handle
1750 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
1751 (SPARC_IMM5_P): Likewise.
1752 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
1753 (enabled): Handle vis4b.
1754 (UNSPEC_DICTUNPACK): New unspec.
1755 (UNSPEC_FPCMPSHL): Likewise.
1756 (UNSPEC_FPUCMPSHL): Likewise.
1757 (UNSPEC_FPCMPDESHL): Likewise.
1758 (UNSPEC_FPCMPURSHL): Likewise.
1759 (cpu_feature): New CPU feature `vis4b'.
1760 (dictunpack{8,16,32}): New insns.
1761 (FPCSMODE): New mode iterator.
1762 (fpcscond): New code iterator.
1763 (fpcsucond): Likewise.
1764 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
1765 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
1766 (fpcmpde{8,16,32}{si,di}shl): Likewise.
1767 (fpcmpur{8,16,32}{si,di}shl): Likewise.
1768 * config/sparc/constraints.md: Define constraints `q' for unsigned
1769 2-bit integer constants and `t' for unsigned 5-bit integer
1771 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
1773 (imm5_operand_dictunpack16): Likewise.
1774 (imm5_operand_dictunpack32): Likewise.
1775 (imm2_operand): Likewise.
1776 * doc/invoke.texi (SPARC Options): Document -mvis4b.
1777 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
1778 ditunpack* and fpcmp*shl builtins.
1780 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1782 * config.gcc: Handle m8 in --with-{cpu,tune} options.
1783 * config.in: Add HAVE_AS_SPARC6 define.
1784 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
1786 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
1788 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
1789 (CPP_CPU_SPEC): Handle m8.
1790 (ASM_CPU_SPEC): Likewise.
1791 * config/sparc/sparc-opts.h (enum processor_type): Add
1793 * config/sparc/sparc.c (m8_costs): New struct.
1794 (sparc_option_override): Handle TARGET_CPU_m8.
1795 (sparc32_initialize_trampoline): Likewise.
1796 (sparc64_initialize_trampoline): Likewise.
1797 (sparc_issue_rate): Likewise.
1798 (sparc_register_move_cost): Likewise.
1799 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
1800 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
1801 (ASM_CPU64_DEFAULT_SPEC): Likewise.
1802 (CPP_CPU_SPEC): Handle M8.
1803 (ASM_CPU_SPEC): Likewise.
1804 (AS_M8_FLAG): Define.
1805 * config/sparc/sparc.md: Add m8 to the cpu attribute.
1806 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
1807 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
1809 * configure: Regenerate.
1810 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
1813 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1815 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
1817 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
1818 ("*movdi_insn_sp32"): Do not set v3pipe.
1819 ("*movsi_insn"): Likewise.
1820 ("*movdi_insn_sp64"): Likewise.
1821 ("*movsf_insn"): Likewise.
1822 ("*movdf_insn_sp32"): Likewise.
1823 ("*movdf_insn_sp64"): Likewise.
1824 ("*zero_extendsidi2_insn_sp64"): Likewise.
1825 ("*sign_extendsidi2_insn"): Likewise.
1826 ("*mov<VM32:mode>_insn"): Likewise.
1827 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1828 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1829 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1830 ("<vlop:code><VL:mode>3"): Likewise.
1831 ("*not_<vlop:code><VL:mode>3"): Likewise.
1832 ("*nand<VL:mode>_vis"): Likewise.
1833 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
1834 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
1835 ("one_cmpl<VL:mode>2"): Likewise.
1836 ("faligndata<VM64:mode>_vis"): Likewise.
1837 ("alignaddrsi_vis"): Likewise.
1838 ("alignaddrdi_vis"): Likweise.
1839 ("alignaddrlsi_vis"): Likewise.
1840 ("alignaddrldi_vis"): Likewise.
1841 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1842 ("bmaskdi_vis"): Likewise.
1843 ("bmasksi_vis"): Likewise.
1844 ("bshuffle<VM64:mode>_vis"): Likewise.
1845 ("cmask8<P:mode>_vis"): Likewise.
1846 ("cmask16<P:mode>_vis"): Likewise.
1847 ("cmask32<P:mode>_vis"): Likewise.
1848 ("pdistn<P:mode>_vis"): Likewise.
1849 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1851 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1853 * config/sparc/sparc.md ("subtype"): New insn attribute.
1854 ("*wrgsr_sp64"): Set insn subtype.
1855 ("*rdgsr_sp64"): Likewise.
1856 ("alignaddrsi_vis"): Likewise.
1857 ("alignaddrdi_vis"): Likewise.
1858 ("alignaddrlsi_vis"): Likewise.
1859 ("alignaddrldi_vis"): Likewise.
1860 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1861 ("fexpand_vis"): Likewise.
1862 ("fpmerge_vis"): Likewise.
1863 ("faligndata<VM64:mode>_vis"): Likewise.
1864 ("bshuffle<VM64:mode>_vis"): Likewise.
1865 ("cmask8<P:mode>_vis"): Likewise.
1866 ("cmask16<P:mode>_vis"): Likewise.
1867 ("cmask32<P:mode>_vis"): Likewise.
1868 ("fchksm16_vis"): Likewise.
1869 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
1870 ("fmean16_vis"): Likewise.
1871 ("fp<plusminus_insn>64_vis"): Likewise.
1872 ("<plusminus_insn>v8qi3"): Likewise.
1873 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1874 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
1875 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
1876 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
1877 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
1878 ("*movqi_insn"): Likewise.
1879 ("*movhi_insn"): Likewise.
1880 ("*movsi_insn"): Likewise.
1881 ("movsi_pic_gotdata_op"): Likewise.
1882 ("*movdi_insn_sp32"): Likewise.
1883 ("*movdi_insn_sp64"): Likewise.
1884 ("movdi_pic_gotdata_op"): Likewise.
1885 ("*movsf_insn"): Likewise.
1886 ("*movdf_insn_sp32"): Likewise.
1887 ("*movdf_insn_sp64"): Likewise.
1888 ("*zero_extendhisi2_insn"): Likewise.
1889 ("*zero_extendqihi2_insn"): Likewise.
1890 ("*zero_extendqisi2_insn"): Likewise.
1891 ("*zero_extendqidi2_insn"): Likewise.
1892 ("*zero_extendhidi2_insn"): Likewise.
1893 ("*zero_extendsidi2_insn_sp64"): Likewise.
1894 ("ldfsr"): Likewise.
1895 ("prefetch_64"): Likewise.
1896 ("prefetch_32"): Likewise.
1897 ("tie_ld32"): Likewise.
1898 ("tie_ld64"): Likewise.
1899 ("*tldo_ldub_sp32"): Likewise.
1900 ("*tldo_ldub1_sp32"): Likewise.
1901 ("*tldo_ldub2_sp32"): Likewise.
1902 ("*tldo_ldub_sp64"): Likewise.
1903 ("*tldo_ldub1_sp64"): Likewise.
1904 ("*tldo_ldub2_sp64"): Likewise.
1905 ("*tldo_ldub3_sp64"): Likewise.
1906 ("*tldo_lduh_sp32"): Likewise.
1907 ("*tldo_lduh1_sp32"): Likewise.
1908 ("*tldo_lduh_sp64"): Likewise.
1909 ("*tldo_lduh1_sp64"): Likewise.
1910 ("*tldo_lduh2_sp64"): Likewise.
1911 ("*tldo_lduw_sp32"): Likewise.
1912 ("*tldo_lduw_sp64"): Likewise.
1913 ("*tldo_lduw1_sp64"): Likewise.
1914 ("*tldo_ldx_sp64"): Likewise.
1915 ("*mov<VM32:mode>_insn"): Likewise.
1916 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1917 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1919 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1921 * config/sparc/sparc.md ("type"): New insn type viscmp.
1922 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
1924 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
1925 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
1926 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1927 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
1929 ("n7_vis_logical_11cycle"): Likewise.
1930 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
1931 * config/sparc/niagara2.md ("niag3_vis": Likewise.
1932 * config/sparc/niagara.md ("niag_vis"): Likewise.
1933 * config/sparc/ultra3.md ("us3_fga"): Likewise.
1934 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
1936 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1938 * config/sparc/sparc.md: New instruction type `bmask'.
1939 (bmaskdi_vis): Use the `bmask' type.
1940 (bmasksi_vis): Likewise.
1941 * config/sparc/ultra3.md (us3_array): Likewise.
1942 * config/sparc/niagara7.md (n7_array): Likewise.
1943 * config/sparc/niagara4.md (n4_array): Likewise.
1944 * config/sparc/niagara2.md (niag2_vis): Likewise.
1945 (niag3_vis): Likewise.
1946 * config/sparc/niagara.md (niag_vis): Likewise.
1948 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1950 * ipa-comdats.c: Remove optimize check from gate.
1951 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
1952 for functions not optimized.
1953 (ipa_fn_summary_read): Skip optimize check.
1954 (ipa_fn_summary_write): Likewise.
1955 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
1957 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
1959 (can_inline_edge_p): Check flag_pcc_struct_return for match.
1960 (check_callers): Give up on caller which is not optimized.
1961 (inline_small_functions): Likewise.
1962 (ipa_inline): Do not give up when not optimizing.
1963 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
1964 away unoptimizes cdtors.
1965 (whole_program_function_and_variable_visibility): Do
1966 ipa_discover_readonly_nonaddressable_vars in LTO mode.
1967 * ipa.c (process_references): Do not check optimize.
1968 (symbol_table::remove_unreachable_nodes): Update optimize check.
1969 (set_writeonly_bit): Update optimize check.
1970 (pass_ipa_cdtor_merge::gate): Do not check optimize.
1971 (pass_ipa_single_use::gate): Remove.
1973 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1975 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
1976 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
1977 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
1978 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
1979 permute_load, permute_store, adjust_extract, adjust_splat,
1980 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
1981 replace_swap_with_copy, dump_swap_insn_table,
1982 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
1983 recombine_lvx_pattern, recombine_stvx_pattern,
1984 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
1985 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
1986 to file rs6000-p8swap.c.
1987 * config/rs6000/rs6000-p8swap.c: New file.
1988 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
1989 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
1990 and rs6000*-*-* targets.
1992 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1994 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
1996 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1998 * lto-wrapper.c (merge_and_complain): Do not merge
1999 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2000 fsigned_zeros, ftrapping_math, fwrapv.
2001 (append_compiler_options): Do not track these options.
2002 (append_linker_options): Likewie
2004 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2006 * cgraphunit.c (cgraph_node::finalize_function): When
2007 !flag_toplevel_reorde set no_reorder flag.
2008 (varpool_node::finalize_decl): Likewise.
2009 (symbol_table::compile): Drop no toplevel reorder path.
2011 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2013 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2014 edges; zero probability is not better than uninitialized.
2016 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2018 * asan.h (asan_sanitize_allocas_p): Declare.
2019 * asan.c (asan_sanitize_allocas_p): New function.
2020 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2021 (handle_builtin_alloca): Likewise.
2022 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2023 if !asan_sanitize_allocas_p.
2024 * params.def (asan-instrument-allocas): Add new option.
2025 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2026 * opts.c (common_handle_option): Disable allocas sanitization for
2029 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2031 * asan.c: Include gimple-fold.h.
2032 (get_last_alloca_addr): New function.
2033 (handle_builtin_stackrestore): Likewise.
2034 (handle_builtin_alloca): Likewise.
2035 (asan_emit_allocas_unpoison): Likewise.
2036 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2037 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2038 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2039 (instrument_builtin_call): Pass gimple iterator to
2040 get_mem_refs_of_builtin_call.
2041 (last_alloca_addr): New global.
2042 * asan.h (asan_emit_allocas_unpoison): Declare.
2043 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2044 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2045 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2046 if function calls alloca.
2047 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2048 * gimple-fold.h (replace_call_with_value): Declare.
2049 * internal-fn.c: Include asan.h.
2050 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2051 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2053 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2055 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2056 (C_SELFTEST_FLAGS): New.
2057 (CPP_SELFTEST_FLAGS): New.
2058 (SELFTEST_DEPS): New, from deps of s-selftest.
2059 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2060 (CPP_SELFTEST_DEPS): New.
2061 (selftest): Add dependency on s-selftest-c++.
2062 (s-selftest): Rename to...
2063 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2064 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2065 than SELFTEST_FLAGS.
2066 (selftest-gdb): Rename to...
2067 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2069 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2070 (selftest-valgrind): Rename to...
2071 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2073 (selftest-valgrind): Reintroduce as an alias for
2074 selftest-c-valgrind.
2075 (s-selftest-c++): New.
2076 (selftest-c++-gdb): New.
2077 (selftest-c++-valgrind): New.
2079 2017-07-06 Olivier Hainque <hainque@adacore.com>
2081 * gcc.c (process_command): When deciding if undefined variables
2082 should be ignored when processing specs, accept "gcc -v" as well.
2084 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2086 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2087 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2089 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2091 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2092 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2093 * config/arm/arm-tables.opt: Regenerate.
2094 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2096 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2098 2017-07-06 Carl Love <cel@us.ibm.com>
2100 * ChangeLog: Clean up from mid air collision
2102 2017-07-06 Carl Love <cel@us.ibm.com>
2104 * config/rs6000/rs6000-c.c: Add support for built-in functions
2105 vector signed int vec_subc (vector signed int, vector signed int);
2106 vector signed __int128 vec_subc (vector signed __int128,
2107 vector signed __int128);
2108 vector unsigned __int128 vec_subc (vector unsigned __int128,
2109 vector unsigned __int128);
2110 vector signed int vec_sube (vector signed int, vector signed int,
2112 vector unsigned int vec_sube (vector unsigned int,
2113 vector unsigned int,
2114 vector unsigned int);
2115 vector signed __int128 vec_sube (vector signed __int128,
2116 vector signed __int128,
2117 vector signed__int128);
2118 vector unsigned __int128 vec_sube (vector unsigned __int128,
2119 vector unsigned __int128,
2120 vector unsigned __int128);
2121 vector signed int vec_subec (vector signed int, vector signed int,
2123 vector unsigned int vec_subec (vector unsigned int,
2124 vector unsigned int,
2125 vector unsigned int);
2126 vector signed __int128 vec_subec (vector signed __int128,
2127 vector signed __int128,
2128 vector signed__int128);
2129 vector unsigned __int128 vec_subec (vector unsigned __int128,
2130 vector unsigned __int128,
2131 vector unsigned __int128);
2132 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2133 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2134 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2135 BU_ALTIVEC_OVERLOAD_X definitions.
2136 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2137 * doc/extend.texi: Update the built-in documentation file for the new
2140 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2143 * diagnostic-show-locus.c (layout::layout): Use start and finish
2144 spelling location for the start and finish of each range.
2145 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2146 Add unused aspect param.
2147 * input.c (expand_location_1): Add "aspect" param, and use it
2148 to access the correct part of the location.
2149 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2151 (expand_location_to_spelling_point): Likewise.
2152 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2153 param, and pass it to expand_location_1.
2155 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2157 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2158 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2159 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2160 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2161 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2162 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2163 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2164 _mm_maskz_getmant_ss): New intrinsics.
2165 (__builtin_ia32_getexpss128_mask): Changed to ...
2166 __builtin_ia32_getexpss128_round ... this.
2167 (__builtin_ia32_getexpsd128_mask): Changed to ...
2168 __builtin_ia32_getexpsd128_round ... this.
2169 * config/i386/i386-builtin-types.def
2170 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2171 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2172 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2173 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2174 __builtin_ia32_getmantss_mask_round): New builtins.
2175 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2176 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2177 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2178 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2179 * config/i386/sse.md
2180 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2181 avx512f_sgetexp<mode><mask_scalar_name>
2182 <round_saeonly_scalar_name> ... this.
2183 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2184 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2185 vgetexp<ssescalarmodesuffix>
2186 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2187 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2188 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2189 avx512f_vgetmant<mode><mask_scalar_name>
2190 <round_saeonly_scalar_name> ... this.
2191 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2192 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2193 vgetmant<ssescalarmodesuffix>
2194 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2195 %0<mask_scalar_operand4>, %1, %2
2196 <round_saeonly_scalar_mask_op4>, %3} ... this.
2197 * config/i386/subst.md (mask_scalar_operand4,
2198 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2199 round_saeonly_scalar_nimm_predicate): New subst attributes.
2201 2017-07-06 Julia Koval <julia.koval@intel.com>
2203 * config/i386/i386.c (ix86_erase_embedded_rounding):
2204 Remove code for old rounding pattern.
2206 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2208 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2210 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2212 * doc/sourcebuild.texi (Test Directives, Variants of
2213 dg-require-support): Add documentation for dg-require-stack-check.
2215 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2217 * config/i386/subst.md (mask_scalar, round_scalar,
2218 round_saeonly_scalar): New meta-templates.
2219 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2220 round_scalar_mask_operand3, round_scalar_mask_op3,
2221 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2222 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2223 round_saeonly_scalar_constraint,
2224 round_saeonly_scalar_prefix): New subst attribute.
2225 * config/i386/sse.md
2226 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2227 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2228 <round_scalar_name> ... this.
2229 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2230 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2231 <round_scalar_name> ... this.
2232 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2233 <sse>_vm<code><mode>3<mask_scalar_name>
2234 <round_saeonly_scalar_name> ... this.
2235 (v<plusminus_mnemonic><ssescalarmodesuffix>
2236 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2237 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2238 v<plusminus_mnemonic><ssescalarmodesuffix>
2239 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2240 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2241 (v<multdiv_mnemonic><ssescalarmodesuffix>
2242 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2243 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2244 v<multdiv_mnemonic><ssescalarmodesuffix>
2245 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2246 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2247 (v<maxmin_float><ssescalarmodesuffix>
2248 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2249 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2250 v<maxmin_float><ssescalarmodesuffix>
2251 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2252 %0<mask_scalar_operand3>, %1, %<iptr>2
2253 <round_saeonly_scalar_mask_op3>} ... this.
2255 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2257 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2258 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2260 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2261 Alan Hayward <alan.hayward@arm.com>
2262 David Sherwood <david.sherwood@arm.com>
2264 * combine.c (simplify_if_then_else): Remove "enum" before
2266 * compare-elim.c (can_eliminate_compare): Likewise.
2267 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2269 (aarch64_lookup_simd_builtin_type): Likewise.
2270 (aarch64_simd_builtin_type): Likewise.
2271 (aarch64_init_simd_builtin_types): Likewise.
2272 (aarch64_simd_expand_args): Likewise.
2273 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2275 (aarch64_reverse_mask): Likewise.
2276 (aarch64_simd_emit_reg_reg_move): Likewise.
2277 (aarch64_gen_adjusted_ldpstp): Likewise.
2278 (aarch64_ccmp_mode_to_code): Likewise.
2279 (aarch64_operands_ok_for_ldpstp): Likewise.
2280 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2281 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2283 (aarch64_min_divisions_for_recip_mul): Likewise.
2284 (aarch64_reassociation_width): Likewise.
2285 (aarch64_get_condition_code_1): Likewise.
2286 (aarch64_simd_emit_reg_reg_move): Likewise.
2287 (aarch64_simd_attr_length_rglist): Likewise.
2288 (aarch64_reverse_mask): Likewise.
2289 (aarch64_operands_ok_for_ldpstp): Likewise.
2290 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2291 (aarch64_gen_adjusted_ldpstp): Likewise.
2292 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2294 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2295 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2296 (arm_lookup_simd_builtin_type): Likewise.
2297 (arm_simd_builtin_type): Likewise.
2298 (arm_init_simd_builtin_types): Likewise.
2299 (arm_expand_builtin_args): Likewise.
2300 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2301 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2302 (ft32_setup_incoming_varargs): Likewise.
2303 (ft32_function_arg): Likewise.
2304 (ft32_function_arg_advance): Likewise.
2305 (ft32_pass_by_reference): Likewise.
2306 (ft32_arg_partial_bytes): Likewise.
2307 (ft32_valid_pointer_mode): Likewise.
2308 (ft32_addr_space_pointer_mode): Likewise.
2309 (ft32_addr_space_legitimate_address_p): Likewise.
2310 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2312 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2313 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2314 (iamcu_alignment): Likewise.
2315 (canonicalize_vector_int_perm): Likewise.
2316 (ix86_noce_conversion_profitable_p): Likewise.
2317 (ix86_mpx_bound_mode): Likewise.
2318 (ix86_operands_ok_for_move_multiple): Likewise.
2319 * config/microblaze/microblaze-protos.h
2320 (microblaze_expand_conditional_branch_reg): Likewise.
2321 * config/microblaze/microblaze.c
2322 (microblaze_expand_conditional_branch_reg): Likewise.
2323 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2325 (rs6000_reassociation_width): Likewise.
2326 (rs6000_invalid_binary_op): Likewise.
2327 (fusion_p9_p): Likewise.
2328 (emit_fusion_p9_load): Likewise.
2329 (emit_fusion_p9_store): Likewise.
2330 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2332 (riscv_hard_regno_mode_ok_p): Likewise.
2333 (riscv_address_insns): Likewise.
2334 (riscv_split_symbol): Likewise.
2335 (riscv_legitimize_move): Likewise.
2336 (riscv_function_value): Likewise.
2337 (riscv_hard_regno_nregs): Likewise.
2338 (riscv_expand_builtin): Likewise.
2339 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2340 (riscv_build_integer): Likewise.
2341 (riscv_split_integer): Likewise.
2342 (riscv_legitimate_constant_p): Likewise.
2343 (riscv_cannot_force_const_mem): Likewise.
2344 (riscv_regno_mode_ok_for_base_p): Likewise.
2345 (riscv_valid_base_register_p): Likewise.
2346 (riscv_valid_offset_p): Likewise.
2347 (riscv_valid_lo_sum_p): Likewise.
2348 (riscv_classify_address): Likewise.
2349 (riscv_legitimate_address_p): Likewise.
2350 (riscv_address_insns): Likewise.
2351 (riscv_load_store_insns): Likewise.
2352 (riscv_force_binary): Likewise.
2353 (riscv_split_symbol): Likewise.
2354 (riscv_force_address): Likewise.
2355 (riscv_legitimize_address): Likewise.
2356 (riscv_move_integer): Likewise.
2357 (riscv_legitimize_const_move): Likewise.
2358 (riscv_legitimize_move): Likewise.
2359 (riscv_address_cost): Likewise.
2360 (riscv_subword): Likewise.
2361 (riscv_output_move): Likewise.
2362 (riscv_canonicalize_int_order_test): Likewise.
2363 (riscv_emit_int_order_test): Likewise.
2364 (riscv_function_arg_boundary): Likewise.
2365 (riscv_pass_mode_in_fpr_p): Likewise.
2366 (riscv_pass_fpr_single): Likewise.
2367 (riscv_pass_fpr_pair): Likewise.
2368 (riscv_get_arg_info): Likewise.
2369 (riscv_function_arg): Likewise.
2370 (riscv_function_arg_advance): Likewise.
2371 (riscv_arg_partial_bytes): Likewise.
2372 (riscv_function_value): Likewise.
2373 (riscv_pass_by_reference): Likewise.
2374 (riscv_setup_incoming_varargs): Likewise.
2375 (riscv_print_operand): Likewise.
2376 (riscv_elf_select_rtx_section): Likewise.
2377 (riscv_save_restore_reg): Likewise.
2378 (riscv_for_each_saved_reg): Likewise.
2379 (riscv_register_move_cost): Likewise.
2380 (riscv_hard_regno_mode_ok_p): Likewise.
2381 (riscv_hard_regno_nregs): Likewise.
2382 (riscv_class_max_nregs): Likewise.
2383 (riscv_memory_move_cost): Likewise.
2384 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2385 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2386 (rl78_addr_space_address_mode): Likewise.
2387 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2389 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2390 (rs6000_reassociation_width): Likewise.
2391 (rs6000_invalid_binary_op): Likewise.
2392 (fusion_p9_p): Likewise.
2393 (emit_fusion_p9_load): Likewise.
2394 (emit_fusion_p9_store): Likewise.
2395 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2396 (ok_for_simple_move_operands): Likewise.
2397 (ok_for_simple_move_strict_operands): Likewise.
2398 (ok_for_simple_arith_logic_operands): Likewise.
2399 (visium_legitimize_reload_address): Likewise.
2400 (visium_select_cc_mode): Likewise.
2401 (output_cbranch): Likewise.
2402 (visium_split_double_move): Likewise.
2403 (visium_expand_copysign): Likewise.
2404 (visium_expand_int_cstore): Likewise.
2405 (visium_expand_fp_cstore): Likewise.
2406 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2407 (visium_function_arg): Likewise.
2408 (visium_function_arg_advance): Likewise.
2409 (visium_libcall_value): Likewise.
2410 (visium_setup_incoming_varargs): Likewise.
2411 (visium_legitimate_constant_p): Likewise.
2412 (visium_legitimate_address_p): Likewise.
2413 (visium_legitimize_address): Likewise.
2414 (visium_secondary_reload): Likewise.
2415 (visium_register_move_cost): Likewise.
2416 (visium_memory_move_cost): Likewise.
2417 (prepare_move_operands): Likewise.
2418 (ok_for_simple_move_operands): Likewise.
2419 (ok_for_simple_move_strict_operands): Likewise.
2420 (ok_for_simple_arith_logic_operands): Likewise.
2421 (visium_function_value_1): Likewise.
2422 (rtx_ok_for_offset_p): Likewise.
2423 (visium_legitimize_reload_address): Likewise.
2424 (visium_split_double_move): Likewise.
2425 (visium_expand_copysign): Likewise.
2426 (visium_expand_int_cstore): Likewise.
2427 (visium_expand_fp_cstore): Likewise.
2428 (visium_split_cstore): Likewise.
2429 (visium_select_cc_mode): Likewise.
2430 (visium_split_cbranch): Likewise.
2431 (output_cbranch): Likewise.
2432 (visium_print_operand_address): Likewise.
2433 * expmed.c (flip_storage_order): Likewise.
2434 * expmed.h (emit_cstore): Likewise.
2435 (flip_storage_order): Likewise.
2436 * genrecog.c (validate_pattern): Likewise.
2437 * hsa-gen.c (gen_hsa_addr): Likewise.
2438 * internal-fn.c (expand_arith_overflow): Likewise.
2439 * ira-color.c (allocno_copy_cost_saving): Likewise.
2440 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2441 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2442 (process_invariant_for_inheritance): Likewise.
2443 * lra-eliminations.c (move_plus_up): Likewise.
2444 * omp-low.c (lower_oacc_reductions): Likewise.
2445 * simplify-rtx.c (simplify_subreg): Likewise.
2446 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2447 (TARGET_CHKP_BOUND_MODE): Likewise..
2448 * targhooks.c (default_chkp_bound_mode): Likewise.
2449 (default_setup_incoming_vararg_bounds): Likewise.
2450 * targhooks.h (default_chkp_bound_mode): Likewise.
2451 (default_setup_incoming_vararg_bounds): Likewise.
2452 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2453 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2454 (have_whole_vector_shift): Likewise.
2455 * tree-vect-stmts.c (vectorizable_load): Likewise.
2456 * doc/tm.texi: Regenerate.
2458 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2460 Graceful degrade if Binutils PR21472 is not available.
2463 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2464 .rodata in flash test fails.
2465 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2466 * confgure: Regenerate.
2467 * config.in: Regenerate.
2468 * config/avr/avr.c (avr_asm_named_section)
2469 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2470 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2471 (avr_asm_init_sections): Same.
2473 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2475 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2476 (fma<VH:mode>4_intrinsic): Likewise.
2477 (*fmsub<VCVTF:mode>4): Likewise.
2478 (*fmsub<VH:mode>4_intrinsic): Likewise.
2480 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2483 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2484 Don't depend on "optimize > 0".
2485 (out_movhi_r_mr, out_movqi_mr_r): Same.
2486 (out_movhi_mr_r, out_movqi_r_mr): Same.
2487 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2488 io_address_operand on "optimize > 0".
2490 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2492 * tree-loop-distribution.c: Add general explanantion on the pass.
2493 (generate_loops_for_partition): Mark distributed loop.
2494 (pg_add_dependence_edges): New parameter. Handle alias data
2495 dependence specially and record it in the parameter if asked.
2496 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2497 (init_partition_graph_vertices, add_partition_graph_edge): New.
2498 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2499 (free_partition_graph_vdata, build_partition_graph): New.
2500 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2501 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2502 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2503 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2504 (version_for_distribution_p, finalize_partitions): New.
2505 (distribute_loop): Handle alias data dependence specially. Factor
2506 out loop fusion code as functions and call these functions.
2508 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2510 * tree-loop-distribution.c (classify_partition): New parameter and
2511 better handle reduction statement.
2512 (rdg_build_partitions): Revise comment.
2513 (distribute_loop): Compute statements in all partitions and pass it
2514 to classify_partition.
2516 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2518 * tree-loop-distribution.c (enum partition_type): New.
2519 (struct partition): New field type.
2520 (partition_merge_into): Add parameter. Update partition type.
2521 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2522 (build_rdg_partition_for_vertex): Compute partition type.
2523 (rdg_build_partitions): Dump partition type.
2524 (distribute_loop): Update calls to partition_merge_into.
2526 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2528 * tree-loop-distribution.c (struct ddr_hasher): New.
2529 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2531 (classify_partition): Call get_data_dependence.
2532 (pg_add_dependence_edges): Ditto.
2533 (distribute_loop): Release data dependence hash table.
2535 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2537 * tree-loop-distribution.c (ref_base_address): Delete.
2538 (similar_memory_accesses): Rename ...
2539 (share_memory_accesses): ... to this. Check if partitions access
2540 the same memory reference.
2541 (distribute_loop): Call share_memory_accesses.
2543 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2545 * tree-loop-distribution.c (struct partition): New field recording
2547 (partition_alloc, partition_free): Init and release data refs.
2548 (partition_merge_into): Merge data refs.
2549 (build_rdg_partition_for_vertex): Collect data refs for partition.
2550 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2552 (distribute_loop): Remve data refs from vertice data of partition
2555 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2557 * tree-loop-distribution.c (params.h): Include header file.
2558 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2559 (datarefs_vec): New global var.
2560 (create_rdg_vertices): Use datarefs_vec directly.
2561 (free_rdg): Don't free data references.
2562 (build_rdg): Update use. Don't free data references.
2563 (distribute_loop): Compute global variable for data references.
2564 Bail out if there are too many data references.
2566 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2568 * tree-loop-distribution.c (loop_nest): New global var.
2569 (build_rdg): Use loop directly, rather than loop nest.
2570 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2572 (distribute_loop): Compute global variable loop nest. Update use.
2574 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2576 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2577 (partition_merge_into): New parameter. Dump reason for fusion.
2578 (distribute_loop): Update use of partition_merge_into.
2580 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2582 * tree-loop-distribution.c (bb_top_order_index): New.
2583 (bb_top_order_index_size, bb_top_order_cmp): New.
2584 (stmts_from_loop): Use topological order.
2585 (pass_loop_distribution::execute): Compute and release topological
2586 order for basic blocks.
2588 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2590 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2593 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2595 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2596 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2597 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2598 * internal-fn.def (LOOP_DIST_ALIAS): New.
2599 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2600 (fold_loop_internal_call): ... this.
2601 (vect_loop_dist_alias_call): New function.
2602 (set_uid_loop_bbs): Call fold_loop_internal_call.
2603 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2606 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2609 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2610 Require dead FLAGS_REG at the beginning of a peephole.
2612 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2615 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2616 arguments in the call to __builtin_ia32_sbb_u32.
2617 (_subborrow_u64): Swap _X and _Y arguments in the call to
2618 __builtin_ia32_sbb_u64.
2620 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2623 * tree-vrp.c (compare_assert_loc): Turn into a function template
2624 with stable template parameter. Only test if a->e is NULL,
2625 !a->e == !b->e has been verified already. Use e == NULL or
2626 e != NULL instead of e or ! e tests. If stable is true, don't use
2627 iterative_hash_expr, on the other side allow a or b or both NULL
2628 and sort the NULLs last.
2629 (process_assert_insertions): Sort using compare_assert_loc<false>
2630 instead of compare_assert_loc, later sort using
2631 compare_assert_loc<true> before calling process_assert_insertions_for
2632 in a loop. Use break instead of continue once seen NULL pointer.
2634 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2636 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2637 Cortex-R7 and Cortex-R8 processors.
2639 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2641 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2642 uninitialized while src is not.
2644 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2646 * common/config/arm/arm-common.c: Adjust include path for
2648 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2649 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2650 (arm-cpu-data.h): Likewise.
2651 (arm-cpu-cdata.h): Likewise.
2652 * config/arm/arm-cpu.h: Delete.
2653 * config/arm/arm-cpu-cdata.h: Delete.
2654 * config/arm/arm-cpu-data.h: Delete.
2656 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2658 * config/arm/arm-cpus.in (cortex-a55): New.
2659 (cortex-a75): Likewise.
2660 (cortex-a75.cortex-a55): Likewise.
2661 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2663 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2664 * config/arm/arm-cpu-cdata.h: Regenerate.
2665 * config/arm/arm-cpu-data.h: Regenerate.
2666 * config/arm/arm-cpu.h: Regenerate.
2667 * config/arm/arm-tables.opt: Regenerate.
2668 * config/arm/arm-tune.md: Regenerate.
2670 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2672 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2674 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2676 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2679 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2681 PR tree-optimization/81292
2682 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2683 full_string_p, also call adjust_related_strinfos if the adjustment
2684 is simple, otherwise invalidate related strinfos.
2686 2017-07-04 Martin Liska <mliska@suse.cz>
2689 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2690 newly created variable as DECL_IGNORED_P.
2692 2017-07-04 Martin Liska <mliska@suse.cz>
2695 * ipa-inline.c (inline_small_functions):
2696 Use xstrdup_for_dump.
2698 2017-07-04 Tom de Vries <tom@codesourcery.com>
2700 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2702 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2705 * config/darwin.c (darwin_function_switched_text_sections):
2708 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2710 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2712 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2714 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2716 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2718 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2719 min_profitable_iters, and th as inclusive lower bounds.
2720 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
2721 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
2722 for min_profitable_iters and min_profitable_estimate.
2723 (vect_transform_loop): Treat th as an inclusive lower bound.
2724 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2726 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2729 * config/darwin.c (darwin_function_switched_text_sections):
2730 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
2731 in two pieces, and suppress the use of buf.
2733 2017-07-03 Nathan Sidwell <nathan@acm.org>
2735 * hash-table.h (hash_table_mod1): Fix indentation.
2737 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2740 * predict.c (force_edge_cold): Be more careful about propagation
2742 * profile-count.h (profile_probability::guessed,
2743 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
2745 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
2747 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
2749 * doc/invoke.texi (rcpc architecture extension): Document it.
2751 2017-07-03 Richard Biener <rguenther@suse.de>
2753 PR tree-optimization/60510
2754 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
2755 the scalar reduction PHI and use it.
2756 (vectorizable_reduction): Properly guard the single_defuse_cycle
2757 path for non-SLP reduction chains where we cannot use it.
2758 Rework reduc_def/index and vector type deduction. Rework
2759 vector operand gathering during reduction op code-gen.
2760 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
2761 chains dissolve the chain and leave it to non-SLP reduction
2764 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2766 * tree-data-ref.h (dr_alignment): Declare.
2767 * tree-data-ref.c (dr_alignment): New function.
2768 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
2769 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
2771 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
2773 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2775 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
2776 and base_misalignment fields.
2777 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
2778 * tree-data-ref.c: Include builtins.h.
2779 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
2780 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
2781 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
2782 * tree-vect-data-refs.c: Include tree-cfg.h.
2783 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
2784 fields instead of calculating an alignment here.
2785 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
2786 innermost_loop_behavior fields.
2788 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2790 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
2792 (DR_STEP_ALIGNMENT): New macro.
2793 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
2794 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
2795 (create_data_ref): Print it.
2796 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
2797 to tell whether the step preserves vector (mis)alignment.
2798 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2799 Move the check for an integer step and generalise to all INTEGER_CST.
2800 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
2801 Print the outer step alignment.
2803 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2805 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
2806 with offset_alignment.
2807 (DR_ALIGNED_TO): Delete.
2808 (DR_OFFSET_ALIGNMENT): New macro.
2809 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
2810 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
2811 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
2812 (create_data_ref): Likewise.
2813 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2814 (vect_analyze_data_refs): Likewise.
2815 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
2816 creating dummy innermost behavior.
2818 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2820 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
2821 with a "innermost_loop_behavior *" and refeence tree.
2822 * tree-data-ref.c (dr_analyze_innermost): Likewise.
2823 (create_data_ref): Update call accordingly.
2824 * tree-predcom.c (find_looparound_phi): Likewise.
2826 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2828 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
2829 fields with dr_wrt_vec_loop.
2830 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
2831 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
2832 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
2833 (vect_dr_behavior): New function.
2834 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2835 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
2836 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
2837 track whether the step preserves the misalignment.
2838 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2839 Use vect_dr_behavior.
2840 (vect_setup_realignment): Update call accordingly.
2841 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
2842 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
2843 call to vect_create_addr_base_for_vector_ref.
2844 (vect_create_cond_for_align_checks): Likewise.
2845 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
2846 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
2847 (vect_recog_mask_conversion_pattern): Likewise.
2848 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
2849 (new_stmt_vec_info): Remove redundant zeroing.
2851 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2853 * common/config/arm/arm-common.c (arm_be8_option): New function.
2854 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
2855 (ISA_ARMv6): Add isa_bit_be8.
2856 * config/arm/arm.h (arm_be8_option): Add prototype.
2857 (BE8_SPEC_FUNCTION): New define.
2858 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
2859 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
2860 (mlittle-endian): Similarly.
2861 (mbe8, mbe32): New options.
2862 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
2863 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
2865 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2867 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
2869 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2871 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
2872 (cleanup_tree_cfg_bb): Use it.
2873 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
2875 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
2877 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2880 * loop-doloop.c (add_test): Update profile.
2882 2017-07-03 Martin Liska <mliska@suse.cz>
2885 * sanopt.c (rewrite_usage_of_param): New function.
2886 (sanitize_rewrite_addressable_params): Likewise.
2887 (pass_sanopt::execute): Call rewrite_usage_of_param.
2889 2017-07-03 Richard Biener <rguenther@suse.de>
2891 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
2892 back to using VIEW_CONVERT_EXPR.
2894 2017-07-03 Martin Liska <mliska@suse.cz>
2897 * doc/extend.texi: Document when a resolver function is
2898 generated for target_clones.
2900 2017-07-03 Martin Liska <mliska@suse.cz>
2902 * asan.c (asan_emit_stack_protection): Unpoison just red zones
2903 and shadow memory of auto variables which are subject of
2904 use-after-scope sanitization.
2905 (asan_expand_mark_ifn): Add do set only when is_poison.
2907 2016-07-03 Richard Biener <rguenther@suse.de>
2909 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
2911 (vect_force_simple_reduction): Record reduction def -> phi mapping.
2912 (vectorizable_reduction): Perform reduction PHI creation when
2913 visiting a reduction PHI and adjust and simplify code generation
2914 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
2915 (vect_transform_loop): Visit reduction PHIs.
2916 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
2917 defs into the SLP tree.
2918 (vect_build_slp_tree): Reduction defs terminate the recursion.
2919 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
2921 (vect_get_vec_defs_for_stmt_copy): Export.
2922 (vect_get_vec_defs): Likewise.
2923 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
2925 (vect_get_vec_defs_for_stmt_copy): Declare.
2926 (vect_get_vec_defs): Likewise.
2928 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2930 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
2931 parameter with a "loop" parameter and use it instead of the
2932 loop containing DR_STMT. Don't check simple_iv when doing
2933 BB analysis. Describe the two analysis modes in the comment.
2935 2017-07-03 Tom de Vries <tom@codesourcery.com>
2937 PR tree-optimization/69468
2938 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
2939 (find_same_succ_bb): Handle ignore_edge_flags.
2941 2017-07-03 Tom de Vries <tom@codesourcery.com>
2943 PR tree-optimization/81192
2944 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
2946 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
2948 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
2950 2017-07-03 Tom de Vries <tom@codesourcery.com>
2952 PR tree-optimization/81192
2953 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
2954 BB_SAME_SUCC (bb) == NULL.
2956 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2958 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
2961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2963 * dumpfile.c: Include profile-count.h
2964 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
2966 (insert_cond_bb): Update profile.
2967 * tree-cfg.h (insert_cond_bb): Update prototype.
2968 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
2969 * tree-dump.c: Do not include tree-cfg.
2971 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2973 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
2975 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2977 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
2980 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2982 * tree-complex.c (expand_complex_div_wide): update profile.
2984 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2985 Alan Hayward <alan.hayward@arm.com>
2986 David Sherwood <david.sherwood@arm.com>
2988 * Makefile.in (MACHMODE_H): Remove insn-modes.h
2989 (CORETYPES_H): New define.
2990 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
2991 (insn-modes-inline.h, s-modes-inline-h): New rules.
2992 (generated_files): Add insn-modes-inline.h.
2993 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
2994 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
2995 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
2996 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
2997 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
2998 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
2999 (build/gencodes.o, build/genconditions.o): Likewise.
3000 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3001 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3002 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3003 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3004 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3005 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3006 * coretypes.h: Include everything up to real.h for generators.
3007 Include insn-modes.h first. Include wide-int-print.h after
3008 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3009 * machmode.h: Don't include insn-modes.h here.
3010 * function-tests.c: Remove includes of signop.h, machmode.h,
3011 double-int.h and wide-int.h.
3013 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3015 * optc-save-gen.awk: Likewise.
3016 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3017 * godump.c: Remove include of wide-int-print.h.
3018 * pretty-print.h: Likewise.
3019 * wide-int-print.cc: Likewise.
3020 * wide-int.cc: Likewise.
3021 * hash-map-tests.c: Remove include of signop.h.
3022 * hash-set-tests.c: Likewise.
3023 * rtl-tests.c: Likewise.
3024 * mkconfig.sh: Remove include of machmode.h.
3025 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3027 (emit_insn_modes_inline_h): ...this new function. Emit the code
3028 into an insn-modes-inline.h header file, adding appropriate
3029 include guards and end comments.
3030 (emit_insn_modes_c_header): Remove include of machmode.h.
3031 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3033 (main): Handle -i flag and call emit_insn_modes_inline_h when
3036 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3038 * tree-ssa-strlen.c (strinfo): Rename the length field to
3039 nonzero_chars. Add a full_string_p field.
3040 (compare_nonzero_chars, zero_length_string_p): New functions.
3041 (get_addr_stridx): Add an offset_out parameter.
3042 Use compare_nonzero_chars.
3043 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3044 (new_strinfo): Update after above changes to strinfo.
3045 (set_endptr_and_length): Set full_string_p.
3046 (get_string_length): Update after above changes to strinfo.
3047 (unshare_strinfo): Update call to new_strinfo.
3048 (maybe_invalidate): Likewise.
3049 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3050 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3051 as a uhwi instead of an shwi. Update after above changes to
3052 strinfo and new_strinfo.
3053 (zero_length_string): Assert that chainsi contains full strings.
3054 Use zero_length_string_p. Update call to new_strinfo.
3055 (adjust_related_strinfos): Update after above changes to strinfo.
3056 Copy full_string_p from origsi.
3057 (adjust_last_stmt): Use zero_length_string_p.
3058 (handle_builtin_strlen): Update after above changes to strinfo and
3059 new_strinfo. Install the lhs as the string length if the previous
3060 entry didn't describe a full string.
3061 (handle_builtin_strchr): Update after above changes to strinfo
3063 (handle_builtin_strcpy): Likewise.
3064 (handle_builtin_strcat): Likewise.
3065 (handle_builtin_malloc): Likewise.
3066 (handle_pointer_plus): Likewise.
3067 (handle_builtin_memcpy): Likewise. Track nonzero characters
3068 that aren't necessarily followed by a nul terminator.
3069 (handle_char_store): Likewise.
3071 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3073 PR tree-optimization/80769
3074 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3075 for malloc and calloc. Document the new invariant that all related
3076 strinfos have delayed lengths or none do.
3077 (verify_related_strinfos): Move earlier in file.
3078 (set_endptr_and_length): New function, split out from...
3079 (get_string_length): ...here. Also set the lengths of related
3081 (zero_length_string): Assert that chainsi has known (rather than
3083 (adjust_related_strinfos): Likewise.
3085 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3087 PR tree-optimization/81136
3088 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3089 assert that two references with the same misalignment have the same
3090 compile-time misalignment if those compile-time misalignments
3093 2017-07-01 Andi Kleen <ak@linux.intel.com>
3095 * print-tree.c (print_node): Print all attributes.
3097 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3099 * cfg.c (scale_bbs_frequencies): New function.
3100 * cfg.h (scale_bbs_frequencies): Declare it.
3101 * cfgloopanal.c (single_likely_exit): Cleanup.
3102 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3104 (scale_loop_profile): Likewise.
3105 (loop_version): Likewise.
3106 (create_empty_loop_on_edge): Update.
3107 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3108 scale_loop_frequencies, scale_loop_profile, loopify,
3109 loop_version): Update prototypes.
3110 * modulo-sched.c (sms_schedule): Update.
3111 * predict.c (unlikely_executed_edge_p): Also check probability.
3112 (probably_never_executed_edge_p): Fix typo.
3113 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3114 * tree-parloops.c (gen_parallel_loop): Update.
3115 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3116 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3117 * tree-ssa-loop-split.c (split_loop): Update.
3118 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3119 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3120 (vect_loop_versioning): Update.
3121 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3123 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3125 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3127 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3129 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3130 to keep profile consistent.
3132 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3134 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3135 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3136 * profile-count.h (max_safe_multiplier): Make unsigned.
3137 (profile_count::guessed_zero): New.
3139 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3141 * bb-reorder.c (fix_up_crossing_landing_pad,
3142 fix_crossing_conditional_branches): Use make_single_succ_edge
3143 to keep profile consistent.
3145 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3147 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3150 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3153 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3154 the right scopes, make sure cond_jump isn't preserved between multiple
3155 iterations. Search for fallthru edge whenever there are 3+ edges and
3156 use find_fallthru_edge for it.
3158 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3160 Patch by Alexander Monakov <amonakov@ispras.ru>
3161 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3162 probabilities consistently.
3164 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3166 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3167 emit_cmp_and_jump_insns.
3169 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3172 * tree-inline.c (expand_call_inline): Combine profile statuses.
3174 2017-06-30 Andrew Pinski <apinski@cavium.com>
3176 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3177 fold_stmt returned true.
3179 2017-06-30 Nathan Sidwell <nathan@acm.org>
3181 * ggc.h (empty_string): Delete.
3182 * cfgexpand.c (expand_asm_stmt): Use plain "".
3183 * optabs.c (expand_asm_memory_barrier): Likewise.
3184 * stringpool.c (empty_string): Delete.
3185 (digit_vector, digit_string): Delete.
3186 (ggc_alloc_string): Use plain "", don't optimize single digit
3187 strings. Use ggc_alloc_atomic.
3189 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3191 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3192 comparison set and one other set, use the cost of the non-comparison
3195 2017-06-30 Nathan Sidwell <nathan@acm.org>
3197 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3200 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3202 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3203 loops. Remove now unneeded calls to gimple_switch_set_label() that
3204 just set removed labels to NULL_TREE.
3206 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3208 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3209 (set_range_info): ...here. Only call set_range_info_raw if domain
3211 (set_nonzero_bits): Call set_range_info_raw.
3212 * tree-ssanames.h (set_range_info_raw): New.
3214 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3217 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3218 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3219 of nonimmediate_operand and <store_mask_constraint> instead of m
3220 for the input operand. For V8FI iterator, always split if input
3221 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3222 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3223 <store_mask_predicate> instead of register_operand and
3224 <store_mask_constraint> instead of v for the input operand. Make
3225 sure both operands aren't MEMs for if not <mask_applied>.
3227 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3229 * lto-wrapper.c (copy_file) Close both file descriptors before
3232 2017-06-30 Martin Liska <mliska@suse.cz>
3235 * multiple_target.c (create_dispatcher_calls): Make ifunc
3236 also for function that don't have calls or are not referenced.
3238 2017-06-30 Richard Biener <rguenther@suse.de>
3240 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3241 analyze the first scalar stmt. Move vector type computation
3242 for the BB case here from ...
3243 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3244 live operation processing in the SLP case properly.
3246 2017-06-30 Richard Biener <rguenther@suse.de>
3248 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3250 2017-06-30 Martin Liska <mliska@suse.cz>
3253 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3254 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3256 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3258 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3261 2017-06-30 Richard Biener <rguenther@suse.de>
3263 PR tree-optimization/81249
3264 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3265 condition reduction result to original scalar type.
3267 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3269 * profile-count.h (enum profile_quality): Fix typos and whitespace
3272 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3274 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3275 type for branch probabilities.
3277 2017-06-29 Julian Brown <julian@codesourcery.com>
3278 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3280 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3281 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3282 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3283 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3285 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3287 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3288 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3289 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3290 CC usage from generic code to here.
3291 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3292 CC usage into the target macros.
3294 2017-06-29 Maya Rashish <coypu@sdf.org>
3296 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3299 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3301 * arm/arm-builtins.c: Include profile-count.h
3302 * except.c (sjlj_emit_function_enter): Use
3303 profile_probability::unlikely.
3305 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3307 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3308 and tocrel_offset be pointer args rather than implicitly using
3310 (legitimate_constant_pool_address_p, rs6000_emit_move,
3311 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3312 tocrel_offset and use in toc_relative_expr_p call.
3313 (print_operand, print_operand_address): Use static tocrel_base_oac
3314 and tocrel_offset_oac.
3315 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3318 2017-06-29 Maya Rashish <coypu@sdf.org>
3320 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3322 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3324 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3325 objects, take into account only the alignment of 'op0' and 'mode1' if
3328 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3330 * ccmp.c (ccmp_tree_comparison_p): New function.
3331 (ccmp_candidate_p): Update to use above function.
3332 (get_compare_parts): New function.
3333 (expand_ccmp_next): Update to use new functions.
3334 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3336 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3337 take mode as argument.
3338 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3339 * expr.c (expand_expr_real_1): Pass mode as argument.
3341 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3343 * combine.c (combine_instructions): Print insns to dump_file, together
3346 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3348 * asan.c (asan_emit_stack_protection): Update.
3349 (create_cond_insert_point): Update.
3350 * auto-profile.c (afdo_propagate_circuit): Update.
3351 * basic-block.h (struct edge_def): Turn probability to
3352 profile_probability.
3353 (EDGE_FREQUENCY): Update.
3354 * bb-reorder.c (find_traces_1_round): Update.
3355 (better_edge_p): Update.
3356 (sanitize_hot_paths): Update.
3357 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3358 (make_single_succ_edge): Update.
3359 (check_bb_profile): Update.
3360 (dump_edge_info): Update.
3361 (update_bb_profile_for_threading): Update.
3362 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3363 probabilitycount to 0.
3364 * cfgbuild.c (compute_outgoing_frequencies): Update.
3365 * cfgcleanup.c (try_forward_edges): Update.
3366 (outgoing_edges_match): Update.
3367 (try_crossjump_to_edge): Update.
3368 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3369 (expand_gimple_tailcall): Update.
3370 (construct_init_block): Use make_single_succ_edge.
3371 (construct_exit_block): Use make_single_succ_edge.
3372 * cfghooks.c (verify_flow_info): Update.
3373 (redirect_edge_succ_nodup): Update.
3374 (split_edge): Update.
3375 (account_profile_record): Update.
3376 * cfgloopanal.c (single_likely_exit): Update.
3377 * cfgloopmanip.c (scale_loop_profile): Update.
3378 (set_zero_probability): Remove.
3379 (duplicate_loop_to_header_edge): Update.
3380 * cfgloopmanip.h (loop_version): Update prototype.
3381 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3382 (force_nonfallthru_and_redirect): Update.
3383 (update_br_prob_note): Update.
3384 (rtl_verify_edges): Update.
3385 (purge_dead_edges): Update.
3386 (rtl_lv_add_condition_to_bb): Update.
3387 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3388 * cgraphunit.c (init_lowered_empty_function): Update.
3389 (cgraph_node::expand_thunk): Update.
3390 * cilk-common.c: Include profile-count.h
3391 * dojump.c (inv): Remove.
3392 (jumpifnot): Update.
3393 (jumpifnot_1): Update.
3394 (do_jump_1): Update.
3396 (do_jump_by_parts_greater_rtx): Update.
3397 (do_compare_rtx_and_jump): Update.
3398 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3399 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3400 * dwarf2cfi.c: Include profile-count.h
3401 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3402 (sjlj_emit_dispatch_table): Likewise.
3403 * explow.c: Include profile-count.h
3404 * expmed.c (emit_store_flag_force): Update.
3405 (do_cmp_and_jump): Update.
3406 * expr.c (compare_by_pieces_d::generate): Update.
3407 (compare_by_pieces_d::finish_mode): Update.
3408 (emit_block_move_via_loop): Update.
3409 (store_expr_with_bounds): Update.
3410 (store_constructor): Update.
3411 (expand_expr_real_2): Update.
3412 (expand_expr_real_1): Update.
3413 * expr.h (try_casesi, try_tablejump): Update prototypes.
3414 * gimple-pretty-print.c (dump_probability): Update.
3415 (dump_profile): New.
3416 (dump_gimple_label): Update.
3417 (dump_gimple_bb_header): Update.
3418 * graph.c (draw_cfg_node_succ_edges): Update.
3419 * hsa-gen.c (convert_switch_statements): Update.
3420 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3421 (find_if_case_1): Update.
3422 (find_if_case_2): Update.
3423 * internal-fn.c (expand_arith_overflow_result_store): Update.
3424 (expand_addsub_overflow): Update.
3425 (expand_neg_overflow): Update.
3426 (expand_mul_overflow): Update.
3427 (expand_vector_ubsan_overflow): Update.
3428 * ipa-cp.c (good_cloning_opportunity_p): Update.
3429 * ipa-split.c (split_function): Use make_single_succ_edge.
3430 * ipa-utils.c (ipa_merge_profiles): Update.
3431 * loop-doloop.c (add_test): Update.
3432 (doloop_modify): Update.
3433 * loop-unroll.c (compare_and_jump_seq): Update.
3434 (unroll_loop_runtime_iterations): Update.
3435 * lra-constraints.c (lra_inheritance): Update.
3436 * lto-streamer-in.c (input_cfg): Update.
3437 * lto-streamer-out.c (output_cfg): Update.
3438 * mcf.c (adjust_cfg_counts): Update.
3439 * modulo-sched.c (sms_schedule): Update.
3440 * omp-expand.c (expand_omp_for_init_counts): Update.
3441 (extract_omp_for_update_vars): Update.
3442 (expand_omp_ordered_sink): Update.
3443 (expand_omp_for_ordered_loops): Update.
3444 (expand_omp_for_generic): Update.
3445 (expand_omp_for_static_nochunk): Update.
3446 (expand_omp_for_static_chunk): Update.
3447 (expand_cilk_for): Update.
3448 (expand_omp_simd): Update.
3449 (expand_omp_taskloop_for_outer): Update.
3450 (expand_omp_taskloop_for_inner): Update.
3451 * omp-simd-clone.c (simd_clone_adjust): Update.
3452 * optabs.c (expand_doubleword_shift): Update.
3453 (expand_abs): Update.
3454 (emit_cmp_and_jump_insn_1): Update.
3455 (expand_compare_and_swap_loop): Update.
3456 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3457 * predict.c (predictable_edge_p): Update.
3458 (edge_probability_reliable_p): Update.
3459 (set_even_probabilities): Update.
3460 (combine_predictions_for_insn): Update.
3461 (combine_predictions_for_bb): Update.
3462 (propagate_freq): Update.
3463 (estimate_bb_frequencies): Update.
3464 (force_edge_cold): Update.
3465 * profile-count.c (profile_count::dump): Add missing space into dump.
3466 (profile_count::debug): Add newline.
3467 (profile_count::differs_from_p): Explicitly convert to unsigned.
3468 (profile_count::stream_in): Update.
3469 (profile_probability::dump): New member function.
3470 (profile_probability::debug): New member function.
3471 (profile_probability::differs_from_p): New member function.
3472 (profile_probability::differs_lot_from_p): New member function.
3473 (profile_probability::stream_in): New member function.
3474 (profile_probability::stream_out): New member function.
3475 * profile-count.h (profile_count_quality): Rename to ...
3476 (profile_quality): ... this one.
3477 (profile_probability): New.
3478 (profile_count): Update.
3479 * profile.c (compute_branch_probabilities): Update.
3480 * recog.c (peep2_attempt): Update.
3481 * sched-ebb.c (schedule_ebbs): Update.
3482 * sched-rgn.c (find_single_block_region): Update.
3483 (compute_dom_prob_ps): Update.
3484 (schedule_region): Update.
3485 * sel-sched-ir.c (compute_succs_info): Update.
3486 * stmt.c (struct case_node): Update.
3487 (do_jump_if_equal): Update.
3488 (get_outgoing_edge_probs): Update.
3489 (conditional_probability): Update.
3490 (emit_case_dispatch_table): Update.
3491 (expand_case): Update.
3492 (expand_sjlj_dispatch_table): Update.
3493 (emit_case_nodes): Update.
3494 * targhooks.c: Update.
3495 * tracer.c (better_p): Update.
3496 (find_best_successor): Update.
3497 * trans-mem.c (expand_transaction): Update.
3498 * tree-call-cdce.c: Update.
3499 * tree-cfg.c (gimple_split_edge): Upate.
3500 (move_sese_region_to_fn): Upate.
3501 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3502 * tree-eh.c (lower_resx): Upate.
3503 (cleanup_empty_eh_move_lp): Upate.
3504 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3505 * tree-inline.c (copy_edges_for_bb): Update.
3506 (copy_cfg_body): Update.
3507 * tree-parloops.c (gen_parallel_loop): Update.
3508 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3509 (gimple_gen_time_profiler): Update.
3510 * tree-ssa-dce.c (remove_dead_stmt): Update.
3511 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3512 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3513 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3514 (unloop_loops): Update.
3515 (try_peel_loop): Update.
3516 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3517 * tree-ssa-loop-split.c (connect_loops): Update.
3518 (split_loop): Update.
3519 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3520 (hoist_guard): Update.
3521 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3522 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3523 (value_replacement): Update.
3524 * tree-ssa-reassoc.c (branch_fixup): Update.
3525 * tree-ssa-tail-merge.c (replace_block_by): Update.
3526 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3527 (create_edge_and_update_destination_phis): Update.
3528 (compute_path_counts): Update.
3529 (recompute_probabilities): Update.
3530 (update_joiner_offpath_counts): Update.
3531 (freqs_to_counts_path): Update.
3532 (duplicate_thread_path): Update.
3533 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3534 (struct switch_conv_info): Update.
3535 (gen_inbound_check): Update.
3536 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3537 (vect_do_peeling): Update.
3538 (vect_loop_versioning): Update.
3539 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3540 (optimize_mask_stores): Update.
3541 * ubsan.c (ubsan_expand_null_ifn): Update.
3542 * value-prof.c (gimple_divmod_fixed_value): Update.
3543 (gimple_divmod_fixed_value_transform): Update.
3544 (gimple_mod_pow2): Update.
3545 (gimple_mod_pow2_value_transform): Update.
3546 (gimple_mod_subtract): Update.
3547 (gimple_mod_subtract_transform): Update.
3548 (gimple_ic): Update.
3549 (gimple_stringop_fixed_value): Update.
3550 (gimple_stringops_transform): Update.
3551 * value-prof.h: Update.
3553 2017-06-29 Carl Love <cel@us.ibm.com>
3555 * config/rs6000/rs6000-c.c: Add support for built-in functions
3556 vector signed int vec_signed (vector float);
3557 vector signed long long vec_signed (vector double);
3558 vector signed int vec_signed2 (vector double, vector double);
3559 vector signed int vec_signede (vector double);
3560 vector signed int vec_signedo (vector double);
3561 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3562 instruction generator.
3563 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3564 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3565 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3567 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3568 vunsignede_v2df): Add define_expands.
3569 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3570 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3571 VEC_UNSIGNEDO): Add definitions.
3572 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3573 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3574 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3575 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3576 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3577 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3578 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3579 vec_unsignede, vec_unsignedo): Add builtin defines.
3580 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3582 * doc/extend.texi: Update the built-in documentation file for the
3583 new built-in functions.
3585 2017-06-29 Richard Biener <rguenther@suse.de>
3587 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3588 reduction chains to LOOP_VINFO_REDUCTIONS.
3589 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3590 SLP reductions after processing reduction chains.
3592 2017-06-29 Nathan Sidwell <nathan@acm.org>
3594 * builtins.c (fold_builtin_FUNCTION): Use
3595 lang_hooks.decl_printable_name.
3597 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3600 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3601 with only one label.
3602 * stmt.c (expand_case): Assert NCASES is greater than one.
3604 2017-06-29 Richard Biener <rguenther@suse.de>
3606 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3608 (group_case_labels): Likewise.
3609 (find_taken_edge): Push sanity checking on val to workers...
3610 (find_taken_edge_cond_expr): ... here
3611 (find_taken_edge_switch_expr): ... and here, handle cases
3612 with just a default label.
3613 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3614 (group_case_labels): Likewise.
3615 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3616 group_case_labels does anything cleanup the CFG again.
3618 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3620 PR tree-optimization/81196
3621 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3622 exit condition comparing two IVs.
3624 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3626 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3627 profile to the dummy entry at the end of the list of architectures.
3628 * config/arm/arm-cpu-cdata.h: Regenerated.
3630 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3631 Michael Collison <michael.collison@arm.com>
3634 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3636 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3637 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3638 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3639 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3640 of shift when the shift amount is masked with constant equal to
3641 the size of the mode.
3642 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3645 2017-06-29 Martin Liska <mliska@suse.cz>
3647 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3649 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3652 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3655 (edge_growth_cache_entry::edge_growth_cache_entry): New
3657 (reset_edge_growth_cache): Update to use constructor.
3659 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3661 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3662 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3663 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3665 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3667 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3668 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3670 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3672 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3673 (*-linux-uclibc*): Add t-uclibc tmake_file.
3674 * config/t-musl: New.
3675 * config/t-uclibc: New.
3677 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3679 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3681 (gen_comm_data): Emit architectural setting of arch_prof.
3682 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3684 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3685 (armv8-m.base, armv8-m.main): Likewise.
3686 * arm-protos.h (arm_build_target): Add profile field.
3687 (arch_option): Likewise.
3688 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3690 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3691 arm_active_target.profile.
3693 2017-06-28 Richard Biener <rguenther@suse.de>
3696 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3697 TYPE_OVERFLOW_WRAPS.
3698 * match.pd (negate_expr_p): Likewise.
3699 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3700 fold_build2, not fold_binary.
3702 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3704 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3705 Convert memory address to Pmode.
3706 (aarch64_print_operand): Assert MEM operands are always Pmode.
3708 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3711 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3712 Remove redundant if.
3713 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3714 * config/arm/aarch-common-protos.h
3715 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3716 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3718 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3721 * multiple_target.c (create_dispatcher_calls): Set the default
3722 clone to be static, not public.
3724 2017-06-28 Richard Biener <rguenther@suse.de>
3726 * tree-vect-loop.c (vectorizable_reduction): Move special
3727 cond reduction IV var creation ...
3728 (vect_create_epilog_for_reduction): ... here. Remove induction_index
3729 parameter. Use STMT_VINFO_VECTYPE.
3730 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
3733 2017-06-28 Martin Liska <mliska@suse.cz>
3736 * ipa-visibility.c (non_local_p): Handle visibility.
3738 2017-06-28 Martin Liska <mliska@suse.cz>
3741 * common.opt: Add IntegerRange to various options.
3742 * opt-functions.awk (integer_range_info): New function.
3743 * optc-gen.awk: Add integer_range_info to cl_options struct.
3744 * opts-common.c (decode_cmdline_option): Handle
3745 CL_ERR_INT_RANGE_ARG.
3746 (cmdline_handle_error): Likewise.
3747 * opts.c (print_filtered_help): Show valid interval in
3748 when --help is provided.
3749 * opts.h (struct cl_option): Add range_min and range_max fields.
3750 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
3752 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
3754 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
3755 (x * C EQ/NE y * C): New transformation.
3757 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
3759 * genmultilib (combination_space): Accept '+' in option names.
3761 2017-06-28 Martin Liska <mliska@suse.cz>
3764 * asan.c (instrument_derefs): Bail out inner references
3765 that are hard register variables.
3767 2017-06-28 Jakub Jelinek <jakub@redhat.com>
3770 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
3771 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
3773 2017-06-28 Richard Biener <rguenther@suse.de>
3775 * tree-vectorizer.h (vect_get_vec_defs): Remove.
3776 (vect_get_slp_defs): Adjust.
3777 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
3779 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
3781 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3782 get_initial_defs_for_reduction instead of vect_get_vec_defs.
3783 (vectorizable_reduction): Adjust.
3784 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
3786 (vect_get_slp_defs): Likewise.
3787 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
3788 (vectorizable_bswap): Adjust.
3789 (vectorizable_call): Likewise.
3790 (vectorizable_conversion): Likewise.
3791 (vectorizable_assignment): Likewise.
3792 (vectorizable_shift): Likewise.
3793 (vectorizable_operation): Likewise.
3794 (vectorizable_store): Likewise.
3795 (vectorizable_condition): Likewise.
3796 (vectorizable_comparison): Likewise.
3798 2017-06-28 Michael Collison <michael.collison@arm.com>
3801 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
3803 (arm_gen_movmemqi): Removed unused variable 'i'.
3804 Convert 'for' loop into 'while' loop.
3805 (arm_expand_prologue): Remove last unnecessary set of insn.
3806 (thumb_pop): Remove unused variable 'pushed_words'.
3807 (thumb_exit): Remove last unnecessary set of regs_to_pop.
3809 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3811 * config/s390/predicates.md: Use s390_rel_address_ok_p.
3812 * config/s390/s390-protos.h: Add prototype of
3813 s390_rel_address_ok_p.
3814 * config/s390/s390.c (s390_got_symbol): New function.
3815 (s390_rel_address_ok_p): New function.
3816 (legitimize_pic_address): Use s390_rel_address_ok_p.
3817 (s390_load_got): Use s390_got_symbol.
3818 (s390_option_override): Issue error if
3819 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
3820 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
3822 * config/s390/s390.opt: New option mpic-data-is-text-relative.
3824 2017-06-27 Andrew Pinski <apinski@cavium.com>
3826 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
3827 (X * copysign (1.0, X)): New pattern.
3828 (X * copysign (1.0, -X)): New pattern.
3829 (copysign (-1.0, CST)): New pattern.
3831 2017-06-27 Joseph Myers <joseph@codesourcery.com>
3833 * genmultilib (combination_space): Remove variable.
3834 Validate reuse rules against regular expression for any sequence
3835 of multilib options in any order.
3837 2017-06-27 Michael Collison <michael.collison@arm.com>
3839 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
3840 call aarch64_split_simd_combine.
3841 * (aarch64_combine_internal<mode>): Delete pattern.
3842 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
3843 Allow register and subreg operands.
3845 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3847 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
3848 specific need, just fallback on defaults.
3849 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
3851 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3852 Olivier Hainque <hainque@adacore.com>
3854 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
3856 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
3857 targets. Pick a default if no particular attempt applied.
3858 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
3861 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3863 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
3864 (x86_64-wrs-vxworks7): Likewise.
3866 2017-06-27 Marek Polacek <polacek@redhat.com>
3869 * ubsan.c (instrument_null): Check get_base_address's result for null.
3871 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3873 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
3875 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3877 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
3878 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
3879 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
3880 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
3882 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
3883 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
3884 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
3885 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
3886 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
3887 BUILT_IN_FEUPDATEENV): New builtins.
3888 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
3889 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
3890 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
3891 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
3893 (builtin_structptr_types): Adjust size.
3894 * tree.c (builtin_structptr_types): Add four entries.
3896 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3897 Olivier Hainque <hainque@adacore.com>
3899 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
3900 (TLS_SYM): New local macro, forcing reference to __tls__ on
3901 link command lines for VxWorks 7 RTPs, triggering initialization
3903 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
3904 OS features TLS support, true for RTPs on VxWorks 7.
3905 * config/vxworks.c (vxworks_override_options): Setup emutls
3908 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3910 * predict.c (test_prediction_value_range): Use -1U instead of -1
3911 to avoid narrowing conversion warning.
3912 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
3913 to avoid narrowing conversion warning.
3914 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
3916 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
3918 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3920 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
3921 64bit configurations.
3922 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
3923 (SIZE_TYPE): Likewise.
3924 * config/vxworks.c (vxworks_emutls_var_fields): Use
3925 long_unsigned_type_node instead of unsigned_type_node as the offset
3926 field type, which is "pointer" mode in emutls.c.
3928 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3931 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
3934 * gimple-fold.c (replace_call_with_call_and_fold): Handle
3935 gimple_vuse copying separately from gimple_vdef copying.
3937 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3939 * value-prof.c (free_hist): Remove call to memset and the enclosing if
3942 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3943 Olivier Hainque <hainque@adacore.com>
3945 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
3946 for all vxworks7 targets.
3947 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
3948 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
3949 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
3950 variations for VX6/VX7 and 32/64bits later on in ...
3951 (VXWORKS_LIB_SPEC): Leverage new macros.
3952 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
3953 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
3955 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3957 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
3958 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
3960 2017-06-26 Carl Love <cel@us.ibm.com>
3962 * config/rs6000/rs6000-c.c: Add support for built-in functions
3963 vector bool char vec_reve (vector bool char);
3964 vector signed char vec_reve (vector signed char);
3965 vector unsigned char vec_reve (vector unsigned char);
3966 vector bool int vec_reve (vector bool int);
3967 vector signed int vec_reve (vector signed int);
3968 vector unsigned int vec_reve (vector unsigned int);
3969 vector bool long long vec_reve (vector bool long long);
3970 vector signed long long vec_reve (vector signed long long);
3971 vector unsigned long long vec_reve (vector unsigned long long);
3972 vector bool short vec_reve (vector bool short);
3973 vector signed short vec_reve (vector signed short);
3974 vector double vec_reve (vector double);
3975 vector float vec_reve (vector float);
3976 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
3977 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
3978 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
3979 (altivec_vreve): New pattern.
3980 * config/rs6000/altivec.h (vec_reve): New define.
3981 * doc/extend.texi (vec_rev): Update the built-in documentation file
3982 for the new built-in functions.
3984 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3986 PR tree-optimization/71815
3987 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
3989 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
3991 (slsr_process_phi): Likewise.
3992 (replace_uncond_cands_and_profitable_phis): Don't replace a
3993 multiply candidate with a stride of 1 (copy or cast).
3994 (phi_incr_cost): Call uses_consumed_by_stmt rather than
3996 (lowest_cost_path): Likewise.
3997 (total_savings): Likewise.
3999 2017-06-26 Richard Biener <rguenther@suse.de>
4002 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4003 Use def_builtin_pure for all gather builtins.
4005 2017-06-26 Richard Biener <rguenther@suse.de>
4007 PR tree-optimization/81203
4008 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4011 2017-06-26 Marek Polacek <polacek@redhat.com>
4014 * doc/invoke.texi: Document -Wmultistatement-macros.
4016 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4018 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4019 arm_neon_ok_no_float_abi effective target.
4021 2017-06-26 Richard Biener <rguenther@suse.de>
4023 PR tree-optimization/80928
4024 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4025 (copy_bbs): Set BB_DUPLICATED flag early.
4026 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4028 (execute_on_shrinking_pred): Likewise.
4029 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4030 BB_DUPLICATED blocks.
4031 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4032 iterate over all PHIs considering removal of *gsi.
4034 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4036 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4039 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4041 * config/rs6000/rs6000-string.c: (expand_block_clear,
4042 do_load_for_compare, select_block_compare_mode,
4043 compute_current_alignment, expand_block_compare,
4044 expand_strncmp_align_check, expand_strn_compare,
4045 expand_block_move, rs6000_output_load_multiple)
4046 Move functions related to string/block move/compare
4048 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4049 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4050 for this function which is now used in two files.
4051 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4052 * config.gcc: Add rs6000-string.o to extra_objs for
4053 targets powerpc*-*-* and rs6000*-*-*.
4055 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4058 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4059 32-bit, since indexed is not valid for DImode.
4060 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4061 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4062 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4063 (define_peephole2 for Altivec d-form store): Likewise.
4066 * multiple_target.c (create_dispatcher_calls): Only create the
4067 dispatcher call if the function is the default clone of a
4070 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4073 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4074 a call, force the call to not be a tail call.
4076 2017-06-23 Jeff Law <law@redhat.com>
4078 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4081 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4083 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4084 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4085 handling for early expansion of vector shifts (sl,sr,sra,rl).
4086 (builtin_function_type): Add vector shift right instructions
4087 to the unsigned argument list.
4089 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4091 rtl-optimizatoin/79286
4092 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4093 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4094 trap. PIC register plus a const unspec without offset can never trap.
4096 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4098 * tree.h (builtin_structptr_type): New type.
4099 (builtin_structptr_types): Declare new array.
4100 * tree.c (builtin_structptr_types): New array.
4101 (free_lang_data, build_common_tree_nodes): Use it.
4103 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4106 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4109 2017-06-22 Matt Turner <mattst88@gmail.com>
4111 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4112 Lake models to skylake case. Assume skylake for unknown
4113 models with clflushopt.
4115 2017-06-22 Jeff Law <law@redhat.com>
4117 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4118 frame sizes that do not satisfy aarch64_uimm12_shift.
4120 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4122 * profile-count.h (apply_probability,
4123 apply_scale, probability_in): Fix checks for zero.
4125 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4127 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4128 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4130 2016-06-22 Richard Biener <rguenther@suse.de>
4132 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4133 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4134 REDUC_MAX_EXPR support.
4135 (vectorizable_reduction): Likewise.
4136 (vect_create_epilog_for_reduction): Likewise.
4138 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4140 * match.pd (A / (1 << B) -> A >> B): New.
4141 * generic-match-head.c: Include optabs-tree.h.
4142 * gimple-match-head.c: Likewise.
4143 * optabs-tree.h (target_supports_op_p): New.
4144 * optabs-tree.c (target_supports_op_p): New.
4146 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4148 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4149 $gcc_cv_ld --help output.
4150 (gcc_cv_ld_demangle): Likewise.
4151 (gcc_cv_ld_eh_frame_hdr): Likewise.
4152 (gcc_cv_ld_pie): Likewise.
4153 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4154 (gcc_cv_ld_buildid): Likewise.
4155 (gcc_cv_ld_sysroot): Likewise.
4156 (ld_bndplt_support): Likewise.
4157 (ld_pushpopstate_support): Likewise.
4158 * configure: Regenerate.
4159 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4161 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4164 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4165 operands indexes to avoid gap between operands and match_dups.
4167 2017-06-21 Andrew Pinski <apinski@cavium.com>
4169 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4170 Increment Arith_shift and Arith_shift_reg by 1.
4171 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4173 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4174 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4175 (aarch64_strip_extend): Add new argument and test for it.
4176 (aarch64_cheap_mult_shift_p): New function.
4177 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4178 add a cost if it is true.
4179 Update calls to aarch64_strip_extend.
4180 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4182 2017-06-21 Andrew Pinski <apinski@cavium.com>
4184 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4186 (thunderxt88): Likewise.
4187 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4188 (thunderx_prefetch_tune): New variable.
4189 (thunderx2t99_prefetch_tune): Update for the correct values.
4190 (thunderxt88_tunings): New variable.
4191 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4192 generic_prefetch_tune.
4193 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4195 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4197 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4198 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4199 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4200 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4201 (aarch64_atomic_cas<mode>, GPI): Likewise.
4203 2017-06-21 Martin Liska <mliska@suse.cz>
4205 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4206 statements on cold and hot labels.
4207 * predict.c (tree_estimate_probability_bb): Remove the
4208 prediction from this place.
4210 2017-06-21 Martin Liska <mliska@suse.cz>
4212 PR tree-optimization/79489
4213 * gimplify.c (maybe_add_early_return_predict_stmt): New
4215 (gimplify_return_expr): Call the function.
4216 * predict.c (tree_estimate_probability_bb): Remove handling
4218 * predict.def: Update comment about early return predictor.
4219 * gimple-predict.h (is_gimple_predict): New function.
4220 * predict.def: Change default value of early return to 66.
4221 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4223 * passes.def: Put pass_strip_predict_hints to the beginning of
4226 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4228 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4229 FUNCTION_DECL declarations.
4230 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4232 (dwaf2out_decl): Likewise.
4233 * godump.c (go_early_global_decl): Skip call to the real debug hook
4234 for FUNCTION_DECL declarations.
4235 * passes.c (rest_of_decl_compilation): Skip call to the
4236 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4237 -fdump-go-spec is passed.
4239 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4241 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4242 Reorder for compactness.
4243 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4244 (def_builtin_pure, def_builtin_pure2): New functions.
4245 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4247 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4249 * match.pd (nop_convert): New predicate.
4250 ((A +- CST1) +- CST2): Allow some NOP conversions.
4252 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4255 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4256 with ctors/dtors if GOVD_SHARED is set.
4258 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4260 * config/aarch64/aarch64.md (movti_aarch64):
4261 Emit mov rather than orr.
4262 (movtf_aarch64): Likewise.
4263 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4264 Emit mov rather than orr.
4266 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4268 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4269 Swap alternatives, make integer dup more expensive.
4271 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4273 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4274 Return true for non-tls symbols.
4276 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4278 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4279 (cortex-a75): Likewise.
4280 (cortex-a75.cortex-a55): Likewise.
4281 * config/aarch64/aarch64-tune.md: Regenerate.
4282 * doc/invoke.texi (-mtune): Document new values for -mtune.
4284 2017-06-21 Tom de Vries <tom@codesourcery.com>
4286 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4288 (Effective-Target Keywords, Other attributes): Suggest using
4289 dg-add-options stack_size feature to get stack limit in stack_size
4290 effective target documentation.
4292 2017-06-21 Julian Brown <julian@codesourcery.com>
4293 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4295 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4296 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4297 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4299 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4300 attribute type list for neon_multiply.
4301 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4302 attribute type list for neon_multiply.
4303 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4304 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4305 attribute type list for neon_multiply.
4306 * config/arm/types.md (crypto_pmull): Add.
4307 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4308 attribute type list.
4310 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4312 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4315 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4317 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4318 to make sure not to dereference a NULL cost_classes_ptr pointer.
4320 2017-06-20 Carl Love <cel@us.ibm.com>
4322 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4323 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4324 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4325 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4326 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4327 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4328 VMULOSW): New enum "unspec" values.
4329 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4330 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4331 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4332 altivec_vmulosw): New patterns.
4333 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4334 VMULOSW): Add definitions.
4336 2017-06-20 Julia Koval <julia.koval@intel.com>
4338 * config/i386/i386.c: Fix rounding expand for new pattern.
4339 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4341 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4343 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4344 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4346 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4348 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4351 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4353 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4354 architecture, then by implementer ID.
4355 * config/aarch64/aarch64-tune.md: Regenerate.
4357 2017-06-20 Richard Biener <rguenther@suse.de>
4360 * fold-const.c (split_tree): Fold to type before negating.
4362 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4364 * diagnostic-show-locus.c
4365 (selftest::test_fixit_deletion_affecting_newline): New function.
4366 (selftest::diagnostic_show_locus_c_tests): Call it.
4368 2017-06-20 Andreas Schwab <schwab@suse.de>
4371 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4374 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4376 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4377 __ARM_FEATURE_COPROC according to support.
4379 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4381 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4382 Rewritten to avoid overflow for > 32-bit pointers.
4385 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4386 by removing enum keyword.
4387 (ubsan_type_descriptor): Likewise. Formatting fix.
4390 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4391 splitter): Require TARGET_SSE2 in the condition.
4393 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4396 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4397 for doing vector set of SFmode on ISA 3.0.
4398 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4399 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4401 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4402 SFmode value into a V4SF variable that was extracted from another
4403 V4SF variable without converting the element to double precision
4404 and back to single precision vector format.
4405 (vsx_insert_extract_v4sf_p9_2): Likewise.
4407 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4409 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4410 in UWHI to avoid undefined overflow.
4413 * ubsan.h (enum ubsan_encode_value_phase): New.
4414 (ubsan_encode_value): Change second argument to
4415 enum ubsan_encode_value_phase with default value of
4416 UBSAN_ENCODE_VALUE_GENERIC.
4417 * ubsan.c (ubsan_encode_value): Change second argument to
4418 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4419 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4420 create_tmp_var_raw instead of create_tmp_var and use a
4422 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4423 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4424 ubsan_encode_value callers.
4427 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4428 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4429 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4431 2017-06-19 Richard Biener <rguenther@suse.de>
4434 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4435 estimates if we changed anything.
4437 2017-06-19 Richard Biener <rguenther@suse.de>
4439 PR tree-optimization/80887
4440 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4441 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4442 simplified lookups, then reset mprts_hook.
4443 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4445 (try_to_simplify): Likewise.
4447 2017-06-19 Martin Liska <mliska@suse.cz>
4450 * gimplify.c (gimplify_switch_expr):
4451 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4453 2017-06-19 Martin Liska <mliska@suse.cz>
4455 * doc/install.texi: Document that PGO runs in 4 stages.
4457 2017-06-19 Martin Liska <mliska@suse.cz>
4460 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4461 to dispatcher function name.
4462 * multiple_target.c (replace_function_decl): New function.
4463 (create_dispatcher_calls): Redirect both edges and references.
4465 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4467 * profile-count.c (profile_count::dump): Dump quality.
4468 (profile_count::differs_from_p): Update for unsigned val.
4469 * profile-count.h (profile_count_quality): New enum.
4470 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4472 2017-06-19 Richard Biener <rguenther@suse.de>
4474 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4475 struct function as arg.
4476 (estimate_numbers_of_iterations): Export overload with loop arg.
4477 (free_numbers_of_iterations_estimates_loop): Use an overload of
4478 free_numbers_of_iterations_estimates instead.
4479 * tree-cfg.c (remove_bb): Adjust.
4480 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4481 * tree-parloops.c (gen_parallel_loop): Likewise.
4482 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4484 (tree_unroll_loops_completely): Likewise.
4485 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4486 Use an overload instead and export.
4487 (estimated_loop_iterations): Adjust.
4488 (max_loop_iterations): Likewise.
4489 (likely_max_loop_iterations): Likewise.
4490 (estimate_numbers_of_iterations): Take struct function as arg
4492 (loop_exits_before_overflow): Adjust.
4493 (free_numbers_of_iterations_estimates_loop): Use an overload.
4494 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4495 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4497 2017-06-19 Richard Biener <rguenther@suse.de>
4500 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4501 RANGE_EXPR conservatively.
4503 2017-06-16 Carl Love <cel@us.ibm.com>
4505 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4506 definitions for vec_float, vec_float2, vec_floato,
4507 vec_floate built-ins.
4508 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4509 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4511 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4512 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4513 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4514 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4515 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4516 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4517 vec_floato): Add builtin defines.
4518 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4519 Update the built-in documentation file for the new built-in
4522 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4524 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4525 (mthumb): Mark as the negative of -marm.
4527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4529 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4531 (ARM Options, -mtune): Document that this accepts the same
4532 extension options as -mcpu.
4533 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4535 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4537 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4538 permitted extensions.
4540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4542 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4543 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4544 (armv8-m.main): Add option +nodsp.
4545 * config/arm/arm-cpu-cdata.h: Regenerated.
4547 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4549 * config/arm/t-fuchsia: New file.
4550 * config.gcc (arm*-*-fuchsia*): Use it.
4552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4554 * config/arm/t-symbian: Rewrite for new option infrastructure.
4556 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4558 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4559 (MULTILIB_REQUIRED): Likewise.
4561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4563 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4564 (MULTILIB_RESUE): Likewise.
4565 (MULTILIB_MATCHES): Likewise.
4566 (MULTLIB_REQUIRED): Likewise.
4568 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4570 * config/arm/t-rtems: Rewrite for new option framework.
4572 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4574 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4575 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4576 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4577 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4578 * config/arm/t-multilib: ... here.
4579 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4580 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4581 armv7-a and armv8*-a when A-profile libraries have not been built.
4582 * config/arm/t-rmprofile: Rewrite.
4584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4586 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4587 with a backslash. Remove the backslash after substituting unescaped
4589 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4591 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4593 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4594 the driver to pass through the default setting of -mfloat-abi.
4595 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4597 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4598 (all_feat_combs): New rule.
4599 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4601 * config/arm/t-aprofile: Rewrite.
4603 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4605 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4606 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4607 fpu is not specified by the user/command-line.
4608 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4609 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4610 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4611 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4612 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4613 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4617 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4618 * config/arm/t-arm-elf: Rewritten.
4620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4622 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4623 have some floating-point instructions.
4624 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4625 (TARGET_MAYBE_HARD_FLOAT): New macro.
4626 * config/arm/arm-builtins.c (arm_init_builtins): Use
4627 TARGET_MAYBE_HARD_FLOAT.
4628 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4630 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4632 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4633 (configargs.h): Include it.
4634 (arm_print_hint_for_fpu_option): New function.
4635 (arm_parse_fpu_option): New function.
4636 (candidate_extension): New class.
4637 (arm_canon_for_multilib): New function.
4638 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4639 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4640 (ARCH_CANONICAL_SPECS): New macro.
4641 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4645 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4646 are set after handling multilib fragments. Set target_cpu_default2
4649 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4651 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4653 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4654 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4656 (check_arch): Likewise.
4657 * config/arm/arm.c (arm_configure_build_target): Handle
4658 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4659 options in the default.
4661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4663 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4664 when an option is an alias of another.
4665 * config/arm/parsecpu.awk (optalias): New parser token.
4666 (gen_comm_data): Mark non-alias options as such. Emit entries
4667 for extension aliases.
4668 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4669 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4670 (armv6kz, armv6zk, armv6t2): Likewise.
4671 (armv7): Make vfpv3-d16 an alias.
4672 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4674 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4675 Sort in canonical order.
4676 (armv8-a): Sort in canonical order.
4677 (armv8.1-a, armv8.2-a): Likewise.
4678 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4680 (cortex-a9): Sort in canonical order.
4681 * config/arm/arm.c (selftests.h): Include it.
4682 (arm_test_cpu_arch_data): New function.
4683 (arm_run_self_tests): New function.
4684 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4685 (targetm): Move declaration to the end of the file.
4686 * arm-cpu-cdata.h: Regenerated.
4688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4690 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4691 call to target_mode_check describing the type of option passed.
4692 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4693 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4694 arm_parse_cpu_option_name to match parameters against list of
4696 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4697 arm_arch_core_flags data structure.
4698 * config/arm/arm-cpu_cdata.h: Regenerated.
4700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4702 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4704 (arm_print_hint_for_cpu_option): Likewise.
4705 (arm_print_hint_for_arch_option): Likewise.
4706 (arm_parse_cpu_option_name): Likewise.
4707 (arm_parse_arch_option_name): Likewise.
4708 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4709 of entries in the all_fpus list.
4710 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4711 (arm_parse_cpu_option_name): Declare.
4712 (arm_parse_arch_option_name): Declare.
4713 (arm_parse_option_features): Declare.
4714 (arm_intialize_isa): Declare.
4715 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4717 (gen_comm_data): ... here. Make definitions non-static.
4718 * config/arm/arm-cpu-data.h: Regenerated.
4719 * config/arm/arm-cpu-cdata.h: Regenerated.
4721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4723 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
4724 (cpu_arch_extension): New structure.
4725 (cpu_arch_option, arch_option, cpu_option): New structures.
4726 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
4728 (gen_data): Generate new format data tables.
4729 * config/arm/arm.c (cpu_tune): New structure.
4730 (cpu_option, processors): Delete.
4731 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
4732 (arm_print_hint_for_cpu_option): ... this and ...
4733 (arm_print_hint_for_arch_option): ... this.
4734 (arm_parse_arch_cpu_name): Delete. Replace with ...
4735 (arm_parse_cpu_option_name): ... this and ...
4736 (arm_parse_arch_option_name): ... this.
4737 (arm_unrecognized_feature): Change type of target parameter to
4739 (arm_parse_arch_cpu_features): Delete. Replace with ...
4740 (arm_parse_option_features): ... this.
4741 (arm_configure_build_target): Rework to use new configuration data
4743 (arm_print_tune_info): Rework for new configuration data tables.
4744 * config/arm/arm-cpu-data.h: Regenerated.
4745 * config/arm/arm-cpu.h: Regenerated.
4747 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4749 * Makefile.in (OBJS): Move sbitmap.o from here ...
4750 (OBJS-libcommon): ... to here.
4752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4754 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
4755 (ISA_ALL_CRYPTO): New macro.
4756 (ISA_ALL_SIMD): New macro
4757 (ISA_ALL_FP): New macro.
4758 * config/arm/arm.c (fpu_bitlist): Update initializer.
4759 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
4761 (arm9e): Add fpu. Add option for nofp
4762 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
4763 (arm926ej-s, arm1026ej-s): Likewise.
4764 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
4765 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
4766 neon-fp16, neon-vfpv4, nofp and nosimd.
4767 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
4768 (cortex-a8): Add fpu. Add option for nofp.
4769 (cortex-a9): Add fpu. Add options for nosimd and nofp.
4770 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
4771 (cortex-r4f): Add fpu.
4772 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
4773 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
4775 (cortex-r8): Likewise.
4776 (cortex-m4): Add fpu. Add option for nofp.
4777 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
4778 (cortex-a17.cortex-a7): Likewise.
4779 (cortex-a32): Add fpu. Add options for crypto and nofp.
4780 (cortex-a35, cortex-a53): Likewise.
4781 (cortex-a57): Add fpu. Add option for crypto.
4782 (cortex-a72, cortex-a73): Likewise.
4783 (exynos-m1): Likewise.
4784 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
4785 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
4786 (cortex-m33): Add fpu. Add option for nofp.
4787 * config/arm/arm-cpu-cdata.h: Regenerated
4788 * config/arm/arm-cpu-data.h: Regenerated.
4790 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4792 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
4793 (armv5te, armv5tej): Likewise.
4794 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
4795 (armv7): Add options fp and vfpv3-d16.
4796 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
4797 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
4799 (armv7ve): Likewise.
4800 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
4801 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
4802 (armv8-a): Add nocrypto option.
4803 (armv8.1-a, armv8.2-a): Likewise.
4804 (armv8-m.main): add options fp, fp.dp and nofp.
4806 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4808 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
4810 (armv8-a+crc): Delete.
4811 (armv8.1-a): Add options simd, crypto and nofp.
4812 (armv8.2-a): Add options fp16, simd, crypto and nofp.
4813 (armv8.2-a+fp16): Delete.
4814 (armv8-m.main): Add option dsp.
4815 (armv8-m.main+dsp): Delete.
4816 (cortex-a8): Add fpu. Add nofp option.
4817 (cortex-a9): Add fpu. Add nofp and nosimd options.
4818 * config/arm/parsecpu.awk (gen_data): Generate option tables and
4819 link to main cpu and architecture data structures.
4820 (gen_comm_data): Only put isa attributes from the main architecture
4822 (option): New statement for architecture and CPU entries.
4823 * arm.c (struct cpu_option): New structure.
4824 (struct processors): Add entry for options.
4825 (arm_unrecognized_feature): New function.
4826 (arm_parse_arch_cpu_name): Ignore any characters after the first
4828 (arm_parse_arch_cpu_feature): New function.
4829 (arm_configure_build_target): Separate out any CPU and architecture
4830 features and parse separately. Don't error out if -mfpu=auto is
4831 used with only an architecture string.
4832 (arm_print_asm_arch_directives): New function.
4833 (arm_file_start): Call it.
4834 * config/arm/arm-cpu-cdata.h: Regenerated.
4835 * config/arm/arm-cpu-data.h: Likewise.
4836 * config/arm/arm-tables.opt: Likewise.
4838 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4840 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
4841 assembler when it is not -mfpu=auto.
4843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4845 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
4846 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
4847 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
4848 (ASM_CPU_SPEC): Rewrite.
4849 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
4850 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
4851 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
4852 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
4853 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
4854 copied string is NUL-terminated. Also strip any characters prefixed
4856 (arm_rewrite_selected_arch): New function.
4857 (arm_rewrite_march): New function.
4859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4861 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
4862 (x_arm_cpu_string, x_arm_tune_string): Likewise.
4863 (march, mcpu, mtune): Convert to string-based options.
4864 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
4865 (arm_parse_arch_cpu_name): New function.
4866 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
4867 identify selected architecture or CPU.
4868 (arm_option_save): New function.
4869 (TARGET_OPTION_SAVE): Redefine.
4870 (arm_option_restore): Restore string options.
4871 (arm_option_print): Print string options.
4873 2017-06-16 Martin Sebor <msebor@redhat.com>
4875 PR tree-optimization/80933
4876 PR tree-optimization/80934
4877 * builtins.c (fold_builtin_3): Do not handle bcmp here.
4878 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
4879 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
4880 (gimple_fold_builtin): Call them.
4882 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4884 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
4885 as unlikely; update profile.
4887 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4889 * predict.c (force_edge_cold): Handle declaring edges impossible
4892 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4894 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
4896 (try_unroll_loop_completely): Fix reporting.
4898 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4900 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
4902 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
4905 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
4906 if given a non-constant argument for an intrinsic which requires a
4909 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4911 * profile.c (compare_freqs): New function.
4912 (branch_prob): Sort edge list.
4913 (find_spanning_tree): Assume that the list is priority sorted.
4915 2017-06-16 Richard Biener <rguenther@suse.de>
4917 PR tree-optimization/81090
4918 * passes.def (pass_record_bounds): Remove.
4919 * tree-pass.h (make_pass_record_bounds): Likewise.
4920 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
4921 make_pass_record_bounds): Likewise.
4922 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
4923 not free niter estimates at the beginning but at the end.
4924 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
4926 2017-06-16 Richard Biener <rguenther@suse.de>
4928 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
4929 initializer to workaround ICE in host GCC 4.8.
4931 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4933 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
4935 (clone_inlined_nodes): Update.
4937 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4939 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
4940 prefetch settings, and enable prefetching by default at -O3.
4942 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4944 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4945 Set flag_prefetch_loop_arrays according to tuning data.
4947 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4949 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
4951 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
4952 [Unrelated to main purpose of the patch] Place the pointer field last
4953 to enable type checking errors when tune structure are wrongly merged.
4954 * config/aarch64/aarch64.c (generic_prefetch_tune,)
4955 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
4956 (thunderx2t99_prefetch_tune): New tune constants.
4957 (tune_params *_tunings): Update all tunings (no functional change).
4958 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
4959 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
4960 from tunings structures.
4962 2017-06-16 Jakub Jelinek <jakub@redhat.com>
4965 * ubsan.c (instrument_null): Add T argument, use it instead
4966 of computing it based on IS_LHS.
4967 (instrument_object_size): Likewise.
4968 (pass_ubsan::execute): Adjust instrument_null and
4969 instrument_object_size callers to pass gimple_get_lhs or
4970 gimple_assign_rhs1 result to it. Use instrument_null instead of
4971 calling get_base_address and instrument_mem_ref. Handle
4972 aggregate call arguments for object-size sanitization.
4974 2017-06-16 Yury Gribov <tetra2005@gmail.com>
4976 PR tree-optimization/81089
4977 * tree-vrp.c (is_masked_range_test): Validate operands of
4980 2017-06-15 Martin Sebor <msebor@redhat.com>
4983 * dumpfile.c (dump_register): Avoid calling memset to initialize
4984 a class with a default ctor.
4985 * gcc.c (struct compiler): Remove const qualification.
4986 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
4987 * hash-table.h: Ditto.
4988 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
4990 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
4991 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
4993 * params.h (struct param_info): Make struct members non-const.
4994 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
4995 with default initialization.
4996 * vec.h (vec_copy_construct, vec_default_construct): New helper
4998 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
4999 with vec_copy_construct.
5000 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5001 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5002 * doc/invoke.texi (-Wclass-memaccess): Document.
5004 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5006 * emit-rtl.h (is_leaf): Update comment about local
5009 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5012 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5013 for a variable to have a section before checking if the section has a
5015 Set section to.persistent if persistent attribute is set.
5016 Warn if .persistent attribute is used on an automatic variable.
5018 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5020 PR rtl-optimization/80474
5021 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5023 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5025 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5028 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5030 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5031 (casesi): Emit bounds checking as RTL.
5032 (casesi_internal_mips16_<mode>): Remove bounds checking.
5034 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5036 * config/xtensa/xtensa.c (xtensa_option_override): Append
5037 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5038 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5039 xtensa_doloop_hooks): Define unconditionally.
5040 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5042 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5043 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5044 for it in xtensa_option_override.
5045 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5046 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5048 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5050 * doc/cppopts.texi: Document '-' special value to -MF.
5052 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5054 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5055 (cortex_a53_fconst): Likewise.
5056 (cortex_a53_fpmul): Likewise.
5057 (cortex_a53_f_load_64): Likewise.
5058 (cortex_a53_f_load_many): Likewise.
5059 (cortex_a53_advsimd_alu): Likewise.
5060 (cortex_a53_advsimd_alu_q): Likewise.
5061 (cortex_a53_advsimd_mul): Likewise.
5062 (cortex_a53_advsimd_mul_q): Likewise.
5063 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5064 Add missing fmul, r2f_cvt and fconst cases.
5066 2017-06-14 Richard Biener <rguenther@suse.de>
5069 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5071 (fold_binary_loc): When associating do not treat pre-existing
5072 TREE_OVERFLOW on literal constants as a reason to allow
5073 TREE_OVERFLOW on associated literal constants.
5075 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5077 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5078 (MASK_FEATURES): New macro.
5079 * config/sparc/sparc.c (sparc_option_override): Remove the special
5080 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5082 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5084 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5085 a division of 0 if non-call exceptions are enabled.
5087 2017-06-14 Andrew Pinski <apinski@cavium.com>
5088 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5091 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5092 Improve vector initialization code gen for only variable case.
5094 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5096 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5098 2017-06-14 Richard Biener <rguenther@suse.de>
5100 PR tree-optimization/81083
5101 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5104 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5106 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5107 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5108 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5109 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5110 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5111 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5113 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5115 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5116 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5118 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5120 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5122 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5124 * config/rs6000/t-rtems: Don't handle SPE.
5126 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5128 * config/rs6000/t-linux: Don't handle SPE.
5130 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5132 * config/rs6000/eabispe.h: Delete file.
5134 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5136 * config/rs6000/t-spe: Delete file.
5138 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5140 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5141 (rs6000_legitimate_offset_address_p): Return false for anything in
5142 V2SImode or V2SFmode.
5144 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5146 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5147 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5148 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5150 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5151 (rs6000_legitimate_offset_address_p): Ditto.
5152 (rs6000_emit_move): Ditto.
5153 (rs6000_init_builtins): Remove V4HI_type_node.
5155 2017-06-13 Martin Liska <mliska@suse.cz>
5158 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5159 (gate_asan): Likewise.
5160 * asan.h (asan_no_sanitize_address_p): Remove the function.
5161 (sanitize_flags_p): New function.
5162 * builtins.def: Fix coding style.
5163 * common.opt: Use renamed enum value.
5164 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5165 * doc/extend.texi: Document no_sanitize attribute.
5166 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5167 to SANITIZE_UNDEFINED_NONDEFAULT.
5168 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5169 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5170 Use sanitize_flags_p.
5171 * gimplify.c (gimplify_function_tree): Likewise.
5172 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5173 * opts.c (parse_no_sanitize_attribute): New function.
5174 (common_handle_option): Use renamed enum value.
5175 * opts.h (parse_no_sanitize_attribute): Declare.
5176 * tree.c (sanitize_flags_p): New function.
5177 * tree.h: Declared here.
5178 * tsan.c: Use sanitize_flags_p.
5179 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5180 (instrument_mem_ref): Likewise.
5181 (instrument_bool_enum_load): Likewise.
5182 (do_ubsan_in_current_function): Remove the function.
5183 (pass_ubsan::execute): Use sanitize_flags_p.
5184 * ubsan.h: Remove do_ubsan_in_current_function
5185 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5186 (dump_function_to_file): Use it here.
5188 2017-06-13 Martin Jambor <mjambor@suse.cz>
5190 PR tree-optimization/80803
5191 PR tree-optimization/81063
5192 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5193 (propagate_subaccesses_across_link): Enqueue subtree whenever
5194 necessary instead of relying on the caller.
5196 2017-06-13 Martin Jambor <mjambor@suse.cz>
5198 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5199 that have a first_link.
5200 (sort_and_splice_var_accesses): Do not check first_link before
5202 (subtree_mark_written_and_enqueue): Likewise.
5203 (propagate_all_subaccesses): Likewise and do not stop at first
5204 parent with a first_link.
5206 2017-06-13 Martin Jambor <mjambor@suse.cz>
5208 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5211 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5213 * match.pd: New pattern.
5215 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5217 * tree-vrp.c (is_masked_range_test): New function.
5218 (register_edge_assert_for): Determine ranges for
5221 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5223 PR tree-optimization/67328
5224 * fold-const.c (maskable_range_p): New function.
5225 (build_range_check): Generate bittests if possible.
5227 2017-06-13 Martin Liska <mliska@suse.cz>
5229 * gimple-pretty-print.c (dump_probability): Add new argument.
5230 (dump_edge_probability): Dump both probability and count.
5231 (dump_gimple_label): Likewise.
5232 (dump_gimple_bb_header): Likewise.
5234 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5237 * config/avr/avr-devices.c: Fix indentation.
5238 * config/avr/gen-avr-mmcu-specs.c: Dito.
5240 2017-06-13 Richard Biener <rguenther@suse.de>
5242 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5243 instead get vector type from stmt_info.
5244 (vectorizable_reduction): Adjust. Remove dead code.
5246 2017-06-13 Richard Biener <rguenther@suse.de>
5249 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5250 case of C * (x * C2 + C3).
5251 (fold_addr_of_array_ref_difference): Properly fold index difference.
5253 2017-06-12 David S. Miller <davem@davemloft.net>
5256 * config/sparc/sparc.md (return expander): Emit frame blockage if
5257 function uses alloca.
5259 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5261 * combine.c (make_field_assignment): Check len rather than the mode
5262 precision when calling force_to_mode.
5264 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5266 Support multilibs and devices that see flash in RAM address range.
5269 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5270 (avr_mcu_t) <flash_pm_offset>: New field.
5271 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5272 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5273 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5274 (AVR_TINY_PM_OFFSET): Remove macro.
5275 * config/avr/avr.opt (-mshort-calls): New option.
5276 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5277 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5278 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5279 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5280 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5281 instead of avr_arch->have_jmp_call.
5282 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5283 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5284 avr_arch->flash_pm_offset to define.
5285 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5286 new field flash_pm_offset. Add entry for avrxmega3.
5287 (avr_texinfo): Add entry for avrxmega3.
5288 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5289 attiny212, attiny214,
5290 attiny412, attiny414, attiny416, attiny417,
5291 attiny814, attiny816, attiny817,
5292 attiny1614, attiny1616, attiny1617,
5293 attiny3214, attiny3216, attiny3217.
5294 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5295 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5296 (avr_print_operand_address) [AVR_TINY]: Same.
5297 (avr_asm_init_sections) <readonly_data_section>: Only patch
5298 callback if avr_arch->flash_pm_offset = 0.
5299 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5300 for rodata if avr_arch->flash_pm_offset != 0.
5301 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5302 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5303 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5304 (m_options): Append opt_rcall.
5305 (m_dirnames): Append dir_rcall.
5306 * config/avr/t-multilib: Regenerate.
5308 * configure.ac [target=avr]: Check whether avrxmega3 default
5309 linker description file works as needed.
5310 * configure: Regenerate.
5311 * doc/avr-mmcu.texi: Regenerate.
5312 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5313 <__AVR_ARCH__>: Document avrxmega3 and 103.
5314 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5315 <__AVR_SHORT_CALLS__>: Document it.
5316 <__AVR_PM_BASE_ADDRESS__>: Document it.
5317 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5318 (AVR Variable Attributes) <progmem>: Document this is
5319 not needed for avrxmega3.
5320 (AVR Named Address Spaces) <__flash>: Dito.
5322 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5324 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5326 2017-06-12 Doug Rupp <rupp@adacore.com>
5328 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5329 Append vxworks-stdint.h to the tm_file list.
5330 * config/vxworks-stdint.h: New file.
5332 2017-06-12 Martin Liska <mliska@suse.cz>
5334 PR tree-optimization/81041
5335 * tree-profile.c (gimple_gen_ic_func_profiler):
5336 Create an extra BB in profile-generate
5337 (gimple_gen_time_profiler): Likewise.
5339 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5341 PR tree-optimization/81003
5342 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5343 (update_range_test): Use it instead of force_gimple_operand_gsi.
5345 2017-06-12 Richard Biener <rguenther@suse.de>
5347 PR tree-optimization/81053
5348 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5349 with backedge value not defined in loop. Simplify def stmt
5352 2017-06-11 Tom de Vries <tom@codesourcery.com>
5355 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5357 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5358 nvptx_cannot_force_const_mem.
5360 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5362 * opts.c (finish_options): Move test for flag_split_stack after
5363 it has been initialized.
5365 2017-06-11 Jason Merrill <jason@redhat.com>
5367 * tree.h (id_equal): New.
5368 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5369 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5370 instead of strcmp of IDENTIFIER_POINTER.
5372 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5374 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5375 (mark_all_inlined_calls_cdtor): Fix formating.
5376 (inline_transform): Rescale profile before inlining.
5378 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5380 * cgraph.h (cgraph_edge::clone): Update prototype.
5381 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5382 (cgraph_node::create_clone): Update.
5383 (cgraph_node::create_version_clone): Update.
5384 * tree-inline.c (copy_bb): Update.
5385 (expand_call_inline): Update.
5387 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5389 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5390 factored out from ...
5391 (rs6000_emit_prologue): ... here.
5393 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5395 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5396 factored out from ...
5397 (rs6000_emit_prologue): ... here.
5399 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5401 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5403 (handle_missing_profiles): Fix computation of tp_first_run.
5404 (counts_to_freqs): Do not touch freqs when count is 0.
5406 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5408 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5411 2017-06-10 Tom de Vries <tom@codesourcery.com>
5413 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5414 attributes): Document signal effective target.
5416 2017-06-10 Tom de Vries <tom@codesourcery.com>
5418 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5419 Document effective target stack_size.
5421 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5423 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5424 to the edit_context if they can be auto-applied.
5426 2017-06-9 Ian Lance Taylor <iant@golang.org>
5428 * opts.c (finish_options): If -fsplit-stack, disable implicit
5429 -forder-blocks-and-partition.
5430 * doc/invoke.texi (Optimize Options): Document that when using
5431 -fsplit-stack -forder-blocks-and-partition is not implicitly
5434 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5436 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5437 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5438 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5439 * builtins.def (abort, trap, unreachable): Declare cold.
5440 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5441 * tree-core.h (ECF_COLD): New.
5442 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5443 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5445 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5447 * predict.c (unlikely_executed_stmt_p): Cleanup.
5449 2017-06-09 Richard Biener <rguenther@suse.de>
5451 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5452 model if the ref is always written to.
5454 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5456 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5458 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5460 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5463 2017-06-09 Tom de Vries <tom@codesourcery.com>
5466 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5467 "target cannot support label values" when encountering LABEL_REF.
5469 2017-06-09 Martin Liska <mliska@suse.cz>
5471 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5472 (gimple_gen_ic_func_profiler): Emit direct comparison
5473 of __gcov_indirect_call_callee with NULL.
5474 (gimple_gen_time_profiler): Change probability from
5475 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5477 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5479 * profile.c (edge_gcov_counts): Turn to pointer.
5480 (compute_branch_probabilities, compute_branch_probabilities): Update.
5481 (branch_prob): Do not clear edge_gcov_count.
5482 * profile.h (edge_gcov_counts): Turn to pointer.
5483 (edge_gcov_count): Update.
5485 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5487 * gimple.h (gimple_check_failed): Mark cold.
5489 2017-06-09 Richard Biener <rguenther@suse.de>
5491 PR tree-optimization/66623
5492 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5493 refactor check_reduction into two parts, properly computing
5494 whether we have to check reduction validity for outer loop
5497 2017-06-09 Richard Biener <rguenther@suse.de>
5499 PR tree-optimization/79483
5500 * graphite-scop-detection.c (order): New global.
5501 (get_order): Compute bb to order mapping that satisfies code
5502 generation constraints.
5503 (cmp_pbbs): New helper.
5504 (build_scops): Start domwalk at entry block, sort generated
5507 2017-06-09 Richard Biener <rguenther@suse.de>
5510 * ipa-polymorphic-call.c
5511 (ipa_polymorphic_call_context::restrict_to_inner_class):
5512 Skip FIELD_DECLs with error_mark_node type.
5513 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5516 2017-06-09 Martin Liska <mliska@suse.cz>
5518 * predict.c (struct branch_predictor): New struct.
5519 (test_prediction_value_range): New test.
5520 (predict_c_tests): New function.
5521 * selftest-run-tests.c (selftest::run_tests): Run the function.
5522 * selftest.h: Declare new tests.
5524 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5527 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5528 gen_add3_insn did not fail.
5529 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5530 r0, construct that number in a temporary reg and add that reg to r0.
5531 If asked to put the result in r0 as well, fail.
5533 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5535 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5536 for early expansion of vec_eqv.
5538 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5541 * ubsan.c (instrument_null): Avoid pointless code temporary.
5542 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5544 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5548 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5551 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5554 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5556 * predict.c (unlikely_executed_edge_p): Move ahead.
5557 (probably_never_executed_edge_p): Use it.
5559 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5562 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5563 gimple_call_builtin_p call.
5565 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5567 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5568 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5569 rtl_check_failed_type2, rtl_check_failed_code1,
5570 rtl_check_failed_code2, rtl_check_failed_code_mode,
5571 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5572 rtvec_check_failed_bounds, rtl_check_failed_flag,
5573 _fatal_insn_not_found, _fatal_insn): Likewise.
5574 * tree.h (tree_contains_struct_check_failed,
5575 tree_check_failed, tree_not_check_failed,
5576 tree_class_check_failed, tree_range_check_failed,
5577 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5578 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5579 tree_operand_check_failed, omp_clause_check_failed,
5580 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5583 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5585 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5586 flag_branch_probabilities.
5587 * ipa-inline.c (edge_badness): Likewise.
5588 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5589 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5590 * predict.c (maybe_hot_frequency_p): Likewise.
5591 (probably_never_executed): Likewise.
5592 * sched-ebb.c (schedule_ebbs): Likewise.
5593 * sched-rgn.c (find_single_block_region): Likewise.
5594 * tracer.c (tail_duplicate): Likewise.
5596 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5598 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5599 longer requires x_flag_profile_use.
5601 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5603 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5604 instead of flag_reorder_blocks_and_partition.
5605 * dbxout.c (dbxout_function_end): Likewise.
5606 * dwarf2out.c (gen_subprogram_die): Likewise.
5607 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5608 * hw-doloop.c (reorg_loops): Likewise.
5609 * varasm.c (assemble_start_function,
5610 assemble_end_function): Likewise.
5611 (decide_function_section): Do not check for
5612 flag_reorder_blocks_and_partition.
5614 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5616 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5618 (chkp_get_hard_register_fake_addr_expr): Ditto.
5619 (chkp_build_addr_expr): Add check for hard reg case.
5620 (chkp_parse_array_and_component_ref): Ditto.
5621 (chkp_find_bounds_1): Ditto.
5622 (chkp_process_stmt): Don't generate bounds store for
5625 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5627 * predict.c (maybe_hot_bb_p): Do not check profile status.
5628 (maybe_hot_edge_p): Likewise.
5629 (probably_never_executed): Check for zero counts even if profile
5631 (unlikely_executed_edge_p): New function.
5632 (unlikely_executed_stmt_p): New function.
5633 (unlikely_executed_bb_p): New function.
5634 (set_even_probabilities): Use unlikely predicates.
5635 (combine_predictions_for_bb): Likewise.
5636 (predict_paths_for_bb): Likewise.
5637 (predict_paths_leading_to_edge): Likewise.
5638 (determine_unlikely_bbs): New function.
5639 (estimate_bb_frequencies): Use it.
5640 (compute_function_frequency): Use zero counts even if profile is
5642 * profile-count.h: Fix typo.
5644 2017-08-08 Julia Koval <julia.koval@intel.com>
5646 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5647 _mm512_mask_cvtsepi16_storeu_epi8,
5648 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5649 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5650 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5651 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5652 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5653 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5654 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5655 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5656 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5657 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5658 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5659 __builtin_ia32_pmovuswb256mem_mask,
5660 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5661 __builtin_ia32_pmovwb512mem_mask): New builtins.
5663 2017-08-08 Julia Koval <julia.koval@intel.com>
5665 PR target/73350,80862
5666 * config/i386/subst.md (round): Fix round pattern.
5667 * config/i386/i386.c (ix86_erase_embedded_rounding):
5668 Fix erasing rounding for the fixed pattern.
5670 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5672 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5674 2017-06-08 Martin Liska <mliska@suse.cz>
5676 PR gcov-profile/80911
5677 * gcov.c (block_info::block_info): New constructor.
5679 2017-06-07 Carl Love <cel@us.ibm.com>
5681 * config/rs6000/rs6000-c: The return type of the following
5682 built-in functions was implemented as int not long long. Fix sign
5683 of return value for the unsigned version of vec_mulo and vec_mule.
5684 vector unsigned long long vec_bperm (vector unsigned long long,
5685 vector unsigned char)
5686 vector signed long long vec_mule (vector signed int,
5688 vector unsigned long long vec_mule (vector unsigned int,
5689 vector unsigned int)
5690 vector signed long long vec_mulo (vector signed int,
5692 vector unsigned long long vec_mulo (vector unsigned int,
5693 vector unsigned int)
5694 * doc/extend.texi: Fix the documentation for the built-in
5697 2017-06-07 Carl Love <cel@us.ibm.com>
5700 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5703 2017-06-07 Carl Love <cel@us.ibm.com>
5705 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5706 support, Generate doublehv for signed int/float for BE case only.
5708 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5710 * doc/invoke.texi (mcx16): Rewrite.
5712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5714 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5715 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5716 *mov<mode>_softfloat, and an anonymous splitter): Use
5717 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5719 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5721 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
5723 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5724 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
5725 (rs6000_debug_reg_global): Adjust.
5726 (rs6000_init_hard_regno_mode_ok): Adjust.
5727 (rs6000_dbx_register_number): Adjust.
5728 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
5729 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5730 Remove SPE_ACC and SPEFSCR.
5731 (REG_ALLOC_ORDER): Ditto.
5732 (FRAME_POINTER_REGNUM): Change to 111.
5733 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
5734 (REG_CLASS_NAMES): Ditto.
5735 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
5736 (REGISTER_NAMES): Ditto.
5737 (ADDITIONAL_REG_NAMES): Ditto.
5738 (rs6000_reg_names): Ditto.
5739 * config/rs6000/rs6000.md: Renumber some register number
5742 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5744 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
5746 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5747 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
5749 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
5750 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5751 Delete the SPE high registers.
5752 (REG_ALLOC_ORDER): Ditto.
5753 (enum reg_class): Remove SPE_HIGH_REGS.
5754 (REG_CLASS_NAMES): Ditto.
5755 (REG_CLASS_CONTENTS): Delete the SPE high registers.
5756 (REGISTER_NAMES): Ditto.
5757 (rs6000_reg_names): Ditto.
5758 * doc/tm.texi.in: Remove SPE as example.
5759 * doc/tm.texi: Regenerate.
5761 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5763 * config/rs6000/8540.md (ppc8540_brinc): Delete.
5764 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
5765 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
5766 * config/rs6000/rs6000.md (type): Remove "brinc".
5768 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5770 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
5771 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
5772 * config/rs6000/linuxspe.h: Delete file.
5773 * config/rs6000/rs6000.md: Don't include spe.md.
5774 * config/rs6000/spe.h: Delete file.
5775 * config/rs6000/spe.md: Delete file.
5776 * config/rs6000/t-rs6000: Remove spe.md.
5778 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5780 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
5781 (reg_or_none500mem_operand): Delete.
5782 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
5783 instead of reg_or_none500mem_operand.
5785 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5787 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5788 handling of SPE flags.
5789 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
5791 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5793 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
5795 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
5796 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
5797 paired_divv2sf3): Similar.
5798 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
5799 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
5800 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
5802 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
5803 Rename the paired_* instruction patterns.
5804 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
5806 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
5807 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
5808 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
5809 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
5810 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
5812 (struct machine_function): Delete field spe_insn_chain_scanned_p.
5813 (spe_func_has_64bit_regs_p): Delete.
5814 (spe_expand_predicate_builtin): Delete.
5815 (spe_expand_evsel_builtin): Delete.
5816 (TARGET_DWARF_REGISTER_SPAN): Do not define.
5817 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
5818 (invalid_e500_subreg): Delete.
5819 (rs6000_legitimize_address): Always force_reg op2 as well, for
5820 paired single memory accesses.
5821 (rs6000_member_type_forces_blk): Delete.
5822 (rs6000_spe_function_arg): Delete.
5823 (rs6000_expand_unop_builtin): Delete SPE handling.
5824 (rs6000_expand_binop_builtin): Ditto.
5825 (spe_expand_stv_builtin): Delete.
5826 (bdesc_2arg_spe): Delete.
5827 (spe_expand_builtin): Delete.
5828 (spe_expand_predicate_builtin): Delete.
5829 (spe_expand_evsel_builtin): Delete.
5830 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
5831 (spe_init_builtins): Delete.
5832 (spe_func_has_64bit_regs_p): Delete.
5833 (savres_routine_name): Delete "info" parameter. Adjust callers.
5834 (rs6000_emit_stack_reset): Ditto.
5835 (rs6000_dwarf_register_span): Delete.
5836 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
5837 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
5838 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
5840 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
5842 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
5843 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
5844 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
5845 mulv2sf3, divv2sf3): Delete expanders.
5847 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5849 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
5851 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5853 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
5854 * config/rs6000/rs6000.c: Ditto.
5856 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5858 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
5859 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
5860 comparison_operator.
5862 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5864 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
5865 * config/rs6000/rs6000.opt: Ditto.
5866 * config/rs6000/t-rtems: Ditto.
5868 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5870 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
5871 TARGET_E500_SINGLE by 0, simplify.
5872 * config/rs6000/rs6000.c: Ditto.
5873 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
5874 (spe_build_register_parallel): Delete.
5875 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
5876 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
5877 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
5878 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
5879 (E500_CONVERT): Delete.
5880 * config/rs6000/spe.md: Remove many patterns and all define_constants.
5882 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5884 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
5885 * config/rs6000/dfp.md: Ditto.
5886 (negdd2, *negdd2_fpr): Merge.
5887 (absdd2, *absdd2_fpr): Merge.
5888 (negtd2, *negtd2_fpr): Merge.
5889 (abstd2, *abstd2_fpr): Merge.
5890 * config/rs6000/e500.h: Delete file.
5891 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
5892 TARGET_FPRS by 1 and simplify.
5893 * config/rs6000/rs6000-c.c: Ditto.
5894 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
5896 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
5898 * config/rs6000/rs6000.md: Ditto.
5899 (floatdidf2, *floatdidf2_fpr): Merge.
5900 (move_from_CR_gt_bit): Delete.
5901 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
5902 (E500_CR_IOR_COMPARE): Delete.
5903 (All patterns that require !TARGET_FPRS): Delete.
5904 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
5906 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5908 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
5910 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5912 * graphds.c (add_edge): Intitialize edge's attached data.
5913 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
5914 pointer parameter. Call pointed function on each edge during
5915 graph traversing. Skip traversing the edge when the function
5917 (graphds_dfs, graphds_scc): Ditto.
5918 (for_each_edge): New parameter. Pass the new parameter to callback
5920 * graphds.h (skip_edge_callback): New function pointer type.
5921 (graphds_dfs, graphds_scc): New function pointer parameter.
5922 (graphds_edge_callback, for_each_edge): New parameter.
5924 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5926 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
5927 out code checking if runtime alias check is possible to below ...
5928 Call the new function.
5929 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
5930 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
5932 2017-06-07 Marek Polacek <polacek@redhat.com>
5935 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
5936 TYPE_OVERFLOW_WRAPS check.
5938 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5940 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
5941 if versioning is required.
5942 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
5943 peeling with the check for versioning.
5945 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5947 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
5948 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
5949 Set true to new parameter if new ssa variable is defined.
5950 (vect_gen_vector_loop_niters): Refactor. Set range information
5951 for the new vector loop bound variable.
5952 (vect_do_peeling): Ditto.
5954 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5956 * tree-affine.c (ssa.h): Include header file.
5957 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
5958 has wrapping overflow behavior.
5960 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5962 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
5964 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5966 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
5967 (tree_to_aff_combination): ... here.
5969 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5971 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
5972 reg_pressure model function.
5973 (ivopts_global_cost_for_size): Delete.
5974 (determine_set_costs, iv_ca_recount_cost): Call new model function
5975 ivopts_estimate_reg_pressure.
5977 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5979 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
5980 expensive than udiv. Remove floating point cases from mod.
5982 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5984 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
5987 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5989 * config/aarch64/aarch64.md
5990 (copysignsf3): Fix mask generation.
5992 2017-06-07 Jakub Jelinek <jakub@redhat.com>
5994 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
5996 (class dump_manager): Add register_dumps method.
5997 * dumpfile.c: Include langhooks.h.
5998 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
5999 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6000 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6001 (dump_manager::dump_register): Start with 512 entries instead of 32.
6002 (dump_manager::register_dumps): New method.
6003 * toplev.c (general_init): Instead of invoking register_dumps
6004 langhook, invoke register_dumps method on the dump manager.
6005 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6008 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6010 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6011 Say that # requires an associated define_split to exist, and that
6012 the define_split must be suitable for use after register allocation.
6014 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6016 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6017 (compute_outgoing_frequencies): Also initialize zero counts.
6018 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6019 around loops; preserve more of profile when nothing changes.
6021 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6023 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6025 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6026 * config/arm/arm-cpu-cdata.h: Regenerate.
6027 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6028 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6029 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6030 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6031 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6033 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6035 * config/arm/t-rmprofile: Likewise.
6036 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6038 2017-06-06 David S. Miller <davem@davemloft.net>
6041 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6042 blockage if function uses alloca.
6044 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6046 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6047 New "uid" fields to hold pretty-print IDs of group and ref.
6048 Memory references are now identified as <group_id>:<ref_id>
6049 instead of using [random] addresses.
6050 (dump_mem_details): Simplify, no functional change.
6051 (dump_mem_ref): Simplify and make output more concise.
6052 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6053 (find_or_create_group): Initialize group uid.
6054 (record_ref): Initialize ref uid. Improve debug output.
6055 (prune_group_by_reuse, should_issue_prefetch_p,)
6056 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6057 (mark_nontemporal_store, determine_loop_nest_reuse):
6058 Improve debug output.
6060 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6062 * dbgcnt.def (prefetch): New debug counter.
6063 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6064 (schedule_prefetches): Stop issueing prefetches if debug counter
6067 2017-06-06 Tom de Vries <tom@codesourcery.com>
6069 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6070 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6072 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6074 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6075 Use aarch64_reg_or_zero predicate for operand 4.
6076 (aarch64_compare_and_swap<mode> define_insn_and_split):
6077 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6078 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6080 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6082 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6083 (arm_compute_save_core_reg_mask): This.
6084 (thumb1_compute_save_reg_mask): Rename into ...
6085 (thumb1_compute_save_core_reg_mask): This.
6086 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6087 (arm_compute_frame_layout): Likewise.
6089 2017-06-06 Richard Biener <rguenther@suse.de>
6091 PR tree-optimization/80974
6092 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6093 keep or clear leaders SSA info.
6095 2017-06-06 Tom de Vries <tom@codesourcery.com>
6097 * config/nvptx/nvptx.c (split_mode_p): New function.
6098 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6100 2017-06-06 Tom de Vries <tom@codesourcery.com>
6102 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6104 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6107 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6110 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6112 * shrink-wrap.c (handle_simple_exit): Update profile.
6113 (try_shrink_wrapping): Upate profile.
6115 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6117 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6118 (tree_guess_outgoing_edge_probabilities): New.
6119 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6120 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6122 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6124 * ipa-split.c (split_function): Initialize return bb profile.
6126 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6128 * profile.c (compute_branch_probabilities): Also initialize
6131 2017-06-06 Richard Biener <rguenther@suse.de>
6133 PR tree-optimization/80928
6134 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6135 (vect_analyze_loop_operations): Properly guard analysis for
6137 (vect_transform_loop): Likewise.
6138 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6139 (vect_model_induction_cost): Do not cost for pure SLP.
6140 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6141 of induction in inner loop vectorization.
6142 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6143 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6144 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6146 (vect_analyze_slp_cost_1): Cost induction.
6147 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6148 (vect_get_slp_vect_defs): Likewise.
6149 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6150 (vect_transform_stmt): Handle SLP reductions.
6151 * tree-vectorizer.h (vectorizable_induction): Adjust.
6153 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6155 * config/rs6000/rs6000.c (make_resolver_func): Update
6156 init_lowered_empty_function call.
6158 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6160 * doc/invoke.texi: Document the -fprofile-abs-path option.
6161 * common.opt (fprofile-abs-path): New option.
6162 * gcov-io.h (gcov_write_filename): Declare.
6163 * gcov-io.c (gcov_write_filename): New function.
6164 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6165 * profile.c (output_location): Likewise.
6167 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6169 * shring-wrap.c: Revert accidental commit.
6171 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6173 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6175 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6177 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6179 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6180 profile in callgraph edge.
6181 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6182 (apply_scale): Likewise.
6183 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6184 Also scale profile when inlining function with zero profile.
6185 (initialize_cfun): Update exit block profile even when it is zero.
6186 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6187 when profile is read.
6189 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6191 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6192 (CLONE_*): New constants to define the processors we can generate
6193 code for with the target_clone attribute.
6194 (rs6000_clone_map): New array to identify which clone processors
6195 the current program is running on.
6196 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6197 target_clone attribute.
6198 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6199 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6200 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6201 (cpu_expand_builtin): Add support for target_clone attribute.
6202 (rs6000_valid_attribute_p): Allow "default" attribute.
6203 (get_decl_name): New debug function to simplify printing the
6204 current function name in debugging statements.
6205 (rs6000_clone_priority): New functions to support the target_clone
6206 attribute, and be able to generate code to switch between ISA 2.05
6207 through ISA 3.0 (power6 through power9).
6208 (rs6000_compare_version_priority): Likewise.
6209 (rs6000_get_function_versions_dispatcher): Likewise.
6210 (make_resolver_func): Likewise.
6211 (add_condition_to_bb): Likewise.
6212 (dispatch_function_versions): Likewise.
6213 (rs6000_generate_version_dispatcher_body): Likewise.
6214 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6215 (fusion_gpr_load_p): Fix a spacing issue.
6216 * doc/extend.texi (Common Function Attributes): Document that the
6217 PowerPC supports the target_clone attribute.
6219 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6221 * config/arm/arm.h: explain F symbol found in description of ARM
6222 register allocation in its legend.
6224 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6226 * config/mips/frame-header-opt.c: Include profile-count.h.
6227 * config/riscv/riscv.c: Include profile-count.h
6229 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6231 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6233 (sm_set_flag_if_changed): Add bbs field.
6234 (execute_sm_if_changed_flag_set): Pass BBS.
6235 (execute_sm): Update.
6237 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6239 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6242 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6244 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6245 (peephole2): New peephole2 to emit the above.
6246 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6248 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6250 * config/aarch64/aarch64.c (define_peephole2 above
6251 *sub_<shift>_<mode>): New peephole.
6253 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6255 * config/i386/i386.c (make_resolver_func): Update.
6256 * Makefile.in: Add profile-count.h and profile-count.o
6257 * auto-profile.c (afdo_indirect_call): Update to new API.
6258 (afdo_set_bb_count): Update.
6259 (afdo_propagate_edge): Update.
6260 (afdo_propagate_circuit): Update.
6261 (afdo_calculate_branch_prob): Update.
6262 (afdo_annotate_cfg): Update.
6263 * basic-block.h: Include profile-count.h
6264 (struct edge_def): Turn count to profile_count.
6265 (struct basic_block_def): Likewie.
6266 (REG_BR_PROB_BASE): Move to profile-count.h
6267 (RDIV): Move to profile-count.h
6268 * bb-reorder.c (max_entry_count): Turn to profile_count.
6269 (find_traces): Update.
6270 (rotate_loop):Update.
6271 (connect_traces):Update.
6272 (sanitize_hot_paths):Update.
6273 * bt-load.c (migrate_btr_defs): Update.
6274 * cfg.c (RDIV): Remove.
6275 (init_flow): Use alloc_block.
6276 (alloc_block): Uninitialize count.
6277 (unchecked_make_edge): Uninitialize count.
6278 (check_bb_profile): Update.
6279 (dump_edge_info): Update.
6280 (dump_bb_info): Update.
6281 (update_bb_profile_for_threading): Update.
6282 (scale_bbs_frequencies_int): Update.
6283 (scale_bbs_frequencies_gcov_type): Update.
6284 (scale_bbs_frequencies_profile_count): New.
6285 * cfg.h (update_bb_profile_for_threading): Update.
6286 (scale_bbs_frequencies_profile_count): Declare.
6287 * cfgbuild.c (compute_outgoing_frequencies): Update.
6288 (find_many_sub_basic_blocks): Update.
6289 * cfgcleanup.c (try_forward_edges): Update.
6290 (try_crossjump_to_edge): Update.
6291 * cfgexpand.c (expand_gimple_tailcall): Update.
6292 (construct_exit_block): Update.
6293 * cfghooks.c (verify_flow_info): Update.
6294 (dump_bb_for_graph): Update.
6295 (split_edge): Update.
6296 (make_forwarder_block): Update.
6297 (duplicate_block): Update.
6298 (account_profile_record): Update.
6299 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6300 (get_estimated_loop_iterations): Update.
6301 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6302 (single_likely_exit): Update.
6303 * cfgloopmanip.c (scale_loop_profile): Update.
6305 (set_zero_probability): Update.
6306 (lv_adjust_loop_entry_edge): Update.
6307 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6308 (purge_dead_edges): Update.
6309 (rtl_account_profile_record): Update.
6310 * cgraph.c (cgraph_node::create): Uninitialize count.
6311 (symbol_table::create_edge): Uninitialize count.
6312 (cgraph_update_edges_for_call_stmt_node): Update.
6313 (cgraph_edge::dump_edge_flags): Update.
6314 (cgraph_node::dump): Update.
6315 (cgraph_edge::maybe_hot_p): Update.
6316 * cgraph.h: Include profile-count.h
6317 (create_clone), create_edge, create_indirect_edge): Update.
6318 (cgraph_node): Turn count to profile_count.
6319 (cgraph_edge0: Likewise.
6320 (make_speculative, clone): Update.
6321 (create_edge): Update.
6322 (init_lowered_empty_function): Update.
6323 * cgraphclones.c (cgraph_edge::clone): Update.
6324 (duplicate_thunk_for_node): Update.
6325 (cgraph_node::create_clone): Update.
6326 * cgraphunit.c (cgraph_node::analyze): Update.
6327 (cgraph_node::expand_thunk): Update.
6328 * final.c (dump_basic_block_info): Update.
6329 * gimple-streamer-in.c (input_bb): Update.
6330 * gimple-streamer-out.c (output_bb): Update.
6331 * graphite.c (print_global_statistics): Update.
6332 (print_graphite_scop_statistics): Update.
6333 * hsa-brig.c: Include basic-block.h.
6334 * hsa-dump.c: Include basic-block.h.
6335 * hsa-gen.c (T sum_slice): Update.
6336 (convert_switch_statements):Update.
6337 * hsa-regalloc.c: Include basic-block.h.
6338 * ipa-chkp.c (chkp_produce_thunks): Update.
6339 * ipa-cp.c (struct caller_statistics): Update.
6340 (init_caller_stats): Update.
6341 (gather_caller_stats): Update.
6342 (ipcp_cloning_candidate_p): Update.
6343 (good_cloning_opportunity_p): Update.
6344 (get_info_about_necessary_edges): Update.
6345 (dump_profile_updates): Update.
6346 (update_profiling_info): Update.
6347 (update_specialized_profile): Update.
6348 (perhaps_add_new_callers): Update.
6349 (decide_about_value): Update.
6350 (ipa_cp_c_finalize): Update.
6351 * ipa-devirt.c (struct odr_type_warn_count): Update.
6352 (struct decl_warn_count): Update.
6353 (struct final_warning_record): Update.
6354 (possible_polymorphic_call_targets): Update.
6355 (ipa_devirt): Update.
6356 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6357 * ipa-icf.c (sem_function::merge): Update.
6358 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6359 * ipa-inline.c (compute_uninlined_call_time): Update.
6360 (compute_inlined_call_time): Update.
6361 (want_inline_small_function_p): Update.
6362 (want_inline_self_recursive_call_p): Update.
6363 (edge_badness): Update.
6364 (lookup_recursive_calls): Update.
6365 (recursive_inlining): Update.
6366 (inline_small_functions): Update.
6367 (dump_overall_stats): Update.
6368 (dump_inline_stats): Update.
6369 * ipa-profile.c (ipa_profile_generate_summary): Update.
6370 (ipa_propagate_frequency): Update.
6371 (ipa_profile): Update.
6372 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6373 * ipa-utils.c (ipa_merge_profiles): Update.
6374 * loop-doloop.c (doloop_modify): Update.
6375 * loop-unroll.c (report_unroll): Update.
6376 (unroll_loop_runtime_iterations): Update.
6377 * lto-cgraph.c (lto_output_edge): Update.
6378 (lto_output_node): Update.
6379 (input_node): Update.
6380 (input_edge): Update.
6381 (merge_profile_summaries): Update.
6382 * lto-streamer-in.c (input_cfg): Update.
6383 * lto-streamer-out.c (output_cfg): Update.
6384 * mcf.c (create_fixup_graph): Update.
6385 (adjust_cfg_counts): Update.
6386 (sum_edge_counts): Update.
6387 * modulo-sched.c (sms_schedule): Update.
6388 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6389 * predict.c (maybe_hot_count_p): Update.
6390 (probably_never_executed): Update.
6391 (dump_prediction): Update.
6392 (combine_predictions_for_bb): Update.
6393 (propagate_freq): Update.
6394 (handle_missing_profiles): Update.
6395 (counts_to_freqs): Update.
6396 (rebuild_frequencies): Update.
6397 (force_edge_cold): Update.
6398 * predict.h: Include profile-count.h
6399 (maybe_hot_count_p, counts_to_freqs): UPdate.
6400 * print-rtl-function.c: Do not include cfg.h
6401 * print-rtl.c: Include basic-block.h
6402 * profile-count.c: New file.
6403 * profile-count.h: New file.
6404 * profile.c (is_edge_inconsistent): Update.
6405 (correct_negative_edge_counts): Update.
6406 (is_inconsistent): Update.
6407 (set_bb_counts): Update.
6408 (read_profile_edge_counts): Update.
6409 (compute_frequency_overlap): Update.
6410 (compute_branch_probabilities): Update; Initialize and deinitialize
6412 (branch_prob): Update.
6413 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6414 (edge_gcov_count): New.
6415 (bb_gcov_count): New.
6416 * shrink-wrap.c (try_shrink_wrapping): Update.
6417 * tracer.c (better_p): Update.
6418 * trans-mem.c (expand_transaction): Update.
6419 (ipa_tm_insert_irr_call): Update.
6420 (ipa_tm_insert_gettmclone_call): Update.
6421 * tree-call-cdce.c: Update.
6422 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6423 (gimple_duplicate_sese_tail): Update.
6424 (gimple_account_profile_record): Update.
6425 (execute_fixup_cfg): Update.
6426 * tree-inline.c (copy_bb): Update.
6427 (copy_edges_for_bb): Update.
6428 (initialize_cfun): Update.
6429 (freqs_to_counts): Update.
6430 (copy_cfg_body): Update.
6431 (expand_call_inline): Update.
6432 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6433 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6434 (try_unroll_loop_completely): Update.
6435 (try_peel_loop): Update.
6436 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6437 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6438 * tree-ssa-loop-split.c (connect_loops): Update.
6439 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6440 * tree-ssa-reassoc.c (branch_fixup): Update.
6441 * tree-ssa-tail-merge.c (replace_block_by): Update.
6442 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6443 (compute_path_counts): Update.
6444 (update_profile): Update.
6445 (recompute_probabilities): Update.
6446 (update_joiner_offpath_counts): Update.
6447 (estimated_freqs_path): Update.
6448 (freqs_to_counts_path): Update.
6449 (clear_counts_path): Update.
6450 (ssa_fix_duplicate_block_edges): Update.
6451 (duplicate_thread_path): Update.
6452 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6453 (struct switch_conv_info): Update.
6454 * tree-tailcall.c (decrease_profile): Update.
6455 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6456 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6457 * value-prof.c (check_counter): Update.
6458 (gimple_divmod_fixed_value): Update.
6459 (gimple_mod_pow2): Update.
6460 (gimple_mod_subtract): Update.
6461 (gimple_ic_transform): Update.
6462 (gimple_stringop_fixed_value): Update.
6463 * value-prof.h (gimple_ic): Update.
6465 2017-06-02 Carl Love <cel@us.ibm.com>
6467 * config/rs6000/rs6000-c: Add support for built-in functions
6468 vector double vec_doublee (vector signed int);
6469 vector double vec_doublee (vector unsigned int);
6470 vector double vec_doublee (vector float);
6471 vector double vec_doubleh (vector signed int);
6472 vector double vec_doubleh (vector unsigned int);
6473 vector double vec_doubleh (vector float);
6474 vector double vec_doublel (vector signed int);
6475 vector double vec_doublel (vector unsigned int);
6476 vector double vec_doublel (vector float);
6477 vector double vec_doubleo (vector signed int);
6478 vector double vec_doubleo (vector unsigned int);
6479 vector double vec_doubleo (vector float);.
6480 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6481 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6483 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6484 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6485 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6487 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6488 vec_doublel, vec_doubleh.
6489 * doc/extend.texi: Update the built-in documentation file for the
6490 new built-in functions.
6492 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6495 * ipa-inline-analysis.c (free_growth_caches): Set
6496 edge_removal_hook_holder to NULL after removing it.
6498 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6500 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6501 comparision with zero.
6503 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6504 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6505 for early expansion of vec_min and vec_max builtins.
6506 (builtin_function_type): Add min/max unsigned variants to those
6507 identified as having unsigned arguments.
6509 2017-06-02 Olivier Hainque <hainque@adacore.com>
6511 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6513 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6515 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6516 Use VALL_F16 iterator rather than VALL.
6518 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6520 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6521 Emit CBNZ inside loop when doing a strong exchange and comparing
6522 against zero. Generate the CC flags after the loop.
6524 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6526 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6527 (dl_section_ref): New.
6528 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6529 On AIX, append an expression to subtract the size of the
6530 section length to dl_section_ref.
6532 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6534 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6535 for early expansion of vector absolute builtins.
6537 2017-06-02 Richard Biener <rguenther@suse.de>
6539 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6540 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6542 2017-06-02 Richard Biener <rguenther@suse.de>
6544 PR tree-optimization/80948
6545 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6548 2017-06-02 Richard Biener <rguenther@suse.de>
6550 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6552 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6553 for inductions as relevant.
6555 2017-06-02 Richard Biener <rguenther@suse.de>
6557 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6558 (vectorizable_induction): ... this. Remove dead code.
6560 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6562 * builtins. (expand_builtin_alloca): Remove second parameter and
6563 infer its value from the first parameter instead.
6564 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6566 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6568 PR rtl-optimization/80903
6569 * loop-doloop.c (add_test): Unshare sequence.
6571 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6573 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6575 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6577 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6579 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6580 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6581 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6582 (xlogue_layout::get_stub_rtx): Make static.
6583 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6584 (xlogue_layout::compute_stub_managed_regs): Rename to...
6585 (xlogue_layout::count_stub_managed_regs): ...this.
6586 (xlogue_layout::is_stub_managed_reg): New function.
6587 (xlogue_layout::m_stub_names): Rename to...
6588 (xlogue_layout::s_stub_names): ...this, make static.
6589 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6590 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6591 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6592 xlogue_layout::s_stub_names): Instantiate statics.
6593 (stub_managed_regs): Remove.
6594 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6595 (disable_call_ms2sysv_xlogues): Rename to...
6596 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6597 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6599 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6600 change after reload_completed.
6601 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6603 (ix86_expand_prologue): Likewise.
6604 (ix86_expand_epilogue): Likewise.
6605 (ix86_expand_split_stack_prologue): Likewise.
6606 (ix86_compute_frame_layout): Remove frame parameter ...
6607 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6608 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6610 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6611 (ix86_frame): Move from here ...
6612 * config/i386/i386.h (ix86_frame): ... to here.
6613 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6614 complete ix86_frame data structure instead. Remove some_ld_name.
6616 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6618 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6619 symbols that hold a DECL_VALUE_EXPR.
6621 2017-06-01 Martin Jambor <mjambor@suse.cz>
6623 PR tree-optimization/80898
6624 * tree-sra.c (process_subtree_disqualification): Removed.
6625 (disqualify_candidate): Do not acll
6626 process_subtree_disqualification.
6627 (subtree_mark_written_and_enqueue): New function.
6628 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6629 RHS has been disqualified and re-queue LHS if necessary. Apart
6630 from that, ignore disqualified RHS.
6632 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6634 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6635 address fetch for z10 or later.
6637 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6639 * config/arc/arc.md (tst_movb): Add guard when splitting.
6641 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6643 * config/arc/arc.c (arc_can_eliminate): Test against
6644 arc_frame_pointer_needed.
6646 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6648 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6649 to prevent store reordering.
6650 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6651 (type): Add block type.
6652 (stack_tie): Define special instruction to be used in
6655 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6657 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6658 constraint. It is not valid for the pattern.
6659 (noncommutative_binary_comparison): Likewise.
6661 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6663 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6666 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6668 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6669 be used by the reg-alloc.
6671 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6673 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6674 reg-alloc when having mul64 or mul32x16 instructions.
6675 (mulsidi3): Likewise.
6676 (umulsidi3): Likewise.
6677 (mulsi32x16): New pattern.
6678 (mulsi64): Likewise.
6679 (mulsidi64): Likewise.
6680 (umulsidi64): Likewise.
6681 (MUL32x16_REG): Define.
6682 (mul64_600): Use MUL32x16_REG.
6683 (mac64_600): Likewise.
6684 (umul64_600): Likewise.
6685 (umac64_600): Likewise.
6687 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6689 * config/arc/arc.md (mulsi3_700): Make it commutative.
6691 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6693 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6695 (*sign_extendsidi2_insn): Likewise for movstosw.
6697 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6699 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6700 the type of the input discriminant value. Convert the
6701 discriminant value of signedness vary.
6703 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6705 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6706 Add to -Wall section.
6708 2017-06-01 Richard Biener <rguenther@suse.de>
6711 * fold-const.c (fold_plusminus_mult_expr): If the factored
6712 factor may be zero use a wrapping type for the inner operation.
6713 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6714 and handle moved defs.
6715 (process_assignment): Properly guard the unary op case. Return a
6716 tri-state indicating that moving the stmt before the call may allow
6717 to continue. Pass through to_move.
6718 (find_tail_calls): Handle moving unrelated defs before
6721 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6724 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
6725 splitter result in the canonical way.
6727 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
6729 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
6730 also for 32bit target. Update insn attributes.
6731 (zero-extendsidi2 splitter): Allow all registers for operand 1.
6733 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
6735 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
6736 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
6737 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
6738 (_mm_maskz_min_ss): New intrinsics.
6740 2017-05-31 Martin Liska <mliska@suse.cz>
6742 * tree-vect-loop.c (vect_create_epilog_for_reduction):
6743 Change comment style to one we normally use.
6744 (vectorizable_reduction): Likewise.
6745 (vectorizable_induction): Likewise.
6746 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6747 (vectorizable_call): Likewise.
6748 (vectorizable_simd_clone_call): Likewise.
6749 (vectorizable_conversion): Likewise.
6750 (vectorizable_assignment): Likewise.
6751 (vectorizable_shift): Likewise.
6752 (vectorizable_operation): Likewise.
6753 (vectorizable_store): Likewise.
6754 (vectorizable_load): Likewise.
6755 * tree-vectorizer.h: Likewise.
6757 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
6759 * passes.c (emergency_dump_function): New.
6760 * tree-pass.h (emergency_dump_function): Declare.
6761 * plugin.c (plugins_internal_error_function): Remove.
6762 * plugin.h (plugins_internal_error_function): Remove declaration.
6763 * toplev.c (internal_error_function): New static function. Use it...
6764 (general_init): ...here.
6766 2017-05-31 Graham Markall <graham.markall@embecosm.com>
6768 * config/arc/arc.c (arc_print_operand): Handle constant operands.
6769 (arc_rtx_costs): Add costs for new patterns.
6770 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
6771 * config/arc/predicates.md: Add _1_2_3_operand predicate.
6773 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6775 * tree-ssa-strlen.c (get_next_strinfo): New function.
6776 (get_stridx_plus_constant): Use it.
6777 (zero_length_string): Likewise.
6778 (adjust_related_strinfos): Likewise.
6779 (adjust_last_stmt): Likewise.
6781 2017-05-31 Richard Biener <rguenther@suse.de>
6784 * config/i386/i386.c (ix86_expand_builtin): Remove assert
6785 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
6787 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6789 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
6790 loop_vinfo argument and use of dependence distance vectors.
6791 Check instead whether the two references differ only in their
6792 initial value and assume that they have the same alignment if the
6793 difference is a multiple of the vector alignment.
6794 (vect_analyze_data_refs_alignment): Update call accordingly.
6796 2017-05-31 Martin Liska <mliska@suse.cz>
6799 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
6801 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6803 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
6804 (create_intersect_range_checks): Move from ...
6805 * tree-data-ref.c (create_intersect_range_checks_index)
6806 (create_intersect_range_checks): ... to here.
6807 (create_runtime_alias_checks): New function factored from ...
6808 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
6809 here. Call above function.
6810 * tree-data-ref.h (create_runtime_alias_checks): New function.
6812 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6814 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
6815 segment length for dr_b and compute it in wide_int.
6817 2017-05-31 Richard Biener <rguenther@suse.de>
6819 PR tree-optimization/80906
6820 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
6821 and pass through iv_map.
6822 (copy_bb_and_scalar_dependences): Adjust.
6823 (translate_pending_phi_nodes): Likewise.
6824 (copy_loop_close_phi_args): Handle code-generating IVs instead
6827 2017-05-30 David Malcolm <dmalcolm@redhat.com>
6829 * diagnostic-color.c (color_dict): Add "type-diff".
6830 (parse_gcc_colors): Update comment.
6831 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6832 -fdiagnostics-show-template-tree and -fno-elide-type.
6833 (GCC_COLORS): Add type-diff to example.
6835 (-fdiagnostics-show-template-tree): New.
6836 (-fno-elide-type): New.
6837 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
6838 the pp_format_decoder callback. Call any m_format_postprocessor's
6840 (pretty_printer::pretty_printer): Initialize
6841 m_format_postprocessor.
6842 (pretty_printer::~pretty_printer): Delete any
6843 m_format_postprocessor.
6844 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
6845 (class format_postprocessor): New class.
6846 (struct pretty_printer::format_decoder): Document the new parameters.
6847 (struct pretty_printer::m_format_postprocessor): New field.
6848 * tree-diagnostic.c (default_tree_printer): Update for new
6849 bool and const char ** params.
6850 * tree-diagnostic.h (default_tree_printer): Likewise.
6852 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
6854 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
6855 (lwa_operand): Delete rs6000_gen_cell_microcode test.
6856 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6857 rs6000_gen_cell_microcode code.
6858 (rs6000_final_prescan_insn): Delete.
6859 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
6860 "warn-cell-microcode" entries.
6861 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
6862 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
6863 throughout. Change cc_reg_not_micro_cr0_operand to
6864 cc_reg_not_cr0_operand throughout.
6865 (*extendhi<mode>2_noload): Delete.
6866 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
6867 (mwarn-cell-microcode): Delete.
6868 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
6869 -mgen-cell-microcode and -mwarn-cell-microcode.
6871 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
6874 * config/i386/constraints.md (Yd): New constraint.
6876 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
6877 and (?Yd, r) alternatives. Update insn attributes.
6878 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
6879 and (?*Yd, r) alternatives. Update insn attributes.
6880 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
6882 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
6884 * gimplify.c (gimplify_modify_expr): Don't create a
6885 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
6888 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6890 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
6892 2017-05-30 Richard Biener <rguenther@suse.de>
6894 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
6895 and reduc_def fields.
6896 (STMT_VINFO_REDUC_TYPE): New define.
6897 (STMT_VINFO_REDUC_DEF): Likewise.
6898 (vect_force_simple_reduction): Adjust prototype.
6899 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
6900 (vect_is_simple_reduction): Remove check_reduction argument.
6901 (vect_force_simple_reduction): Adjust and set
6902 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6903 (vectorizable_reduction): Do not re-do reduction analysis
6904 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6905 * tree-parloops.c (gather_scalar_reductions): Adjust.
6907 2017-05-30 Richard Biener <rguenther@suse.de>
6910 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
6913 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
6915 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
6916 Introduce unknown_misalignment parameter and remove vf.
6917 (vect_peeling_hash_get_lowest_cost):
6918 Pass unknown_misalignment parameter.
6919 (vect_enhance_data_refs_alignment):
6920 Fix unsupportable data ref treatment.
6922 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6924 * tree-vect-data-refs.c (vect_get_data_access_cost):
6925 Workaround for SLP handling.
6926 (vect_enhance_data_refs_alignment):
6927 Compute costs for doing no peeling at all, compare to the best
6928 peeling costs so far and avoid peeling if cheaper.
6930 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6932 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
6933 Return peeling info and set costs to zero for unlimited cost
6935 (vect_enhance_data_refs_alignment): Also inspect all datarefs
6936 with unknown misalignment. Compute and costs for unknown
6937 misalignment, compare them to the costs for known misalignment
6938 and choose the cheapest for peeling.
6940 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6942 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
6943 (vect_get_peeling_costs_all_drs): Create function.
6944 (vect_peeling_hash_get_lowest_cost):
6945 Use vect_get_peeling_costs_all_drs.
6946 (vect_peeling_supportable): Create function.
6947 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
6949 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6951 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
6952 DR_HAS_NEGATIVE_STEP.
6953 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
6954 (vect_enhance_data_refs_alignment): Use.
6955 (vect_duplicate_ssa_name_ptr_info): Use.
6956 * tree-vectorizer.h (dr_misalignment): Use.
6957 (known_alignment_for_access_p): Use.
6959 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6962 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
6964 (has_section_name): New function.
6966 2017-05-30 Martin Liska <mliska@suse.cz>
6969 * auto-profile.c (get_function_decl_from_block): Fix
6972 2017-05-30 Richard Biener <rguenther@suse.de>
6975 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
6977 2017-05-30 Martin Liska <mliska@suse.cz>
6979 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
6980 * dumpfile.h (struct dump_file_info): Remove ctors.
6982 2017-05-30 Martin Liska <mliska@suse.cz>
6984 * predict.def: Fix GNU coding style.
6986 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6988 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
6989 Mark 'to' argument with ATTRIBUTE_UNUSED.
6991 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6993 * config/xtensa/xtensa.c (xtensa_emit_call): Use
6994 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
6995 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
6998 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7000 * doc/install.texi (Options specification): Restore entry of
7001 --enable-sjlj-exceptions.
7003 2017-05-27 Michael Eager <eager@eagercon.com>
7006 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7008 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7010 * config/microblaze/microblaze.h
7011 (FIXED_REGISTERS): Update in macro.
7012 (CALL_USED_REGISTERS): Update in macro.
7014 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7016 * doc/install.texi: Add links to macOS binary distributions.
7018 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7022 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7024 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7026 2017-05-26 Martin Liska <mliska@suse.cz>
7028 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7030 2017-05-26 Martin Liska <mliska@suse.cz>
7032 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7033 always leading ';; '.
7034 (dump_bb_info): Likewise.
7035 (brief_dump_cfg): Likewise.
7036 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7037 * dumpfile.c: Remove usage of TDF_VERBOSE.
7038 * dumpfile.h (enum dump_kind): Likewise.
7039 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7040 * print-tree.c (debug_verbose): Remove.
7041 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7042 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7043 * tree-diagnostic.c (default_tree_printer): Replace
7044 TDF_DIAGNOSTIC with TDF_SLIM.
7046 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7048 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7049 in parameter loop, rather than loop_vinfo.
7050 (create_intersect_range_checks): Ditto.
7051 (vect_create_cond_for_alias_checks): Update call to above functions.
7053 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7055 PR tree-optimization/80815
7056 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7057 for merging runtime alias checks. Handle negative DR_STEPs.
7059 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7061 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7063 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7064 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7065 out code pruning runtime alias checks.
7066 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7067 factored out from above.
7068 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7070 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7072 (prune_runtime_alias_test_list): New decalaration.
7074 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7076 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7077 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7078 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7079 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7080 (operator==, comp_dr_with_seg_len_pair): Ditto.
7081 (vect_prune_runtime_alias_test_list): Ditto.
7083 2017-05-26 Martin Liska <mliska@suse.cz>
7086 * params.def: Bound partial-inlining-entry-probability param.
7088 2017-05-26 Marek Polacek <polacek@redhat.com>
7091 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7094 2017-05-26 Richard Biener <rguenther@suse.de>
7096 PR tree-optimization/80842
7097 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7100 2017-05-26 Richard Biener <rguenther@suse.de>
7102 PR tree-optimization/80844
7103 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7105 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7107 * doc/md.texi (Machine Constraints): Update x86 family
7108 machine constraints section to match 'config/i386/constraints.md'.
7110 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7112 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7114 2017-05-25 Nathan Sidwell <nathan@acm.org>
7116 * doc/invoke.texi (--enable-languages): Update documentation.
7118 2017-05-25 Martin Liska <mliska@suse.cz>
7120 * dumpfile.c: Add TDF_FOLDING.
7121 * dumpfile.h (enum dump_kind): Likewise.
7122 * genmatch.c (dt_simplify::gen_1): Use it.
7124 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7126 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7128 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7130 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7131 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7133 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7135 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7136 * match.pd (X == C): Rewrite it here.
7137 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7138 with_certain_nonzero_bits2): New predicates.
7139 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7141 2017-05-24 Nathan Sidwell <nathan@acm.org>
7143 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7146 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7149 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7151 * config/powerpcspe: New port. Files are copied from the rs6000
7152 port, with "rs6000" in filenames replaced by "powerpcspe".
7154 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7156 PR rtl-optimization/80754
7157 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7159 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7161 * config/sparc/sparc.md (length): Return the correct value for -mflat
7162 sibcalls to match output_sibcall.
7164 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7168 * config/rs6000/rs6000.c (struct machine_function): Add new field
7170 (rs6000_get_separate_components): Init that field, use it.
7171 (rs6000_components_for_bb): Use the field.
7173 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7175 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7177 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7180 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7182 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7185 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7186 address against address_operand predicate.
7187 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7189 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7191 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7192 parameters passed indirectly.
7194 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7196 * config/i386/i386.md (*movdi_internal): Remove SSE4
7197 alternative 18 (?r, *v). Update insn attributes.
7198 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7199 Update insn attributes.
7200 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7201 Update insn attributes.
7202 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7203 alternative 1 (r, v). Remove isa attribute.
7204 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7205 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7206 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7208 2017-05-23 Tom de Vries <tom@codesourcery.com>
7210 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7213 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7215 * cgraphunit.c (symbol_table::process_new_functions): Update.
7216 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7217 (inline_generate_summary): Rename to ...
7218 (ipa_fn_summary_generate): ... this one.
7219 (inline_read_summary): Rename to ...
7220 (ipa_fn_summary_read): ... this one.
7221 (inline_write_summary): Rename to ...
7222 (ipa_fn_summary_write): ... this one.
7223 (inline_free_summary): Rename to ...
7224 (ipa_free_fn_summary): ... this one.
7225 (pass_data_local_fn_summary, pass_local_fn_summary,
7226 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7227 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7228 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7229 make_pass_ipa_fn_summary): New.
7230 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7231 inline_write_summary, inline_free_summary): Remove.
7232 (ipa_free_fn_summary) : New.
7233 * ipa-inline.c (ipa_inline): Update.
7234 (pass_ipa_inline): Do not generate summaries.
7235 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7237 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7238 and add pass_ipa_fn_summary.
7239 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7241 (make_pass_inline_parameters): Remove.
7243 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7245 * omp-low.c (struct omp_context): Remove "default_kind" member.
7248 * omp-offload.c (execute_oacc_device_lower): Remove the
7249 parallelism dimensions function attributes for unparallelized
7250 OpenACC kernels constructs.
7252 2017-05-23 Martin Liska <mliska@suse.cz>
7254 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7256 (cgraph_edge::make_speculative): Likewise.
7257 (cgraph_edge::resolve_speculation): Likewise.
7258 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7259 (cgraph_node::dump): Likewise.
7260 * cgraph.h: Likewise.
7261 * cgraphunit.c (analyze_functions): Likewise.
7262 (symbol_table::compile): Likewise.
7263 * ipa-cp.c (print_all_lattices): Likewise.
7264 (determine_versionability): Likewise.
7265 (initialize_node_lattices): Likewise.
7266 (ipcp_verify_propagated_values): Likewise.
7267 (estimate_local_effects): Likewise.
7268 (update_profiling_info): Likewise.
7269 (create_specialized_node): Likewise.
7270 (perhaps_add_new_callers): Likewise.
7271 (decide_about_value): Likewise.
7272 (decide_whether_version_node): Likewise.
7273 (identify_dead_nodes): Likewise.
7274 (ipcp_store_bits_results): Likewise.
7275 * ipa-devirt.c (dump_targets): Likewise.
7276 (ipa_devirt): Likewise.
7277 * ipa-icf.c (sem_item::dump): Likewise.
7278 (sem_function::equals): Likewise.
7279 (sem_variable::equals): Likewise.
7280 (sem_item_optimizer::read_section): Likewise.
7281 (sem_item_optimizer::execute): Likewise.
7282 (congruence_class::dump): Likewise.
7283 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7284 (dump_inline_summary): Likewise.
7285 (estimate_node_size_and_time): Likewise.
7286 (inline_analyze_function): Likewise.
7287 * ipa-inline-transform.c (inline_call): Likewise.
7288 * ipa-inline.c (report_inline_failed_reason): Likewise.
7289 (want_early_inline_function_p): Likewise.
7290 (edge_badness): Likewise.
7291 (update_edge_key): Likewise.
7292 (inline_small_functions): Likewise.
7293 * ipa-profile.c (ipa_profile): Likewise.
7294 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7295 (ipa_make_edge_direct_to_target): Likewise.
7296 (remove_described_reference): Likewise.
7297 (ipa_impossible_devirt_target): Likewise.
7298 (propagate_controlled_uses): Likewise.
7299 (ipa_print_node_params): Likewise.
7300 (ipcp_transform_function): Likewise.
7301 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7302 (propagate_pure_const): Likewise.
7303 * ipa-reference.c (generate_summary): Likewise.
7304 (read_write_all_from_decl): Likewise.
7305 (propagate): Likewise.
7306 (ipa_reference_read_optimization_summary): Likewise.
7307 * ipa-utils.c (ipa_merge_profiles): Likewise.
7308 * ipa.c (walk_polymorphic_call_targets): Likewise.
7309 (symbol_table::remove_unreachable_nodes): Likewise.
7310 (ipa_single_use): Likewise.
7311 * passes.c (execute_todo): Likewise.
7312 * predict.c (drop_profile): Likewise.
7313 * symtab.c (symtab_node::get_dump_name): New function.
7314 (symtab_node::dump_name): Likewise.
7315 (symtab_node::dump_asm_name): Likewise.
7316 (symtab_node::dump_references): Likewise.
7317 (symtab_node::dump_referring): Likewise.
7318 (symtab_node::dump_base): Likewise.
7319 (symtab_node::debug_symtab): Likewise.
7320 * tree-sra.c (convert_callers_for_node): Likewise.
7321 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7322 * value-prof.c (init_node_map): Likewise.
7324 2017-05-23 Martin Liska <mliska@suse.cz>
7326 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7327 and symtab_node::debug_symtab to symbol_table::debug.
7328 * cgraphunit.c (analyze_functions): Use the renamed function.
7329 (symbol_table::compile): Likewise.
7330 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7331 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7332 * passes.c (execute_todo): Likewise.
7333 * symtab.c (symbol_table::dump): New function.
7334 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7336 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7338 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7339 that nonconst implies exec.
7341 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7343 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7344 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7345 (inline_edge_summary_vec): Turn into ...
7346 (ipa_call_summaries): ... this one.
7347 (redirect_to_unreachable, edge_set_predicate,
7348 evaluate_properties_for_edge, inline_summary_alloc,
7349 reset_ipa_call_summary, reset_inline_summary,
7350 inline_summary_t::duplicate): Update.
7351 (inline_edge_duplication_hook): Turn to ...
7352 (ipa_call_summary_t::duplicate): ... this one.
7353 (inline_edge_removal_hook): Turn to ...
7354 (ipa_call_summary_t::remove): ... this one.
7355 (dump_inline_edge_summary): Turn to ...
7356 (dump_ipa_call_summary): ... this one.
7357 (estimate_function_body_sizes): Update.
7358 (inline_update_callee_summaries): Update.
7359 (remap_edge_change_prob): Update.
7360 (remap_edge_summaries): Update.
7361 (inline_merge_summary): Update.
7362 (do_estimate_edge_time): Update.
7363 (inline_generate_summary): Update.
7364 (inline_read_section): Update.
7365 (inline_read_summary): Update.
7366 (inline_free_summary): Update.
7367 * ipa-inline.c (can_inline_edge_p): Update.
7368 (compute_inlined_call_time): Update.
7369 (want_inline_small_function_p): Update.
7370 (edge_badness): Update.
7371 (early_inliner): Update.
7372 * ipa-inline.h (inline_edge_summary): Turn to ...
7373 (ipa_call_summary): ... this one.
7374 (ipa_call_summary_t): New class.
7375 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7376 (ipa_call_summaries): New.
7377 (inline_edge_summary): Remove.
7378 (estimate_edge_growth): Update.
7379 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7380 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7381 * ipa-split.c (execute_split_functions): Update.
7382 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7384 2017-05-23 Tom de Vries <tom@codesourcery.com>
7386 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7387 attributes): Document rdrand effective target.
7389 2017-05-23 Tom de Vries <tom@codesourcery.com>
7391 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7392 attributes): Sort alphabetically.
7394 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7396 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7398 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7401 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7402 V2DF/V2DI splat into two separate patterns, one that handles
7403 registers, and the other that only handles memory. Drop support
7404 for splatting from a GPR on ISA 2.07 and then splitting the
7405 splat into direct move and splat.
7406 (vsx_splat_<mode>_reg): Likewise.
7407 (vsx_splat_<mode>_mem): Likewise.
7409 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7411 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7413 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7416 * omp-low.c (finish_taskreg_remap): New function.
7417 (finish_taskreg_scan): If unit size of ctx->record_type
7418 is non-constant, unshare the size expression and replace
7419 decls in it with possible outer var refs.
7422 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7423 GOVD_SHARED rather than GOVD_PRIVATE with it.
7424 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7425 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7428 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7429 as last argument to build_outer_var_ref for pointer bases of array
7432 2017-05-19 Martin Sebor <msebor@redhat.com>
7434 * print-tree.c (print_node): Print DECL_READ_P flag.
7436 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7438 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7439 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7440 * cgraph.c: Likewise.
7441 * cgraphunit.c: Likewise.
7442 * gengtype.c: Likewise.
7443 * ipa-cp.c: Likewise.
7444 * ipa-devirt.c: Likewise.
7445 * ipa-icf.c: Likewise.
7446 * ipa-predicate.c: Likewise.
7447 * ipa-profile.c: Likewise.
7448 * ipa-prop.c: Likewise.
7449 * ipa-split.c: Likewise.
7451 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7452 edge_predicate_pool, dump_inline_hints,
7453 inline_summary::account_size_time, redirect_to_unreachable,
7454 edge_set_predicate, set_hint_predicate,
7455 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7456 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7457 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7458 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7459 ipa_call_summary_t::remove, initialize_growth_caches,
7460 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7461 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7462 mark_modified, unmodified_parm_1, unmodified_parm,
7463 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7464 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7465 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7466 will_be_nonconstant_predicate, record_modified_bb_info,
7467 get_minimal_bb, record_modified, param_change_prob,
7468 phi_result_unknown_predicate, predicate_for_phi_result,
7469 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7470 estimate_function_body_sizes, compute_inline_parameters,
7471 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7472 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7473 inline_update_callee_summaries, remap_edge_change_prob,
7474 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7475 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7476 inline_analyze_function, inline_summary_t::insert,
7477 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7478 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7479 inline_free_summary): Move to ipa-fnsummary.h
7480 (predicate_t): Remove.
7481 * ipa-fnsummary.c: New file.
7482 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7483 (enum inline_hints_vals, inline_hints, agg_position_info,
7484 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7485 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7486 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7487 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7488 inline_read_summary, inline_write_summary, inline_free_summary,
7489 inline_analyze_function, initialize_inline_failed,
7490 inline_merge_summary, inline_update_overall_summary,
7491 compute_inline_parameters): Move to ipa-fnsummary.h
7492 * ipa-fnsummary.h: New file.
7493 * ipa-inline-transform.h: Include ipa-inline.h.
7494 * ipa-inline.c: LIkewise.
7496 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7498 * ipa-inline.c (edge_badness): Use inlined_time instead of
7499 inline_summaries->get.
7501 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7503 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7505 2017-05-22 Nathan Sidwell <nathan@acm.org>
7507 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7508 (fdump-lang): Document 'raw' option.
7509 * dumpfile.h (TDI_tu): Delete.
7510 * dumpfile.c (dump_files): Remove translation-unit.
7511 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7513 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7515 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7516 command option from $(AWK) call.
7517 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7518 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7519 [FORMAT]: Remove handling of variable.
7520 * config/avr/t-multilib: Regenerate.
7522 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7524 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7526 (dump_inline_summary): Do not print self_time.
7527 (estimate_function_body_sizes): Do not set self_time.
7528 (compute_inline_parameters): Likewise.
7529 (inline_read_section, inline_write_summary): Do not stream self_time.
7530 * ipa-inline.h (inline_summary): Drop self_time.
7532 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7534 * ipa-inline-analysis.c (account_size_time): Rename to ...
7535 (inline_summary::account_size_time): ... this one.
7536 (reset_ipa_call_summary): Turn to ...
7537 (ipa_call_summary::reset): ... this one.
7538 (reset_inline_summary): Turn to ...
7539 (inline_summary::reset): ... this one.
7540 (inline_summary_t::remove): Update.
7541 (inline_summary_t::duplicate): Update.
7542 (ipa_call_summary_t::remove): Update.
7543 (dump_inline_summary): Update.
7544 (estimate_function_body_sizes): Update.
7545 (compute_inline_parameters): Update.
7546 (estimate_node_size_and_time): Update.
7547 (inline_merge_summary): Update.
7548 (inline_update_overall_summary): Update.
7549 (inline_read_section): Update.
7550 (inline_write_summary): Update.
7551 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7552 add account_size_time and reset member functions.
7553 (ipa_call_summary): Add reset function.
7554 * ipa-predicate.h (predicate::operator &): Constify.
7556 2017-05-22 Richard Biener <rguenther@suse.de>
7558 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7560 2017-05-19 Jason Merrill <jason@redhat.com>
7562 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7564 2017-05-19 Marek Polacek <polacek@redhat.com>
7567 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7568 TYPE_OVERFLOW_WRAPS checks.
7570 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7572 * tree-core.h (enum omp_clause_default_kind): Add
7573 "OMP_CLAUSE_DEFAULT_PRESENT".
7574 * tree-pretty-print.c (dump_omp_clause): Handle it.
7575 * gimplify.c (enum gimplify_omp_var_data): Add
7576 "GOVD_MAP_FORCE_PRESENT".
7577 (gimplify_adjust_omp_clauses_1): Map it to
7578 "GOMP_MAP_FORCE_PRESENT".
7579 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7581 * gimplify.c (oacc_default_clause): Clarify.
7583 2017-05-19 Nathan Sidwell <nathan@acm.org>
7585 LANG_HOOK_REGISTER_DUMPS
7586 * toplev.c (general_init): Call register dump lang hook.
7587 * doc/invoke.texi: Document -fdump-lang option family.
7588 * dumpfile.c (dump_files): Remove class dump here.
7589 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7590 * dumpfile.h (tree_dump_index): Remove TDI_class.
7591 * langhooks-def.h (lhd_register_dumps): Declare.
7592 (LANG_HOOKS_REGISTER_DUMPS): Define.
7593 (LANG_HOOKS_INITIALIZER): Add it.
7594 * langhooks.c (lhd_register_dumps): Define.
7595 * langhooks.h (struct lang_hooks): Add register_dumps.
7597 2017-05-19 Nathan Sidwell <nathan@acm.org>
7599 * context.h (context::set_passes): New.
7600 * context.c (context::context): Do not create pass manager.
7601 * toplev.c (general_init): Create pass manager here.
7603 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7605 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7606 use this splitter if two add or or instructions would also work for
7607 the constant we want to generate.
7609 2017-05-19 Richard Biener <rguenther@suse.de>
7612 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7613 predicate evaluation.
7615 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7617 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7619 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7622 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7624 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7625 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7626 (inline_edge_summary_vec): Turn into ...
7627 (ipa_call_summaries): ... this one.
7628 (redirect_to_unreachable, edge_set_predicate,
7629 evaluate_properties_for_edge, inline_summary_alloc,
7630 reset_ipa_call_summary, reset_inline_summary,
7631 inline_summary_t::duplicate): Update.
7632 (inline_edge_duplication_hook): Turn to ...
7633 (ipa_call_summary_t::duplicate): ... this one.
7634 (inline_edge_removal_hook): Turn to ...
7635 (ipa_call_summary_t::remove): ... this one.
7636 (dump_inline_edge_summary): Turn to ...
7637 (dump_ipa_call_summary): ... this one.
7638 (estimate_function_body_sizes): Update.
7639 (inline_update_callee_summaries): Update.
7640 (remap_edge_change_prob): Update.
7641 (remap_edge_summaries): Update.
7642 (inline_merge_summary): Update.
7643 (do_estimate_edge_time): Update.
7644 (inline_generate_summary): Update.
7645 (inline_read_section): Update.
7646 (inline_read_summary): Update.
7647 (inline_free_summary): Update.
7648 * ipa-inline.c (can_inline_edge_p): Update.
7649 (compute_inlined_call_time): Update.
7650 (want_inline_small_function_p): Update.
7651 (edge_badness): Update.
7652 (early_inliner): Update.
7653 * ipa-inline.h (inline_edge_summary): Turn to ...
7654 (ipa_call_summary): ... this one.
7655 (ipa_call_summary_t): New class.
7656 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7657 (ipa_call_summaries): New.
7658 (inline_edge_summary): Remove.
7659 (estimate_edge_growth): Update.
7660 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7661 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7662 * ipa-split.c (execute_split_functions): Update.
7663 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7665 2017-05-19 Richard Biener <rguenther@suse.de>
7668 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7670 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7672 * config/rs6000/rs6000.c (struct machine_function): Add field
7673 fpr_is_wrapped_separately.
7674 (rs6000_get_separate_components): Use 64 components. Handle the
7676 (rs6000_components_for_bb): Handle the FPR components.
7677 (rs6000_emit_prologue_components): Handle the FPR components.
7678 (rs6000_emit_epilogue_components): Handle the FPR components.
7679 (rs6000_set_handled_components): Handle the FPR components.
7680 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7681 that are already separately shrink-wrapped.
7682 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7683 that are already separately shrink-wrapped.
7685 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7688 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7691 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7692 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7693 cases where the register allocator uses a move and an offsettable
7694 memory operation to/from a FPR register on ISA 2.06/2.07.
7695 (define_peephole2 for Altivec d-form store): Likewise.
7697 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7700 * config/i386/mmx.md (*mov<mode>_internal): Enable
7701 alternatives 11, 12, 13 and 14 also for 32bit targets.
7702 Remove alternatives 15, 16, 17 and 18.
7703 * config/i386/sse.md (vec_concatv2di): Change
7704 alternative (!x, *y) to (x, ?!*Yn).
7706 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7708 * dumpfile.h (enum dump_kind): Remove stray comma.
7710 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7712 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7713 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7714 predicate::num_conditions
7715 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7716 (CHANGED): turn into predicate::changed.
7717 (agg_position_info): Move to ipa-predicate.h
7718 (add_condition, predicate::add_clause, predicate::operator &=,
7719 predicate::or_with, predicate::evaluate, predicate::probability,
7720 dump_condition, dump_clause, predicate::dump,
7721 predicate::remap_after_duplication, predicate::remap_after_inlining,
7722 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
7723 (evaluate_conditions_for_known_args): Update.
7724 (set_cond_stmt_execution_predicate): Update.
7725 * ipa-inline.h: Include ipa-predicate.h
7726 (condition, inline_param_summary, conditions, agg_position_info,
7727 predicate): Move to ipa-predicate.h
7728 * ipa-predicate.c: New file.
7729 * ipa-predicate.h: New file.
7731 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
7733 * final.c (leaf_function_p): Check we are not in a sequence.
7735 2017-05-18 Martin Liska <mliska@suse.cz>
7737 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
7738 * dumpfile.c (dump_register): Use new enum dump_kind.
7739 (get_dump_file_name): Likewise.
7740 (dump_enable_all): Likewise.
7741 (dump_switch_p_1): Likewise.
7742 (enable_rtl_dump_file): Remove usage of TDF_RTL.
7743 * dumpfile.h (enum dump_kind): New enum type.
7744 (struct dump_file_info): Create constructor and
7745 format fields and comments.
7746 * passes.c (pass_manager::register_one_dump_file):
7748 * statistics.c (statistics_early_init): Likewise.
7749 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
7750 TDF_TREE with TDF_SLIM.
7751 (gather_memory_references_ref): Likewise.
7753 2017-05-18 Martin Liska <mliska@suse.cz>
7755 * vec.h (struct vnull): Use it.
7757 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7759 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
7760 (true_predicate, false_predicate, true_predicate_p,
7761 false_predicate_p): Remove.
7762 (single_cond_predicate, not_inlined_predicate): Turn to member function
7764 (add_condition): Update.
7765 (add_clause): Turn to...
7766 (predicate::add_clause): ... this one; update; allow passing NULL
7768 (and_predicates): Turn to ...
7769 (predicate::operator &=): ... this one.
7770 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
7771 (or_predicates): Turn to ...
7772 (predicate::or_with): ... this one.
7773 (evaluate_predicate): Turn to ...
7774 (predicate::evaluate): ... this one.
7775 (predicate_probability): Turn to ...
7776 (predicate::probability): ... this one.
7777 (dump_condition): Update.
7778 (dump_predicate): Turn to ...
7779 (predicate::dump): ... this one.
7780 (account_size_time): Update.
7781 (edge_set_predicate): Update.
7782 (set_hint_predicate): UPdate.
7783 (evaluate_conditions_for_known_args): Update.
7784 (evaluate_properties_for_edge): Update.
7785 (remap_predicate_after_duplication): Turn to...
7786 (predicate::remap_after_duplication): ... this one.
7787 (remap_hint_predicate_after_duplication): Update.
7788 (inline_summary_t::duplicate): UPdate.
7789 (dump_inline_edge_summary): Update.
7790 (dump_inline_summary): Update.
7791 (set_cond_stmt_execution_predicate): Update.
7792 (set_switch_stmt_execution_predicate): Update.
7793 (compute_bb_predicates): Update.
7794 (will_be_nonconstant_expr_predicate): Update.
7795 (will_be_nonconstant_predicate): Update.
7796 (phi_result_unknown_predicate): Update.
7797 (predicate_for_phi_result): Update.
7798 (array_index_predicate): Update.
7799 (estimate_function_body_sizes): Update.
7800 (estimate_node_size_and_time): Update.
7801 (estimate_ipcp_clone_size_and_time): Update.
7802 (remap_predicate): Rename to ...
7803 (predicate::remap_after_inlining): ... this one.
7804 (remap_hint_predicate): Update.
7805 (inline_merge_summary): Update.
7806 (inline_update_overall_summary): Update.
7807 (estimate_size_after_inlining): Update.
7808 (read_predicate): Rename to ...
7809 (predicate::stream_in): ... this one.
7810 (read_inline_edge_summary): Update.
7811 (write_predicate): Rename to ...
7812 (predicate::stream_out): ... this one.
7813 (write_inline_edge_summary): Update.
7814 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
7815 (clause_t): Turn to uint32_t
7816 (predicate): Turn to class; implement constructor and operators
7818 (size_time_entry): Update.
7819 (inline_summary): Update.
7820 (inline_edge_summary): Update.
7822 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
7824 * fold-const.c (fold_binary_loc): Move transformation...
7825 * match.pd (C - X CMP X): ... here.
7827 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
7829 * config/sparc/sparc.c (sparc_option_override): Set function
7830 alignment for -mcpu=niagara7 to 64 to match the I$ line.
7831 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
7833 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
7835 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
7837 2017-05-18 Marek Polacek <polacek@redhat.com>
7840 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
7841 (pass_ubsan::execute): Call gimple_assign_single_p instead of
7842 gimple_assign_load_p.
7844 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
7847 * real.c (do_compare): Give decimal_do_compare preference over
7848 comparing just the signs.
7850 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
7852 * doc/md.texi (Canonicalization of Instructions): Describe the
7853 canonical form of instructions that inherently set a condition
7856 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
7859 * tree-cfg.c: Move deletion of unreachable case statements to after
7860 the merging of consecutive case labels.
7862 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7864 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
7865 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
7866 restoring of callee-saved registers.
7868 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
7870 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
7871 * config/visium/visium.c (single_set_and_flags): Likewise.
7872 * config/visium/visium.md (Substitutions): Likewise.
7874 2017-05-17 Martin Liska <mliska@suse.cz>
7876 * cfg.c: Introduce dump_flags_t type and
7877 use it instead of int type.
7879 * cfghooks.c: Likewise.
7880 * cfghooks.h (struct cfg_hooks): Likewise.
7881 * cfgrtl.c: Likewise.
7882 * cfgrtl.h: Likewise.
7883 * cgraph.c (cgraph_node::get_body): Likewise.
7884 * coretypes.h: Likewise.
7885 * domwalk.c: Likewise.
7886 * domwalk.h: Likewise.
7887 * dumpfile.c (struct dump_option_value_info): Likewise.
7888 (dump_enable_all): Likewise.
7889 (dump_switch_p_1): Likewise.
7890 (opt_info_switch_p): Likewise.
7891 * dumpfile.h (enum tree_dump_index): Likewise.
7892 (struct dump_file_info): Likewise.
7893 * genemit.c: Likewise.
7894 * generic-match-head.c: Likewise.
7895 * gengtype.c (open_base_files): Likewise.
7896 * gimple-pretty-print.c: Likewise.
7897 * gimple-pretty-print.h: Likewise.
7898 * graph.c (print_graph_cfg): Likewise.
7899 * graphite-scop-detection.c (dot_all_sese): Likewise.
7900 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7901 * loop-unroll.c (report_unroll): Likewise.
7902 * passes.c (pass_manager::register_one_dump_file): Likewise.
7903 * print-tree.c: Likewise.
7904 * statistics.c: Likewise.
7905 * tree-cfg.c: Likewise.
7906 * tree-cfg.h: Likewise.
7907 * tree-dfa.c: Likewise.
7908 * tree-dfa.h: Likewise.
7909 * tree-dump.c (dump_function): Likewise.
7910 * tree-dump.h (struct dump_info): Likewise.
7911 * tree-pretty-print.c: Likewise.
7912 * tree-pretty-print.h: Likewise.
7913 * tree-ssa-live.c: Likewise.
7914 * tree-ssa-live.h: Likewise.
7915 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
7916 * tree-vect-loop.c: Likewise.
7917 * tree-vect-slp.c: Likewise.
7919 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
7920 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7922 PR tree-optimization/80457
7923 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
7924 of all arguments to a statement as scalar_to_vec operations.
7925 (vectorizable_call): Adjust call to vect_model_simple_cost for
7927 (vectorizable_conversion): Likewise.
7928 (vectorizable_assignment): Likewise.
7929 (vectorizable_shift): Likewise.
7930 (vectorizable_operation): Likewise.
7931 (vectorizable_comparison): Likewise.
7932 (vect_is_simple_cond): Record the def types for operands.
7933 (vectorizable_condition): Likewise, call vect_model_simple_cost.
7934 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
7935 for statement argument count.
7937 2017-05-16 Carl Love <cel@us.ibm.com>
7939 * config/rs6000/rs6000-c: Add support for built-in functions
7940 vector unsigned long long vec_bperm (vector unsigned long long,
7941 vector unsigned char)
7942 vector signed long long vec_mule (vector signed int,
7944 vector unsigned long long vec_mule (vector unsigned int,
7945 vector unsigned int)
7946 vector signed long long vec_mulo (vector signed int,
7948 vector unsigned long long vec_mulo (vector unsigned int,
7949 vector unsigned int)
7950 vector signed char vec_sldw (vector signed char,
7953 vector unsigned char vec_sldw (vector unsigned char,
7954 vector unsigned char,
7956 vector signed short vec_sldw (vector signed short,
7957 vector signed short,
7959 vector unsigned short vec_sldw (vector unsigned short,
7960 vector unsigned short,
7962 vector signed int vec_sldw (vector signed int,
7965 vector unsigned int vec_sldw (vector unsigned int,
7966 vector unsigned int,
7968 vector signed long long vec_sldw (vector signed long long,
7969 vector signed long long,
7971 vector unsigned long long vec_sldw (vector unsigned long long,
7972 vector unsigned long long,
7974 * config/rs6000/rs6000-c: Add support for built-in functions
7975 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
7976 * config/rs6000/altivec.h: Add defintion for vec_sldw.
7977 * doc/extend.texi: Update the built-in documentation for the
7978 new built-in functions.
7980 2017-05-16 Marek Polacek <polacek@redhat.com>
7984 * tree.c (save_expr): Don't fold the expression.
7986 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
7988 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
7989 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
7990 and (?*y,m). Update insn attributes.
7992 2017-05-16 Martin Liska <mliska@suse.cz>
7994 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
7995 flags argument of print_gimple_stmt, print_gimple_expr,
7996 print_generic_stmt and print_generic_expr.
7997 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
7998 * coretypes.h: Likewise.
7999 * except.c (dump_eh_tree): Likewise.
8000 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8001 * gimple-pretty-print.h: Likewise.
8002 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8003 (backprop::push_to_worklist): Likewise.
8004 (backprop::pop_from_worklist): Likewise.
8005 (backprop::process_use): Likewise.
8006 (backprop::intersect_uses): Likewise.
8007 (note_replacement): Likewise.
8008 * gimple-ssa-store-merging.c
8009 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8010 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8011 (pass_store_merging::execute): Likewise.
8012 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8013 (ssa_base_cand_dump_callback): Likewise.
8014 (dump_incr_vec): Likewise.
8015 (replace_refs): Likewise.
8016 (replace_mult_candidate): Likewise.
8017 (create_add_on_incoming_edge): Likewise.
8018 (create_phi_basis): Likewise.
8019 (insert_initializers): Likewise.
8020 (all_phi_incrs_profitable): Likewise.
8021 (introduce_cast_before_cand): Likewise.
8022 (replace_one_candidate): Likewise.
8023 * gimplify.c (gimplify_expr): Likewise.
8024 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8025 (set_rename): Likewise.
8026 (rename_uses): Likewise.
8027 (copy_loop_phi_nodes): Likewise.
8028 (add_close_phis_to_merge_points): Likewise.
8029 (copy_loop_close_phi_args): Likewise.
8030 (copy_cond_phi_args): Likewise.
8031 (graphite_copy_stmts_from_block): Likewise.
8032 (translate_pending_phi_nodes): Likewise.
8033 * graphite-poly.c (print_pdr): Likewise.
8034 (dump_gbb_cases): Likewise.
8035 (dump_gbb_conditions): Likewise.
8036 (print_scop_params): Likewise.
8037 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8038 (build_cross_bb_scalars_use): Likewise.
8039 (gather_bbs::before_dom_children): Likewise.
8040 * hsa-dump.c (dump_hsa_immed): Likewise.
8041 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8042 (get_replacement_map): Likewise.
8043 * ipa-inline-analysis.c (dump_condition): Likewise.
8044 (estimate_function_body_sizes): Likewise.
8045 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8046 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8047 * ipa-prop.c (ipa_dump_param): Likewise.
8048 (ipa_print_node_jump_functions_for_edge): Likewise.
8049 (ipa_modify_call_arguments): Likewise.
8050 (ipa_modify_expr): Likewise.
8051 (ipa_dump_param_adjustments): Likewise.
8052 (ipa_dump_agg_replacement_values): Likewise.
8053 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8054 * ipa-pure-const.c (check_stmt): Likewise.
8055 (pass_nothrow::execute): Likewise.
8056 * ipa-split.c (execute_split_functions): Likewise.
8057 * omp-offload.c (dump_oacc_loop_part): Likewise.
8058 (dump_oacc_loop): Likewise.
8059 * trans-mem.c (tm_log_emit): Likewise.
8060 (tm_memopt_accumulate_memops): Likewise.
8061 (dump_tm_memopt_set): Likewise.
8062 (dump_tm_memopt_transform): Likewise.
8063 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8064 (print_loop): Likewise.
8065 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8066 (chkp_gather_checks_info): Likewise.
8067 (chkp_get_check_result): Likewise.
8068 (chkp_remove_check_if_pass): Likewise.
8069 (chkp_use_outer_bounds_if_possible): Likewise.
8070 (chkp_reduce_bounds_lifetime): Likewise.
8071 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8072 (chkp_mark_completed_bounds): Likewise.
8073 (chkp_register_incomplete_bounds): Likewise.
8074 (chkp_mark_invalid_bounds): Likewise.
8075 (chkp_maybe_copy_and_register_bounds): Likewise.
8076 (chkp_build_returned_bound): Likewise.
8077 (chkp_get_bound_for_parm): Likewise.
8078 (chkp_build_bndldx): Likewise.
8079 (chkp_get_bounds_by_definition): Likewise.
8080 (chkp_generate_extern_var_bounds): Likewise.
8081 (chkp_get_bounds_for_decl_addr): Likewise.
8082 * tree-chrec.c (chrec_apply): Likewise.
8083 * tree-data-ref.c (dump_data_reference): Likewise.
8084 (dump_subscript): Likewise.
8085 (dump_data_dependence_relation): Likewise.
8086 (analyze_overlapping_iterations): Likewise.
8087 * tree-inline.c (expand_call_inline): Likewise.
8088 (tree_function_versioning): Likewise.
8089 * tree-into-ssa.c (dump_defs_stack): Likewise.
8090 (dump_currdefs): Likewise.
8091 (dump_names_replaced_by): Likewise.
8092 (dump_update_ssa): Likewise.
8093 (update_ssa): Likewise.
8094 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8095 * tree-parloops.c (build_new_reduction): Likewise.
8096 (try_create_reduction_list): Likewise.
8097 (ref_conflicts_with_region): Likewise.
8098 (oacc_entry_exit_ok_1): Likewise.
8099 (oacc_entry_exit_single_gang): Likewise.
8100 * tree-pretty-print.h: Likewise.
8101 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8102 (get_scalar_evolution): Likewise.
8103 (add_to_evolution): Likewise.
8104 (get_loop_exit_condition): Likewise.
8105 (analyze_evolution_in_loop): Likewise.
8106 (analyze_initial_condition): Likewise.
8107 (analyze_scalar_evolution): Likewise.
8108 (instantiate_scev): Likewise.
8109 (number_of_latch_executions): Likewise.
8110 (gather_chrec_stats): Likewise.
8111 (final_value_replacement_loop): Likewise.
8112 (scev_const_prop): Likewise.
8113 * tree-sra.c (dump_access): Likewise.
8114 (disqualify_candidate): Likewise.
8115 (create_access): Likewise.
8117 (maybe_add_sra_candidate): Likewise.
8118 (create_access_replacement): Likewise.
8119 (analyze_access_subtree): Likewise.
8120 (analyze_all_variable_accesses): Likewise.
8121 (sra_modify_assign): Likewise.
8122 (initialize_constant_pool_replacements): Likewise.
8123 (find_param_candidates): Likewise.
8124 (decide_one_param_reduction): Likewise.
8125 (replace_removed_params_ssa_names): Likewise.
8126 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8127 * tree-ssa-copy.c (dump_copy_of): Likewise.
8128 (copy_prop_visit_cond_stmt): Likewise.
8129 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8130 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8131 (record_equivalences_from_stmt): Likewise.
8132 * tree-ssa-dse.c (compute_trims): Likewise.
8133 (delete_dead_call): Likewise.
8134 (delete_dead_assignment): Likewise.
8135 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8136 (forward_propagate_into_cond): Likewise.
8137 (pass_forwprop::execute): Likewise.
8138 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8139 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8141 (move_computations_worker): Likewise.
8142 (execute_sm): Likewise.
8143 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8144 (remove_exits_and_undefined_stmts): Likewise.
8145 (remove_redundant_iv_tests): Likewise.
8146 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8147 (adjust_iv_update_pos): Likewise.
8148 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8149 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8150 (value_replacement): Likewise.
8151 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8152 * tree-ssa-pre.c (print_pre_expr): Likewise.
8153 (get_representative_for): Likewise.
8154 (create_expression_by_pieces): Likewise.
8155 (insert_into_preds_of_block): Likewise.
8156 (eliminate_insert): Likewise.
8157 (eliminate_dom_walker::before_dom_children): Likewise.
8158 (eliminate): Likewise.
8159 (remove_dead_inserted_code): Likewise.
8160 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8161 * tree-ssa-reassoc.c (get_rank): Likewise.
8162 (eliminate_duplicate_pair): Likewise.
8163 (eliminate_plus_minus_pair): Likewise.
8164 (eliminate_not_pairs): Likewise.
8165 (undistribute_ops_list): Likewise.
8166 (eliminate_redundant_comparison): Likewise.
8167 (update_range_test): Likewise.
8168 (optimize_range_tests_var_bound): Likewise.
8169 (optimize_vec_cond_expr): Likewise.
8170 (rewrite_expr_tree): Likewise.
8171 (rewrite_expr_tree_parallel): Likewise.
8172 (linearize_expr): Likewise.
8173 (break_up_subtract): Likewise.
8174 (linearize_expr_tree): Likewise.
8175 (attempt_builtin_powi): Likewise.
8176 (attempt_builtin_copysign): Likewise.
8177 (transform_stmt_to_copy): Likewise.
8178 (transform_stmt_to_multiply): Likewise.
8179 (dump_ops_vector): Likewise.
8180 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8181 (print_scc): Likewise.
8182 (set_ssa_val_to): Likewise.
8183 (visit_reference_op_store): Likewise.
8184 (visit_use): Likewise.
8185 (sccvn_dom_walker::before_dom_children): Likewise.
8186 (run_scc_vn): Likewise.
8187 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8189 (expr_hash_elt::print): Likewise.
8190 (const_and_copies::pop_to_marker): Likewise.
8191 (const_and_copies::record_const_or_copy_raw): Likewise.
8192 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8193 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8194 (dump_predicates): Likewise.
8195 (find_uninit_use): Likewise.
8196 (warn_uninitialized_phi): Likewise.
8197 (pass_late_warn_uninitialized::execute): Likewise.
8198 * tree-ssa.c (verify_vssa): Likewise.
8199 (verify_ssa): Likewise.
8200 (maybe_optimize_var): Likewise.
8201 * tree-vrp.c (dump_value_range): Likewise.
8202 (dump_all_value_ranges): Likewise.
8203 (dump_asserts_for): Likewise.
8204 (register_edge_assert_for_2): Likewise.
8205 (vrp_visit_cond_stmt): Likewise.
8206 (vrp_visit_switch_stmt): Likewise.
8207 (vrp_visit_stmt): Likewise.
8208 (vrp_visit_phi_node): Likewise.
8209 (simplify_cond_using_ranges_1): Likewise.
8210 (fold_predicate_in): Likewise.
8211 (evrp_dom_walker::before_dom_children): Likewise.
8212 (evrp_dom_walker::push_value_range): Likewise.
8213 (evrp_dom_walker::pop_value_range): Likewise.
8214 (execute_early_vrp): Likewise.
8216 2017-05-16 Richard Biener <rguenther@suse.de>
8218 * dwarf2out.c (loc_list_from_tree_1): Do not create
8219 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8221 2017-05-16 Richard Biener <rguenther@suse.de>
8223 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8225 (note_variable_value_in_expr): If we resolved the decl ref
8226 do not push to the stack.
8228 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8230 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8231 operations in fast-math mode.
8232 (vaddq_f16): Likewise.
8233 (vmul_f16): Likewise.
8234 (vmulq_f16): Likewise.
8235 (vsub_f16): Likewise.
8236 (vsubq_f16): Likewise.
8237 * config/arm/neon.md (add<mode>3): New.
8239 (fma:<VH:mode>3): New. Also remove outdated comment.
8242 2017-05-16 Martin Liska <mliska@suse.cz>
8246 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8247 (odr_types_equivalent_p): Likewise.
8249 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8251 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8253 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8256 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8257 non-interunit SSE move alternatives with '?'.
8258 (zero-extendsidi peephole2): New peephole to skip intermediate
8259 general register in SSE zero-extend sequence.
8261 2017-05-15 Jeff Law <law@redhat.com>
8263 * reorg.c (relax_delay_slots): Create a new variable to hold
8264 the temporary target rather than clobbering TARGET_LABEL.
8266 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8267 missing argument to extract_bit_field call.
8268 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8270 2017-05-15 Martin Liska <mliska@suse.cz>
8273 * gcc.c (process_command): Do not allow empty argument of -o option.
8275 2017-05-15 Renlin Li <renlin.li@arm.com>
8277 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8278 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8279 * config/aarch64/constraints.md (Usf): Add long call check.
8280 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8281 (call_value): Likewise.
8282 (sibcall): Likewise.
8283 (sibcall_value): Likewise.
8285 (call_value_insn): New.
8286 (sibcall_insn): Update rtx pattern.
8287 (sibcall_value_insn): Likewise.
8288 (call_internal): Remove.
8289 (call_value_internal): Likewise.
8290 (sibcall_internal): Likewise.
8291 (sibcall_value_internal): Likewise.
8292 (call_reg): Likewise.
8293 (call_symbol): Likewise.
8294 (call_value_reg): Likewise.
8295 (call_value_symbol): Likewise.
8297 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8300 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8302 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8304 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8305 compatible with CCGOCmode and with CCZmode.
8307 2017-05-14 Martin Sebor <msebor@redhat.com>
8310 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8311 (gimple_fold_builtin_snprintf): Same.
8312 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8313 (gimple_fold_builtin_snprintf): Same.
8314 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8316 (is_call_safe): New function.
8317 (try_substitute_return_value): Call it.
8318 (try_simplify_call): New function.
8319 (pass_sprintf_length::handle_gimple_call): Call it.
8321 2017-05-14 Martin Sebor <msebor@redhat.com>
8324 * builtins.c (expand_builtin_stpncpy): Simplify.
8326 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8328 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8329 * config/i386/i386.h
8330 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8331 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8332 (struct machine_function): Add new members call_ms2sysv,
8333 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8334 (struct machine_frame_state): New fields sp_realigned and
8335 sp_realigned_offset.
8336 * config/i386/i386.c
8337 (enum xlogue_stub): New enum.
8338 (enum xlogue_stub_sets): New enum.
8339 (class xlogue_layout): New class.
8340 (struct ix86_frame): New fields stack_realign_allocate_offset,
8341 stack_realign_offset and outlined_save_offset. Modify comments to
8342 detail stack layout when using out-of-line stubs.
8343 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8344 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8345 -mcall-ms2sysv-xlogues.
8346 (stub_managed_regs): New static variable.
8347 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8348 registers managed by out-of-line stub.
8349 (disable_call_ms2sysv_xlogues): New function.
8350 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8351 m->call_ms2sysv when appropriate and compute frame layout for
8353 (sp_valid_at, fp_valid_at): New inline functions.
8354 (choose_basereg): New function.
8355 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8357 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8358 Use align parameter of choose_baseaddr to generated aligned SSE movs
8360 (pro_epilogue_adjust_stack): Modify to track
8361 machine_frame_state::sp_realigned.
8362 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8363 (ix86_nsaved_sseregs): Likewise.
8364 (ix86_emit_save_regs): Likewise.
8365 (ix86_emit_save_regs_using_mov): Likewise.
8366 (ix86_emit_save_sse_regs_using_mov): Likewise.
8367 (get_scratch_register_on_entry): Likewise.
8368 (gen_frame_set): New function.
8369 (gen_frame_load): Likewise.
8370 (gen_frame_store): Likewise.
8371 (emit_outlined_ms2sysv_save): Likewise.
8372 (emit_outlined_ms2sysv_restore): Likewise.
8373 (ix86_expand_prologue): Modify stack re-alignment code and call
8374 emit_outlined_ms2sysv_save when appropriate.
8375 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8376 parameter rtx_insn *insn, which allows the function to be used to only
8378 (ix86_expand_epilogue): Modify validity checks of frame and stack
8379 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8380 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8381 * config/i386/predicates.md
8382 (save_multiple): New predicate.
8383 (restore_multiple): Likewise.
8384 * config/i386/sse.md
8385 (save_multiple<mode>): New pattern.
8386 (save_multiple_realign<mode>): Likewise.
8387 (restore_multiple<mode>): Likewise.
8388 (restore_multiple_and_return<mode>): Likewise.
8389 (restore_multiple_leave_return<mode>): Likewise.
8390 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8392 2017-05-14 Julia Koval <julia.koval@intel.com>
8394 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8395 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8396 (__builtin_ia32_xsetbv): New builtins.
8397 * config/i386/i386.c (ix86_expand_special_args_builtin):
8399 (ix86_expand_builtin): Special expand for new intrinsics.
8400 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8401 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8402 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8404 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8406 * cfganal.c (inverted_post_order_compute): Change argument type
8408 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8409 * df-core.c (rest_of_handle_df_initialize): Adjust.
8410 (rest_of_handle_df_finish): Likewise.
8411 (df_analyze_1): Likewise.
8412 (df_analyze): Likewise.
8413 (loop_inverted_post_order_compute): Change argument to be a vec *.
8414 (df_analyze_loop): Adjust.
8415 (df_get_n_blocks): Likewise.
8416 (df_get_postorder): Likewise.
8417 * df.h (struct df_d): Change field to be a vec.
8418 * lcm.c (compute_laterin): Adjust.
8419 (compute_available): Likewise.
8420 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8421 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8422 * tree-ssa-pre.c (compute_antic): Likewise.
8424 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8426 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8427 (depth_first_search::depth_first_search): Change structure init
8428 function to this constructor.
8429 (depth_first_search::add_bb): Rename function to this member.
8430 (depth_first_search::execute): Likewise.
8431 (flow_dfs_compute_reverse_finish): Adjust.
8433 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8435 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8436 (longest_simple_path): Likewise.
8437 * shrink-wrap.c (spread_components): Likewise.
8438 (disqualify_problematic_components): Likewise.
8439 (emit_common_heads_for_components): Likewise.
8440 (emit_common_tails_for_components): Likewise.
8441 (insert_prologue_epilogue_for_components): Likewise.
8443 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8445 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8448 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8450 * df-core.c (df_set_blocks): Start using auto_bitmap.
8451 (df_compact_blocks): Likewise.
8452 * df-problems.c (df_rd_confluence_n): Likewise.
8453 * df-scan.c (df_insn_rescan_all): Likewise.
8454 (df_process_deferred_rescans): Likewise.
8455 (df_update_entry_block_defs): Likewise.
8456 (df_update_exit_block_uses): Likewise.
8457 (df_entry_block_bitmap_verify): Likewise.
8458 (df_exit_block_bitmap_verify): Likewise.
8459 (df_scan_verify): Likewise.
8460 * lra-constraints.c (lra_constraints): Likewise.
8461 (undo_optional_reloads): Likewise.
8462 (lra_undo_inheritance): Likewise.
8463 * lra-remat.c (calculate_gen_cands): Likewise.
8464 (do_remat): Likewise.
8465 * lra-spills.c (assign_spill_hard_regs): Likewise.
8466 (spill_pseudos): Likewise.
8467 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8468 (bitmap_set_subtract_values): Likewise.
8470 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8472 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8473 management with auto_bitmap.
8474 (fix_inter_tick): Likewise.
8475 (fix_recovery_deps): Likewise.
8476 * ira.c (add_store_equivs): Likewise.
8477 (find_moveable_pseudos): Likewise.
8478 (split_live_ranges_for_shrink_wrap): Likewise.
8479 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8480 (rtx_reuse_manager::seen_def_p): Likewise.
8481 (rtx_reuse_manager::set_seen_def): Likewise.
8482 * print-rtl.h (class rtx_reuse_manager): Likewise.
8484 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8486 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8488 (migrate_btr_def): Likewise.
8489 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8490 * df-core.c (loop_post_order_compute): Likewise.
8491 (loop_inverted_post_order_compute): Likewise.
8492 * hsa-common.h: Likewise.
8493 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8494 * init-regs.c (initialize_uninitialized_regs): Likewise.
8495 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8496 (inline_small_functions): Likewise.
8497 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8498 * ira.c (combine_and_move_insns): Likewise.
8499 (build_insn_chain): Likewise.
8500 * loop-invariant.c (find_invariants): Likewise.
8501 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8502 * predict.c (tree_predict_by_opcode): Likewise.
8503 (predict_paths_leading_to): Likewise.
8504 (predict_paths_leading_to_edge): Likewise.
8505 (estimate_loops_at_level): Likewise.
8506 (estimate_loops): Likewise.
8507 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8508 (spread_components): Likewise.
8509 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8510 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8511 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8512 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8513 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8514 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8515 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8516 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8517 (mark_threaded_blocks): Likewise.
8518 (thread_through_all_blocks): Likewise.
8519 * tree-ssa.c (verify_ssa): Likewise.
8520 (execute_update_addresses_taken): Likewise.
8521 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8523 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8525 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8527 (post_order_compute): Likewise.
8528 (inverted_post_order_compute): Likewise.
8529 (pre_and_rev_post_order_compute_fn): Likewise.
8531 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8533 * genrecog.c (int_set::int_set): Explicitly construct our
8534 auto_vec base class.
8535 * vec.h (auto_vec::auto_vec): New constructor.
8537 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8539 * bitmap.h (class auto_bitmap): New constructor taking
8540 bitmap_obstack * argument.
8542 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8544 * bitmap.h (class auto_bitmap): Change type of m_bits to
8545 bitmap_head, and adjust ctor / dtor and member operators.
8547 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8549 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8550 when returned register mode doesn't match original mode.
8552 2017-05-12 Jeff Law <law@redhat.com>
8553 Jakub Jelinek <jakub@redhat.com>
8555 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8556 we look for cc setter after the compare-elim changes.
8557 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8558 within the vector to match what compare-elim now expects.
8559 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8560 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8562 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8563 after the compare-elim changes.
8564 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8565 the vector to match what compare-elim now expects.
8566 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8567 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8568 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8569 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8570 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8572 * config/visium/visium.c (single_set_and_flags): Fix where
8573 we look for cc setter after the compare-elim changes.
8574 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8575 with the vector to match what compare-elim now expects.
8576 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8577 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8578 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8579 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8580 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8581 (neg<mode>2_insn_set_overflow): Likewise.
8583 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8586 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8587 maybe_expand_insn call, set ops[0].target. If still set after call,
8588 set alt_rtl. Add extra arg to recursive calls.
8589 (extract_bit_field): Add alt_rtl argument. Pass to
8591 * expmed.h (extract_bit_field): Fix prototype.
8592 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8593 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8594 to extract_bit_field_calls.
8595 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8596 Pass alt_rtl to extract_bit_field calls.
8597 * calls.c (store_unaligned_arguments_into_psuedos)
8598 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8599 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8601 * optabs.h (struct expand_operand): Add target bitfield.
8603 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8605 * compare-elim.c (try_eliminate_compare): Canonicalize
8606 operation with embedded compare to
8607 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8608 (set (reg) (operation)].
8610 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8612 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8615 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8616 cost of adding a carry flag for ADC instruction.
8617 [case MINUS]: Ignore the cost of subtracting a carry flag
8618 for SBB instruction.
8620 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8622 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8624 * config/rs6000/bmiintrin.h: New file.
8625 * config/rs6000/bmi2intrin.h: New file.
8626 * config/rs6000/x86intrin.h: New file.
8628 2017-05-12 Jeff Law <law@redhat.com>
8630 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8633 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8636 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8638 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8640 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8641 early expansion of vector divide builtins.
8642 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8643 builtins identified as having unsigned arguments.
8645 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8647 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8648 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8649 expansion of vector logical operations (and, andc, or, xor,
8652 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8654 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8655 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8657 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8659 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8660 early GIMPLE expansion of vector multiplies.
8662 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8664 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8665 TARGET_HAVE_MOVT conditional.
8666 (movt splitter): Likewise.
8668 2017-05-12 Richard Biener <rguenther@suse.de>
8670 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8671 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8672 Fold all stmts not inplace.
8674 2017-05-12 Richard Biener <rguenther@suse.de>
8676 PR tree-optimization/80713
8677 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8678 inserted_exprs bit for not removed stmts.
8680 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8683 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8684 parallelized" attribute for parallelized OpenACC kernels.
8685 * omp-offload.c (execute_oacc_device_lower): Use it.
8687 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8688 Set "oacc kernels" attribute.
8689 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8690 parameter. Adjust all users.
8691 (oacc_fn_attrib_kernels_p): Remove function.
8692 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8693 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8694 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8695 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8696 assert "oacc kernels" attribute is set.
8698 2017-05-11 Carl Love <cel@us.ibm.com>
8700 * config/rs6000/rs6000-c: Add support for built-in functions
8701 vector unsigned char vec_popcnt (vector signed char)
8702 vector unsigned char vec_popcnt (vector unsigned char)
8703 vector unsigned short vec_popcnt (vector signed short)
8704 vector unsigned short vec_popcnt (vector unsigned short)
8705 vector unsigned int vec_popcnt (vector signed int)
8706 vector unsigned int vec_popcnt (vector unsigned int)
8707 vector unsigned long long vec_popcnt (vector signed long long)
8708 vector unsigned long long vec_popcnt (vector unsigned long long)
8709 vector signed long long vec_slo (vector signed long long,
8711 vector signed long long vec_slo (vector signed long long,
8712 vector unsigned char)
8713 vector unsigned long long vec_slo (vector unsigned long long,
8715 vector unsigned long long vec_slo (vector unsigned long long,
8716 vector unsigned char)
8717 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8718 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8719 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8720 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
8721 * doc/extend.texi: Update the built-in documentation file for the
8722 new built-in functions.
8724 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8726 * attribs.h (sorted_attr_string): Move machine independent
8727 functions for target clone support from the i386 port to common
8728 code. Rename ix86_function_versions to common_function_versions.
8729 Rename make_name to make_unique_name.
8730 (common_function_versions): Likewise.
8731 (make_unique_name): Likewise.
8732 (make_dispatcher_decl): Likewise.
8733 (is_function_default_version): Likewise.
8734 * attribs.c (attr_strcmp): Likewise.
8735 (sorted_attr_string): Likewise.
8736 (common_function_versions): Likewise.
8737 (make_unique_name): Likewise.
8738 (make_dispatcher_decl): Likewise.
8739 (is_function_default_version): Likewise.
8740 * config/i386/i386.c (attr_strcmp): Likewise.
8741 (sorted_attr_string): Likewise.
8742 (ix86_function_versions): Likewise.
8743 (make_name): Likewise.
8744 (make_dispatcher_decl): Likewise.
8745 (is_function_default_version): Likewise.
8746 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
8748 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8751 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8752 Account for direct move costs for vec_construct of integer
8755 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
8758 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
8759 (UNSPEC_STX_ATOMIC): Ditto.
8760 (loaddi_via_sse): New insn.
8761 (storedi_via_sse): Ditto.
8762 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
8763 Update corresponding peephole2 patterns.
8764 (atomic_storedi_fpu): Ditto.
8766 2017-05-11 Julia Koval <julia.koval@intel.com>
8768 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
8769 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
8771 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
8772 (__builtin_ia32_rsqrt14ss_mask): New builtins.
8773 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
8775 2017-05-11 Nathan Sidwell <nathan@acm.org>
8777 * graphite-poly.c: Include dumpfile.h.
8779 * dumpfle.h (dump_function): Declare here ...
8780 * tree-dump.h (dump_function): ... not here.
8781 * dumpfile.c: #include tree-cfg.h.
8782 (dump_function): Move here from ...
8783 * tree-dump.c (dump_function): ... here.
8784 * gimplify.c: #include splay-tree.h, not tree-dump.h.
8785 * graphite-poly.c: Don't include tree-dump.h.
8786 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
8787 * print-tree.c: Likewise.
8788 * stor-layout.c: Likewise.
8789 * tree-nested.c: Likewise.
8791 * dumpfile.c (dump_start): Use TDF_FLAGS.
8792 (dump_enable_all): Fix TDF_KIND check thinko.
8794 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8796 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8797 array entries to represent two legal parameterizations of the
8798 overloaded __builtin_cmpb function, as represented by the
8799 P6_OV_BUILTIN_CMPB constant.
8800 (altivec_resolve_overloaded_builtin): Add special case handling
8801 for the __builtin_cmpb function, as represented by the
8802 P6_OV_BUILTIN_CMPB constant.
8803 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
8804 (BU_P6_64BIT_2): New macro.
8805 (BU_P6_OVERLOAD_2): New macro
8806 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
8807 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
8808 (CMPB): Add overload support to represent both 32-bit and 64-bit
8809 compare-bytes function.
8810 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8811 support for TARGET_CMPB.
8812 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
8813 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8814 documentation of the __builtin_cmpb overloaded built-in function.
8816 2017-05-11 Richard Biener <rguenther@suse.de>
8818 PR tree-optimization/80705
8819 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
8820 bases are not vectorizable.
8822 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8824 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
8825 when counting register pressure.
8827 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8829 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
8830 (iv_ca_more_deps): Renamed to ...
8831 (iv_ca_compare_deps): ... this.
8832 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
8834 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8836 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
8838 (determine_group_iv_costs): ... here.
8839 (find_inv_vars_cb): Record inv var if it's not recorded before.
8841 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8843 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
8844 (get_shiftadd_cost): Ditto.
8846 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8848 * tree-ssa-address.c: Include header file.
8849 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
8851 (add_to_parts): Refactor.
8852 (addr_to_parts): New parameter. Update use of move_hint_to_base.
8853 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
8856 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8858 PR tree-optimization/53090
8859 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
8861 (extract_cond_operands): Detect condition with IV on both sides
8862 and return COMP_IV_EXPR_2.
8863 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
8864 (rewrite_use_compare): Simplify by removing call to function
8865 extract_cond_operands.
8867 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8869 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
8870 (extract_cond_operands): Detect condition comparing against non-
8871 invariant bound and return appropriate enum value.
8872 (find_interesting_uses_cond): Update use of extract_cond_operands.
8873 Handle its return value accordingly.
8874 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
8876 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8878 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
8879 nonlinear iv_use computation in loop invariant sensitive way.
8881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8883 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
8884 (find_iv_candidates): Call relate_compare_use_with_all_cands.
8886 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8888 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
8889 (dump_cand): Support iv_cand.inv_exprs.
8890 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
8892 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
8895 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8897 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
8899 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
8900 as local function. Include necessary header files.
8901 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
8903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8905 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
8907 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8909 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
8910 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
8911 RSHIFT_EXPR and BIT_NOT_EXPR.
8913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8915 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
8916 (adjust_setup_cost): New parameter supporting round up adjustment.
8917 (struct address_cost_data): Delete.
8918 (force_expr_to_var_cost): Don't bound cost with spill_cost.
8919 (split_address_cost, ptr_difference_cost): Delete.
8920 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
8921 (struct ainc_cost_data): New struct.
8922 (get_address_cost_ainc): New function.
8923 (get_address_cost, get_computation_cost): Reimplement.
8924 (determine_group_iv_cost_address): Record inv_expr for all uses of
8926 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
8927 (iv_ca_has_deps): Reimplemented to ...
8928 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
8930 (iv_ca_extend): Call iv_ca_more_deps.
8932 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8934 * tree-ssa-address.c (struct mem_address): Move to header file.
8935 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
8936 * tree-ssa-address.h (struct mem_address): Move from C file.
8937 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
8939 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8941 * tree-affine.h (aff_combination_type): New interface.
8942 (aff_combination_zero_p): Remove static.
8943 (aff_combination_const_p): New interface.
8944 (aff_combination_singleton_var_p): New interfaces.
8946 2017-05-11 Richard Biener <rguenther@suse.de>
8948 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8949 Skip unreachable blocks and destinations.
8950 (eliminate): Move stmt removal and fixup ...
8951 (fini_eliminate): ... here. Skip inserted exprs.
8952 (pass_pre::execute): Move fini_pre after fini_eliminate.
8953 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
8954 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
8955 PRE to get rid of dead code that has invalid SSA form and
8956 split critical edges again.
8958 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8960 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8962 2017-05-11 Richard Biener <rguenther@suse.de>
8964 * passes.c (execute_function_todo): Verify loops if they are
8965 said to be up-to-date.
8966 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
8967 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
8969 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
8972 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
8973 handle calling assemble_external ourself.
8976 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
8977 modes with zero size. Enhance comment.
8979 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8981 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
8982 built-ins for vec_xl and vec_xst with short and char pointer
8985 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8987 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
8988 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
8989 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
8990 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
8991 (_mm_maskz_min_round_ss): New intrinsics.
8992 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
8993 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
8994 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
8995 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
8996 (__builtin_ia32_minss_mask_round): New builtins.
8997 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8998 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8999 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9001 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9002 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9004 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9007 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9009 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9010 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9011 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9012 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9013 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9014 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9015 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9016 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9017 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9018 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9019 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9020 (__builtin_ia32_mulss_mask_round): New builtins.
9021 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9022 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9023 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9025 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9026 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9028 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9031 2017-05-10 Julia Koval <julia.koval@intel.com>
9033 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9034 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9035 (_mm256_setr_m128i): New intrinsics.
9037 2017-05-10 Julia Koval <julia.koval@intel.com>
9039 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9040 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9041 (_mm_maskz_rcp14_ss): New intrinsics.
9042 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9043 (__builtin_ia32_rcp14ss_mask): New builtins.
9044 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9046 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9048 PR tree-optimization/51513
9049 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9050 (assert_unreachable_fallthru_edge_p): Use it.
9051 (group_case_labels_stmt): Likewise.
9052 * tree-cfg.h: Prototype it.
9053 * stmt.c: Include cfghooks.h and tree-cfg.h.
9054 (emit_case_dispatch_table) <gap_label>: New local variable.
9055 Use it to fill dispatch table gaps.
9056 Test for default_label before updating probabilities.
9057 (expand_case) <default_label>: Remove unneeded initialization.
9058 Test for unreachable default case statement and remove its edge.
9059 Set default_label accordingly.
9060 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9062 2017-05-10 Carl Love <cel@us.ibm.com>
9064 * config/rs6000/rs6000-c: Add support for built-in functions
9065 vector signed char vec_neg (vector signed char)
9066 vector signed short int vec_neg (vector short int)
9067 vector signed int vec_neg (vector signed int)
9068 vector signed long long vec_neg (vector signed long long)
9069 vector float vec_neg (vector float)
9070 vector double vec_neg (vector double)
9071 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9073 * config/rs6000/altivec.h: Add define for vec_neg
9074 * doc/extend.texi: Update the built-in documentation for the
9075 new built-in functions.
9077 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9079 PR tree-optimization/77644
9080 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9082 2017-05-10 Nathan Sidwell <nathan@acm.org>
9084 * dumpfile.h (TDI_lang_all): New.
9085 (TDF_KIND): New. Renumber others
9086 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9088 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9090 (get_dump_file_name): Adjust suffix generation.
9091 (dump_enable_all): Use TDF_KIND.
9092 * doc/invoke.texi (-fdump-lang-all): Document.
9094 * dumpfile.h: Tabify.
9096 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9099 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9100 Move member access before delete.
9102 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9104 * tree-inline.c (expand_call_inline): Split block at stmt
9107 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9110 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9111 are now unused after splitting mov{sf,sd}_hardfloat.
9112 (f32_lr2): Likewise.
9114 (f32_lm2): Likewise.
9116 (f32_li2): Likewise.
9119 (f32_sr2): Likewise.
9121 (f32_sm2): Likewise.
9123 (f32_si2): Likewise.
9126 (f32_vsx): Likewise.
9128 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9129 For movsf, order stores so the VSX stores occur before the GPR
9130 store which encourages the register allocator to use a traditional
9131 FPR instead of a GPR. For movsd, order the stores so that the GPR
9132 store comes before the VSX stores to allow the power6 to work.
9133 This is due to the power6 not having a 32-bit integer store
9134 instruction from a FPR.
9135 (movsf_hardfloat): Likewise.
9136 (movsd_hardfloat): Likewise.
9138 2017-05-09 Martin Sebor <msebor@redhat.com>
9140 PR translation/80280
9141 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9144 PR translation/80280
9145 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9146 data member added in r247778.
9147 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9149 2017-05-09 Nathan Sidwell <nathan@acm.org>
9151 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9153 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9156 2017-05-09 Marek Polacek <polacek@redhat.com>
9158 * doc/invoke.texi: Fix typo.
9160 2017-05-09 Richard Biener <rguenther@suse.de>
9162 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9163 (vrp_val_is_min): Likewise.
9164 (set_value_range_to_value): Likewise.
9165 (set_value_range_to_nonnegative): Likewise.
9166 (gimple_assign_nonzero_p): Likewise.
9167 (gimple_stmt_nonzero_p): Likewise.
9168 (vrp_int_const_binop): Likewise. Remove unreachable case.
9169 (adjust_range_with_scev): Adjust comments.
9170 (compare_range_with_value): Likewise.
9171 (extract_range_from_phi_node): Likewise.
9172 (test_for_singularity): Likewise.
9174 2017-05-09 Richard Biener <rguenther@suse.de>
9176 * tree-vrp.c (get_single_symbol): Add assert that we don't
9177 get overflowed constants as invariant part.
9178 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9179 checks. Use wi::cmp instead of recursing for integer constants.
9180 (compare_values): Just ignore whether we assumed undefined
9181 overflow instead of failing the compare.
9182 (extract_range_for_var_from_comparison_expr): Add comment before the
9183 TREE_NO_WARNING sets.
9184 (test_for_singularity): Likewise.
9185 (extract_range_from_comparison): Do not disable optimization
9186 when we assumed undefined overflow.
9187 (extract_range_basic): Remove init of unused var.
9189 2017-05-09 Richard Biener <rguenther@suse.de>
9191 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9192 (extract_range_from_multiplicative_op_1): Adjust.
9193 (extract_range_from_binary_expr_1): Use int_const_binop.
9195 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9198 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9199 rs6000_store_data_bypass_p in seven define_bypass directives and
9200 in several comments.
9201 * config/rs6000/rs6000-protos.h: Add prototype for
9202 rs6000_store_data_bypass_p function.
9203 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9204 function implements slightly different (rs6000-specific) semantics
9205 than store_data_bypass_p, returning false rather than aborting
9206 with assertion error when arguments do not satisfy the
9207 requirements of store data bypass.
9208 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9209 rs6000_store_data_bypass_p.
9211 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9213 * config/xtensa/xtensa-protos.h
9214 (xtensa_initial_elimination_offset): New declaration.
9215 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9216 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9217 macro definition, add case for FRAME_POINTER_REGNUM when
9218 FRAME_GROWS_DOWNWARD.
9219 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9220 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9221 xtensa_initial_elimination_offset.
9223 2017-05-08 Nathan Sidwell <nathan@acm.org>
9225 * doc/invoke.texi: Alphabetize -fdump options.
9227 2017-05-08 Martin Sebor <msebor@redhat.com>
9229 PR translation/80280
9230 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9232 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9234 * target.def (compute_frame_layout): New optional target hook.
9235 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9236 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9237 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9239 * reload1.c (verify_initial_elim_offsets): Likewise.
9240 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9241 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9242 (arm_get_frame_offsets): Split up into this ...
9243 (arm_compute_frame_layout): ... and this function.
9245 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9247 * config/aarch64/constraints.md (Usa): New constraint.
9248 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9250 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9252 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9253 with_multilib_list after it has been checked.
9255 2017-05-08 Richard Biener <rguenther@suse.de>
9257 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9258 (bitmap_set_subtract_values): Likewise.
9260 2017-05-08 Richard Biener <rguenther@suse.de>
9262 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9263 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9265 (gimple_stmt_nonzero_warnv_p): Rename to ...
9266 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9268 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9269 (extract_range_basic): Adjust, do not disable propagation on
9270 strict overflow sensitive simplification.
9271 (vrp_visit_cond_stmt): Likewise.
9273 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9275 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9276 body size unconditionally.
9278 2017-05-07 Jeff Law <law@redhat.com>
9281 2017-05-06 Jeff Law <law@redhat.com>
9282 PR tree-optimization/78496
9283 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9286 PR tree-optimization/78496
9287 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9288 (simplify_stmt_using_ranges): Call it.
9289 (vrp_dom_walker::before_dom_children): Extract equivalences
9290 from an ASSERT_EXPR with an equality comparison against a
9293 2017-05-06 Jeff Law <law@redhat.com>
9295 PR tree-optimization/78496
9296 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9299 PR tree-optimization/78496
9300 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9301 (simplify_stmt_using_ranges): Call it.
9302 (vrp_dom_walker::before_dom_children): Extract equivalences
9303 from an ASSERT_EXPR with an equality comparison against a
9306 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9308 * lra-constraints.c (lra_copy_reg_equiv): New function.
9309 (split_reg): Use it to copy equivalence information from the
9310 original register to the spill register.
9312 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9314 PR rtl-optimization/75964
9315 * simplify-rtx.c (simplify_const_relational_operation): Remove
9316 invalid handling of comparisons of integer ABS.
9318 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9320 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9322 (init_regs): Remove declaration.
9323 (function_arg_advance_32): Initialize error_p as boolean variable.
9325 2017-05-05 Nathan Sidwell <nathan@acm.org>
9327 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9328 lines. Use for (;;).
9330 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9332 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9333 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9334 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9335 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9336 VF=2 that require versioning.
9338 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9340 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9343 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9345 * diagnostic.h (diagnostic_override_option_index): Convert from
9346 macro to inline function.
9348 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9350 * diagnostic.c (last_module_changed_p): New function.
9351 (set_last_module): New function.
9352 (diagnostic_report_current_module): Convert macro usage to
9353 the above functions.
9354 * diagnostic.h (diagnostic_context::last_module): Strengthen
9355 from const line_map * to const line_map_ordinary *.
9356 (diagnostic_last_module_changed): Delete macro.
9357 (diagnostic_set_last_module): Delete macro.
9359 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9361 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9362 with diagnostic_report_diagnostic.
9363 (diagnostic_n_impl_richloc): Likewise.
9364 * diagnostic.h (report_diagnostic): Delete macro.
9365 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9366 with diagnostic_report_diagnostic.
9367 * substring-locations.c (format_warning_va): Likewise.
9369 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9371 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9372 save/restor of format_spec. Move option-printing code to...
9373 (print_option_information): ...this new function, and
9374 reimplement by simply printing to the pretty_printer,
9375 rather than appending to the format string.
9377 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9379 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9380 handling logic into...
9381 (update_effective_level_from_pragmas): ...this new function.
9383 2017-05-04 Andrew Waterman <andrew@sifive.com>
9385 * config/riscv/riscv.opt (mstrict-align): New option.
9386 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9387 (SLOW_UNALIGNED_ACCESS): Define.
9388 (riscv_slow_unaligned_access): Declare.
9389 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9391 (riscv_slow_unaligned_access): New variable.
9392 (rocket_tune_info): Set slow_unaligned_access to true.
9393 (optimize_size_tune_info): Set slow_unaligned_access to false.
9394 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9395 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9396 (riscv_option_override): Set riscv_slow_unaligned_access.
9397 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9399 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9401 * config/riscv/riscv.md: Unify indentation.
9403 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9408 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9410 (extendsi<mode>2): Add support for doing sign extension via
9411 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9412 don't have ISA 3.0 instructions.
9413 (extendsi<mode>2 splitter): Likewise.
9414 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9415 generate the normal insns since SImode can now go in vector
9416 registers. Disallow the special UNSPECs needed for previous
9417 machines to hide SImode being used. Add new insns
9418 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9419 (fix_trunc<mode>si2_stfiwx): Likewise.
9420 (fix_trunc<mode>si2_internal): Likewise.
9421 (fixuns_trunc<mode>si2): Likewise.
9422 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9423 (fctiw<u>z_<mode>_smallint): Likewise.
9424 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9425 of floating point to 32-bit integer from doing a direct move to
9426 the GPR registers to do a store.
9427 (fctiwz_<mode>): Break long line.
9429 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9431 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9432 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9433 (addr_list, addr_offset_valid_p): New.
9434 (split_address_groups): Check offset validity with above function.
9435 (gt-tree-ssa-loop-ivopts.h): Include header file.
9437 2017-05-05 Nathan Sidwell <nathan@acm.org>
9439 * config.gcc (arm*-*-*): Add missing 'fi'.
9441 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9443 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9444 included in -fopt-info does not matter.
9445 * doc/optinfo.texi (-fopt-info): Fix description of default
9446 behavour. Explicitly say order of options included in -fopt-info
9449 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9451 * config.gcc: Allow combinations of aprofile and rmprofile values for
9452 --with-multilib-list.
9453 * config/arm/t-multilib: New file.
9454 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9455 variables. Remove setting of ISA and floating-point ABI in
9456 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9457 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9458 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9459 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9461 * config/arm/t-rmprofile: Likewise except for the matches changes.
9462 * doc/install.texi (--with-multilib-list): Document the combination of
9463 aprofile and rmprofile values and warn about pitfalls in doing that.
9465 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9467 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9468 (movdi_aarch64): Likewise.
9470 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9472 PR tree-optimization/80632
9473 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9475 (build_arrays): Initialize it for virtual phis.
9476 (fix_phi_nodes): Use it for virtual phis.
9478 PR tree-optimization/80558
9479 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9480 [x, y] op z into [x op, y op z] for op & or | if conditions
9483 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9484 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9487 * config/arm/arm.md (use_literal_pool): Remove.
9488 (64-bit immediate split): No longer takes cost into consideration
9489 if arm_disable_literal_pool is enabled.
9490 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9491 used when arm_disable_literal_pool is enabled.
9492 (arm_max_const_double_inline_cost): Remove use of
9493 arm_disable_literal_pool.
9494 (push_minipool_fix): Add assert.
9495 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9496 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9497 (no_literal_pool_sf_immediate): New.
9499 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9501 PR tree-optimization/80613
9502 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9503 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9505 2017-05-05 Richard Biener <rguenther@suse.de>
9507 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9509 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9511 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9512 of this flag from insn conditions due to removal from r247495.
9514 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9516 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9518 (arm_early_store_addr_dep_ptr): Likewise.
9519 * config/arm/aarch-common-protos.h
9520 (arm_early_load_addr_dep_ptr): Add prototype.
9521 (arm_early_store_addr_dep_ptr): Likewise.
9522 * config/arm/cortex-a53.md: Add new bypasses.
9524 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9526 * tree.c (next_type_uid): Change type to unsigned.
9527 (type_hash_canon): Decrement back next_type_uid if
9528 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9529 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9532 2017-05-04 Martin Sebor <msebor@redhat.com>
9534 * builtins.c: Fix a trivial typo in a comment.
9537 * builtins.c (check_sizes): Adjust to handle reading past the end.
9538 Avoid printing excessive upper bound of ranges. Use %E to print
9539 tree nodes instead of converting them to %wu.
9540 (expand_builtin_memchr): New function.
9541 (compute_dest_size): Rename...
9542 (compute_objsize): ...to this.
9543 (expand_builtin_memcpy): Adjust.
9544 (expand_builtin_mempcpy): Adjust.
9545 (expand_builtin_strcat): Adjust.
9546 (expand_builtin_strcpy): Adjust.
9547 (check_strncat_sizes): Adjust.
9548 (expand_builtin_strncat): Adjust.
9549 (expand_builtin_strncpy): Adjust and simplify.
9550 (expand_builtin_memset): Adjust.
9551 (expand_builtin_bzero): Adjust.
9552 (expand_builtin_memcmp): Adjust.
9553 (expand_builtin): Handle memcmp.
9554 (maybe_emit_chk_warning): Check strncat just once.
9556 2017-05-04 Martin Sebor <msebor@redhat.com>
9558 PR preprocessor/79214
9561 * builtins.c (check_sizes): Add inlining context and issue
9562 warnings even when -Wno-system-headers is set.
9563 (check_strncat_sizes): Same.
9564 (expand_builtin_strncat): Same.
9565 (expand_builtin_memmove): New function.
9566 (expand_builtin_stpncpy): Same.
9567 (expand_builtin): Handle memmove and stpncpy.
9569 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9571 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9572 which is not used any more.
9574 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9576 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9578 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9580 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9581 (cortexa53_tunings): Likewise.
9582 (cortexa57_tunings): Likewise.
9583 (cortexa72_tunings): Likewise.
9584 (cortexa73_tunings): Likewise.
9586 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9588 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9589 Set loop alignment to 8.
9591 2017-05-04 Martin Sebor <msebor@redhat.com>
9593 PR translation/80280
9594 * builtins.c (expand_builtin_object_size): Add missing quoting to
9595 %D and like directives.
9596 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9597 (hsa_type_for_tree_type): Same.
9598 (verify_function_arguments): Same.
9599 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9600 * varasm.c (get_section): Same.
9603 2017-05-04 Martin Sebor <msebor@redhat.com>
9605 PR translation/80280
9606 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9608 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9610 * config/aarch64/aarch64.c (generic_addrcost_table):
9611 Change HI/TI mode setting.
9613 2017-05-04 Martin Jambor <mjambor@suse.cz>
9615 PR tree-optimization/80622
9616 * tree-sra.c (comes_initialized_p): New function.
9617 (build_accesses_from_assign): Only set write lazily when
9618 comes_initialized_p is false.
9619 (analyze_access_subtree): Use comes_initialized_p.
9620 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9621 instead of testing for PARM_DECL.
9623 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9625 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9626 constraint on operand 0 to allow more general addressing modes.
9627 Adjust output template.
9628 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9630 * config/aarch64/aarch64-protos.h
9631 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9632 * config/aarch64/constraints.md (Dp): New address constraint.
9633 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9636 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9638 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9639 update use of estimate_ipcp_clone_size_and_time.
9640 (estimate_local_effects): Update use of
9641 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9642 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9643 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9644 Return nonspecialized time.
9646 2017-05-04 Richard Biener <rguenther@suse.de>
9648 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9649 for the last VUSE which def dominates the PHI. Directly call
9651 (get_continuation_for_phi_1): Remove.
9653 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9655 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9656 to explain the use of truncating division. Cap the number of
9657 iterations to the maximum given by nb_iterations_upper_bound,
9660 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9662 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9663 * configure: Regenerate.
9664 * config.in: Regenerate.
9665 * config/i386/driver-mingw32.c: new file.
9666 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9667 * config.host: Link driver-mingw32.o on MinGW host.
9668 * doc/install.texi: Document new --enable-mingw-wildcard configure
9671 2017-05-04 Marek Polacek <polacek@redhat.com>
9673 PR tree-optimization/80612
9674 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9676 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9677 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9679 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9680 (movt splitter): Likewise.
9681 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9682 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9683 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9684 block for Thumb-1 with MOVT.
9685 (thumb2_legitimate_address_p): Move code block ...
9686 (can_avoid_literal_pool_for_label_p): ... into this new function.
9687 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9689 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9690 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9691 "M-profile targets with the MOVT instruction".
9693 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9695 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9696 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9697 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9699 2017-05-04 Martin Liska <mliska@suse.cz>
9701 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9704 2017-05-04 Richard Biener <rguenther@suse.de>
9706 * tree.c (array_at_struct_end_p): Handle arrays at struct
9707 end with flexarrays more conservatively. Refactor and treat
9708 arrays of arrays or aggregates more strict. Fix
9709 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9710 * tree.c (array_at_struct_end_p): Adjust prototype.
9711 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9712 * gimple-fold.c (get_range_strlen): Likewise.
9713 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9715 2017-05-04 Richard Biener <rguenther@suse.de>
9717 PR tree-optimization/31130
9718 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9720 (supports_overflow_infinity): Likewise.
9721 (is_negative_overflow_infinity): Likewise.
9722 (is_positive_overflow_infinity): Likewise.
9723 (is_overflow_infinity): Likewise.
9724 (stmt_overflow_infinity): Likewise.
9725 (overflow_infinity_range_p): Likewise.
9726 (usable_range_p): Remove as always returning true.
9727 (make_overflow_infinity): Remove.
9728 (negative_overflow_infinity): Likewise.
9729 (positive_overflow_infinity): Likewise.
9730 (avoid_overflow_infinity): Likewise.
9731 (set_value_range): Adjust accordingly.
9732 (set_value_range_to_nonnegative): Likewise, remove now unused
9733 overflow_infinity arg.
9734 (vrp_operand_equal_p): Adjust.
9735 (update_value_range): Likewise.
9736 (range_int_cst_singleton_p): Likewise.
9737 (operand_less_p): Likewise.
9738 (compare_values_warnv): Likewise.
9739 (extract_range_for_var_from_comparison_expr): Likewise.
9740 (vrp_int_const_binop): Likewise.
9741 (zero_nonzero_bits_from_vr): Likewise.
9742 (extract_range_from_multiplicative_op_1): Likewise.
9743 (extract_range_from_binary_expr_1): Likewise.
9744 (extract_range_from_unary_expr): Likewise.
9745 (extract_range_from_comparison): Likewise.
9746 (extract_range_basic): Likewise.
9747 (adjust_range_with_scev): Likewise.
9748 (compare_ranges): Likewise.
9749 (compare_range_with_value): Likewise.
9750 (dump_value_range): Likewise.
9751 (test_for_singularity): Likewise, remove strict_overflow_p parameter
9753 (simplify_cond_using_ranges): Adjust.
9755 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
9757 * brig-builtins.def: Added a builtin for class_f64.
9758 * builtin-types.def: Added a builtin type needed by class_f64.
9760 2017-05-03 Jason Merrill <jason@redhat.com>
9762 * timevar.def: Add TV_CONSTEXPR.
9764 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9766 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
9768 2017-05-03 Martin Jambor <mjambor@suse.cz>
9770 * ipa-prop.c (ipa_update_after_lto_read): Removed.
9771 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
9772 * ipa-cp.c (ipcp_propagate_stage): Do not call
9773 ipa_update_after_lto_read.
9774 * ipa-inline.c (ipa_inline): Likewise.
9776 2017-05-03 Martin Jambor <mjambor@suse.cz>
9778 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
9779 tag. Added a default constructor and a destructor.
9780 (ipa_edge_args_sum_t): New class;
9781 (ipa_edge_args_sum): Declare.
9782 (ipa_edge_args_vector): Remove declaration.
9783 (IPA_EDGE_REF): Use ipa_edge_args_sum.
9784 (ipa_free_edge_args_substructures): Remove declaration.
9785 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
9786 (ipa_edge_args_info_available_for_edge_p): Likewise.
9787 * ipa-prop.c (ipa_edge_args_vector): Removed.
9788 (edge_removal_hook_holder): Likewise.
9789 (edge_duplication_hook_holder): Likewise.
9790 (ipa_edge_args_sum): New variable.
9791 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
9792 ipa_edge_args_vector.
9793 (ipa_free_edge_args_substructures): Likewise.
9794 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
9795 ipa_edge_args_vector.
9796 (ipa_edge_removal_hook): Turned into method
9797 ipa_edge_args_sum_t::remove.
9798 (ipa_edge_duplication_hook): Turned into method
9799 ipa_edge_args_sum_t::duplicate.
9800 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
9801 registering edge hooks.
9802 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
9803 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
9804 ipa_edge_args_sum instead of ipa_edge_args_vector.
9805 * ipa-profile.c (ipa_profile): Likewise.
9807 2017-05-03 Martin Jambor <mjambor@suse.cz>
9809 * symbol-summary.h (function_summary): New method exists.
9810 (function_summary::symtab_removal): Deallocate through release.
9811 (call_summary): New class.
9812 (gt_ggc_mx): New overload.
9813 (gt_pch_nx): Likewise.
9814 (gt_pch_nx): Likewise.
9816 2017-05-03 Jeff Law <law@redhat.com>
9818 PR tree-optimization/78496
9819 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
9820 from simplify_cond_using_ranges. Split off code to walk
9821 backwards through casts into ...
9822 (simplify_cond_using_ranges_2): New function.
9823 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
9824 (execute_vrp): After identifying jump threads, call
9825 simplify_cond_using_ranges_2.
9827 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
9830 * ipa-inline.h (inline_summary): Add ctor.
9831 (create_ggc): Do not use ggc_cleared_alloc.
9833 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
9834 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9836 * gcc.c (handle_braces): Support escaping in switch matching
9838 * doc/invoke.texi (Spec Files): Document it.
9839 Remove superfluous @code markup in items.
9841 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9843 * diagnostic-show-locus.c (struct column_range): New struct.
9844 (get_affected_columns): New function.
9845 (get_printed_columns): New function.
9846 (struct correction): New struct.
9847 (correction::ensure_capacity): New function.
9848 (correction::ensure_terminated): New function.
9849 (struct line_corrections): New struct.
9850 (line_corrections::~line_corrections): New dtor.
9851 (line_corrections::add_hint): New function.
9852 (layout::print_trailing_fixits): Reimplement in terms of the new
9854 (selftest::test_overlapped_fixit_printing): New function.
9855 (selftest::diagnostic_show_locus_c_tests): Call it.
9857 2017-05-03 Nathan Sidwell <nathan@acm.org>
9859 Canonicalize canonical type hashing
9860 * tree.h (type_hash_canon_hash): Declare.
9861 * tree.c (type_hash_list, attribute_hash_list): Move into
9862 type_hash_canon_hash.
9863 (build_type_attribute_qual_variant): Break out hash code calc into
9864 type_hash_canon_hash.
9865 (type_hash_canon_hash): New. Generic type hash computation.
9866 (build_range_type_1, build_array_type_1, build_function_type,
9867 build_method_type_directly, build_offset_type, build_complex_type,
9868 make_vector_type): Call it.
9870 2017-05-03 Richard Biener <rguenther@suse.de>
9872 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
9873 When all DRs have unknown misaligned do not always peel
9874 when there is a store but apply the same costing model as if
9875 there were only loads.
9877 2017-05-03 Richard Biener <rguenther@suse.de>
9880 PR tree-optimization/80492
9881 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9882 compare_base_decls returning dont-know properly.
9884 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
9886 * config/arm/iterators.md (CCSI): New mode iterator.
9887 (arch): New mode attribute.
9888 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
9889 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
9890 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
9891 code iterator for success result mode.
9892 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
9893 the corresponding new insn generators.
9895 2017-05-03 Bin Cheng <bin.cheng@arm.com>
9898 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9899 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9901 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
9903 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
9904 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
9905 (DDR_A): Wrap DDR argument in brackets.
9906 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
9907 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
9908 (DDR_REVERSED_P): Likewise.
9910 2017-05-03 Jakub Jelinek <jakub@redhat.com>
9912 PR tree-optimization/79472
9913 * tree-switch-conversion.c (struct switch_conv_info): Add
9914 contiguous_range and default_case_nonstandard fields.
9915 (collect_switch_conv_info): Compute contiguous_range and
9916 default_case_nonstandard fields, don't clear final_bb if
9917 contiguous_range and only the default case doesn't have the required
9919 (check_all_empty_except_final): Set default_case_nonstandard instead
9920 of failing if contiguous_range and the default case doesn't have empty
9922 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
9923 and only the default case doesn't have the required constants. Skip
9925 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
9926 if default_case_nonstandard.
9927 (build_constructors): Build constant 1 just once. Assert that default
9928 values aren't inserted in between cases if contiguous_range. Skip
9930 (build_arrays): Skip virtual phis.
9931 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
9932 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
9933 Handle virtual phis.
9934 (gen_inbound_check): Handle default_case_nonstandard case.
9935 (process_switch): Adjust check_final_bb caller. Call
9936 gather_default_values with the first non-default case instead of
9937 default case if default_case_nonstandard.
9939 2017-05-02 Nathan Sidwell <nathan@acm.org>
9941 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
9942 check. Fix formatting.
9944 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
9946 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
9947 errors when comparing specialized and unspecialized times.
9949 2017-05-02 David Malcolm <dmalcolm@redhat.com>
9951 * diagnostic-show-locus.c
9952 (layout::should_print_annotation_line_p): Make private.
9953 (layout::print_annotation_line): Make private.
9954 (layout::annotation_line_showed_range_p): Make private.
9955 (layout::show_ruler): Make private.
9956 (layout::print_source_line): Make private. Pass in line and
9957 line_width, rather than calling location_get_source_line. Drop
9959 (layout::print_leading_fixits): New method.
9960 (layout::print_any_fixits): Rename to...
9961 (layout::print_trailing_fixits): ...this, and make private.
9962 Don't print newline fixits.
9963 (diagnostic_show_locus): Move logic for printing one row into...
9964 (layout::print_line): ...this new function. Move the
9965 location_get_source_line call and error-handling from
9966 print_source_line to here. Call print_leading_fixits, and rename
9967 print_any_fixits to print_trailing_fixits.
9968 (selftest::test_fixit_insert_containing_newline): Update now that
9969 newlines are partially supported.
9970 (selftest::test_fixit_insert_containing_newline_2): New test.
9971 (selftest::test_fixit_replace_containing_newline): Update comments.
9972 (selftest::diagnostic_show_locus_c_tests): Call the new test.
9973 * edit-context.c (class added_line): New class.
9974 (class edited_line): Describe newline handling in comment.
9975 (edited_line::actually_edited_p): New method.
9976 (edited_line::print_content): Delete redundant decl.
9977 (edited_line::m_predecessors): New field.
9978 (edited_file::print_content): Call edited_line::print_content.
9979 (edited_file::print_diff): Update to support newlines.
9980 (edited_file::print_diff_hunk): Likewise.
9981 (edited_file::print_run_of_changed_lines): New function.
9982 (edited_file::print_diff_line): Convert to...
9983 (print_diff_line): ...this.
9984 (edited_file::get_effective_line_count): New function.
9985 (edited_line::edited_line): Initialize new field m_predecessors.
9986 (edited_line::~edited_line): Clean up m_predecessors.
9987 (edited_line::apply_fixit): Handle newlines.
9988 (edited_line::get_effective_line_count): New function.
9989 (edited_line::print_content): New function.
9990 (edited_line::print_diff_lines): New function.
9991 (selftest::test_applying_fixits_insert_containing_newline): New
9993 (selftest::test_applying_fixits_replace_containing_newline): New
9995 (selftest::insert_line): New function.
9996 (selftest::test_applying_fixits_multiple_lines): Add example of
9998 (selftest::edit_context_c_tests): Call the new tests.
10000 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10002 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10003 parameter cand. Update dump information.
10004 (get_computation_cost): Update uses.
10006 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10008 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10009 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10010 (get_computation_at, rewrite_use_address): Update use of
10011 get_computation_aff.
10013 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10015 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10016 (get_computation): Delete.
10017 (get_computation_cost): Implement like get_computation_cost_at.
10018 Use get_computation_at.
10019 (get_computation_cost_at): Delete.
10020 (rewrite_use_nonlinear_expr): Use get_computation_at.
10021 (rewrite_use_compare, remove_unused_ivs): Ditto.
10023 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10025 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10027 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10029 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10030 (ivopts_global_cost_for_size): Rename parameter and update uses.
10031 (iv_ca_recount_cost): Update uses.
10032 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10033 candidates seperately in n_invs and n_cands.
10034 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10036 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10038 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10039 (find_inv_vars_cb): New.
10040 (find_depends): Renamed to ...
10041 (find_inv_vars): ... this.
10042 (add_candidate_1, force_var_cost): Call find_inv_vars.
10043 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10045 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10047 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10048 inv_vars. Add inv_exprs.
10049 (struct iv_cand): Rename depends_on to inv_vars.
10050 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10051 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10052 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10053 (dump_cand): Dump inv_vars.
10054 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10055 (record_invariant, find_depends, add_candidate_1): Ditto.
10056 (set_group_iv_cost, force_var_cost): Ditto.
10057 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10058 (get_computation_cost_at, get_computation_cost): Ditto.
10059 (determine_group_iv_cost_generic): Ditto.
10060 (determine_group_iv_cost_address): Ditto.
10061 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10062 (determine_group_iv_costs): Ditto.
10063 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10064 (iv_ca_set_remove_invariants): Renamed to ...
10065 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10066 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10067 (iv_ca_set_add_invariants): Renamed to ...
10068 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10069 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10070 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10071 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10072 (create_new_ivs): Remove useless dump.
10074 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10076 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10078 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10079 (iv_ca_set_no_cp, create_new_iv): Ditto.
10081 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10083 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10085 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10087 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10088 function tree_check2.
10090 2017-05-02 Martin Liska <mliska@suse.cz>
10092 * doc/gcov.texi: Add missing preposition.
10093 * gcov.c (function_info::function_info): Properly fill up
10094 all member variables.
10096 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10098 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10100 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10102 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10104 2017-05-02 Martin Liska <mliska@suse.cz>
10107 * lto-streamer-in.c (lto_read_tree_1): Remove
10108 LTO_STREAMER_DEBUG.
10109 * lto-streamer.c (struct tree_hash_entry): Likewise.
10110 (struct tree_entry_hasher): Likewise.
10111 (tree_entry_hasher::hash): Likewise.
10112 (tree_entry_hasher::equal): Likewise.
10113 (lto_streamer_init): Likewise.
10114 (lto_orig_address_map): Likewise.
10115 (lto_orig_address_get): Likewise.
10116 (lto_orig_address_remove): Likewise.
10117 * lto-streamer.h: Likewise.
10118 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10119 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10121 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10123 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10124 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10125 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10126 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10127 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10128 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10129 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10130 (mm_maskz_sub_ss): New intrinsics.
10131 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10132 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10133 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10134 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10135 (__builtin_ia32_subss_mask_round): New builtins.
10136 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10137 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10138 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10140 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10141 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10143 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10146 2017-05-02 Martin Jambor <mjambor@suse.cz>
10148 PR tree-optimization/78687
10149 * tree-sra.c (access): New field parent.
10150 (process_subtree_disqualification): New function.
10151 (disqualify_candidate): Call it.
10152 (build_accesses_from_assign): Reset write flag if creating an
10154 (build_access_subtree): Fill in parent field and also prpagate
10155 down grp_write flag.
10156 (create_artificial_child_access): New parameter set_grp_write, set
10157 grp_write to its value.
10158 (propagate_subaccesses_across_link): Also propagate grp_write flag
10160 (propagate_all_subaccesses): Push the closest parent back to work
10161 queue if add_access_to_work_queue returned true.
10163 2017-05-02 Richard Biener <rguenther@suse.de>
10165 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10166 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10167 -fstrict-overflow documentation.
10168 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10169 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10170 flag_strict_overflow.
10171 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10172 * lto-opts.c (lto_write_options): Do not stream it.
10173 * lto-wrapper.c (merge_and_complain): Do not handle it.
10174 * opts.c (default_options_table): Do not set -fstrict-overflow.
10175 (finish_options): Likewise do not clear it when sanitizing.
10176 * simplify-rtx.c (simplify_const_relational_operation): Do not
10177 test flag_strict_overflow.
10179 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10181 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10182 using enabled attribute.
10183 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10184 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10185 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10186 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10187 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10188 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10189 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10190 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10191 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10192 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10194 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10196 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10198 2017-05-02 Richard Biener <rguenther@suse.de>
10200 PR tree-optimization/80591
10202 2017-04-10 Richard Biener <rguenther@suse.de>
10204 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10207 2017-05-02 Richard Biener <rguenther@suse.de>
10209 PR tree-optimization/80549
10210 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10211 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10212 headers if they do not have a preheader.
10214 2017-05-02 Martin Liska <mliska@suse.cz>
10217 * common.opt: Fix typo.
10218 * doc/invoke.texi: Likewise.
10220 2017-05-01 Jan Beulich <jbeulich@suse.com>
10222 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10223 swapping, add (x,x,m,x,n) alternative.
10225 2017-05-01 Nathan Sidwell <nathan@acm.org>
10227 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10228 unnecessary unadjusted_alignment check.
10230 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10233 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10235 * gimplify.c (gimplify_cilk_detach): New function.
10236 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10237 * tree-core.h: Document EXPR_CILK_SPAWN.
10238 * tree.h (EXPR_CILK_SPAWN): Define.
10240 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10242 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10243 to use new fixit_hint representation, using the "replace" logic.
10244 (get_line_span_for_fixit_hint): Likewise.
10245 (layout::print_any_fixits): Likewise.
10246 (selftest::test_one_liner_many_fixits): Rename to...
10247 (selftest::test_one_liner_many_fixits_1): ...this, and update
10248 comment and expected output to reflect that the multiple fix-it
10249 hints are now consolidated into one insertion.
10250 (selftest::test_one_liner_many_fixits_2): New test.
10251 (selftest::test_diagnostic_show_locus_one_liner): Update for
10253 (selftest::test_fixit_consolidation): Update for fix-it API
10255 * diagnostic.c (print_parseable_fixits): Likewise.
10256 * edit-context.c (edited_line::m_line_events): Convert from
10257 auto_vec <line_event *> to auto_vec <line_event>.
10258 (class line_event): Convert from abstract base class to a concrete
10259 class, taking over the role of replace_event.
10260 (class insert_event): Delete.
10261 (class replace_event): Rename to class line_event. Convert to
10263 (edit_context::add_fixits): Reimplement.
10264 (edit_context::apply_insert): Delete.
10265 (edit_context::apply_replace): Rename to...
10266 (edit_context::apply_fixit): ...this. Convert to half-open range.
10267 (edited_file::apply_insert): Delete.
10268 (edited_file::apply_replace): Rename to...
10269 (edited_file::apply_fixit): ...this.
10270 (edited_line::~edited_line): Drop deletion of events.
10271 (edited_line::apply_insert): Delete.
10272 (edited_line::apply_replace): Rename to...
10273 (edited_line::apply_fixit): ...this. Convert to half-open range.
10274 Update for change to type of m_line_events.
10275 * edit-context.h (edit_context::apply_insert): Delete.
10276 (edit_context::apply_replace): Rename to...
10277 (edit_context::apply_fixit): ...this.
10279 2017-05-01 Martin Sebor <msebor@redhat.com>
10281 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10284 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10287 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10288 __get_cpuid_max returns 0.
10289 (__get_cpuid_count): Ditto.
10291 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10293 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10294 replacement expression is another instance of one of its arguments.
10296 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10299 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10300 check for stack push/pop autoinc.
10301 * config/i386/i386.c (ix86_agi_dependent): Return false
10302 if the only reason why modified_in_p returned true is that
10303 addr is SP based and set_insn is a push or pop.
10305 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10307 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10310 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10313 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10314 (account_size_time): Use two predicates - exec_pred and
10316 (evaluate_conditions_for_known_args): Compute both clause and
10318 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10319 (inline_summary_t::duplicate): Update.
10320 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10322 (compute_inline_parameters): Likewise.
10323 (estimate_edge_size_and_time): Update caluclation of time.
10324 (estimate_node_size_and_time): Compute both time and nonspecialized
10326 (estimate_ipcp_clone_size_and_time): Update.
10327 (inline_merge_summary): Update.
10328 (do_estimate_edge_time): Update.
10329 (do_estimate_edge_size): Update.
10330 (do_estimate_edge_hints): Update.
10331 (inline_read_section, inline_write_summary): Stream both new predicates.
10332 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10334 (compute_inlined_call_time): Cleanup.
10335 (big_speedup_p): Update.
10336 (edge_badness): Update.
10337 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10338 (size_time_entry): Replace predicate by exec_predicate and
10339 nonconst_predicate.
10340 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10341 (estimate_edge_time): Return also nonspec_time.
10342 (reset_edge_growth_cache): Update.
10344 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10346 PR rtl-optimization/80491
10347 * ifcvt.c (noce_process_if_block): When looking for x setter
10348 with missing else_bb, don't check only the insn right before
10349 cond_earliest, but look for the last insn that x is modified in
10350 within the same bb.
10352 PR rtl-optimization/80491
10353 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10355 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10357 PR tree-optimization/80487
10358 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10360 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10362 PR tree-optimization/79697
10363 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10364 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10365 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10367 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10368 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10370 2017-04-28 Martin Sebor <msebor@redhat.com>
10372 PR tree-optimization/80523
10373 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10374 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10376 (maybe_warn, format_directive, parse_directive): Use new functions.
10377 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10379 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10381 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10383 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10385 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10386 target_header_dir): Set correctly.
10387 * configure: Regenerated.
10388 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10389 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10390 instead of SYSTEM_HEADER_DIR.
10392 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10394 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10395 (estimate_local_effects): Likewise.
10396 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10397 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10398 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10399 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10400 do_estimate_edge_time, estimate_edge_time): Likewise.
10401 * ipa-inline-analysis.c (estimate_node_size_and_time,
10402 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10403 (estimate_time_after_inlining): Remove.
10405 2017-04-28 Martin Liska <mliska@suse.cz>
10407 * doc/gcov.texi: Enhance documentation of gcov.
10409 2017-04-28 Martin Liska <mliska@suse.cz>
10411 * doc/gcov.texi: Sort options in alphabetic order.
10412 * doc/gcov-dump.texi: Likewise.
10413 * doc/gcov-tool.texi: Likewise.
10414 * gcov.c (print_usage): Likewise.
10415 * gcov-dump.c (print_usage): Likewise.
10416 * gcov-tool.c (print_merge_usage_message): Likewise.
10417 (print_rewrite_usage_message): Likewise.
10418 (print_overlap_usage_message): Likewise.
10420 2017-04-28 Martin Liska <mliska@suse.cz>
10422 PR gcov-profile/53915
10423 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10425 2017-04-28 Martin Liska <mliska@suse.cz>
10427 PR gcov-profile/79891
10428 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10429 is marked by compiler as living on a line.
10430 (get_cycles_count): Remove usage of the union.
10431 (output_intermediate_file): Likewise.
10432 (find_source): Fix GNU coding style.
10433 (accumulate_line_counts): Remove old non-all block mode.
10434 (output_lines): Remove usage of the union.
10435 * profile.c (output_location): Include all BBs, even if
10436 belonging to a same line (and file) as a previous BB.
10438 2017-04-28 Martin Liska <mliska@suse.cz>
10440 * gcov.c (process_args): Handle new argument 'w'.
10441 (read_graph_file): Assign ID to BBs.
10442 (output_branch_count): Display BB # if verbose flag is set.
10443 (output_lines): Likewise for arcs.
10444 (print_usage): Add '--verbose' option help.
10445 * doc/gcov.texi: Document --verbose (-w) option.
10447 2017-04-28 Martin Liska <mliska@suse.cz>
10449 * gcov.c (struct block_location_info): New struct.
10450 (process_file): Fill up the new structure.
10451 (read_graph_file): Replace usage of encoding by the newly added
10453 (add_line_counts): Likewise.
10454 (accumulate_line_counts): Remove usage of the union.
10455 (function_info::function_info): New function.
10456 (function_info::~function_info): Likewise.
10457 (process_file): Call delete instead of release_function.
10458 (release_function): Release the function.
10459 (release_structures): Call delete instead of release_function.
10460 (solve_flow_graph): Replace usage of num_blocks.
10461 (find_exception_blocks): Likewise.
10462 (output_lines): Fix GNU coding style.
10464 2017-04-28 Martin Liska <mliska@suse.cz>
10467 * coverage.c (coverage_remove_note_file): New function.
10468 * coverage.h: Declare the function.
10469 * toplev.c (finalize): Clean if an error has been seen.
10471 2017-04-28 Martin Liska <mliska@suse.cz>
10473 PR gcov-profile/80031
10474 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10475 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10476 * gcov.c (read_graph_file): Read just number of blocks.
10477 * profile.c (branch_prob): Do not stream 0 flags per a basic
10480 2017-04-28 Martin Liska <mliska@suse.cz>
10482 * gcov-dump.c (tag_*): Add new argument to declarations.
10483 (dump_gcov_file): Likewise.
10484 (tag_blocks): Add and use new argument depth.
10485 (tag_arcs): Likewise.
10486 (tag_lines): Likewise.
10487 (tag_counters): Likewise.
10488 (tag_summary): Likewise.
10489 (dump_working_sets): Use depth to do a proper indentation.
10491 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10494 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10495 * symtab.c (symtab_node::debug_symtab): Move definition here.
10497 2017-04-28 Richard Biener <rguenther@suse.de>
10499 * lto-streamer.h (LTO_major_version): Bump to 7.
10501 2017-04-28 Richard Biener <rguenther@suse.de>
10503 * tree-vrp.c (assert_info): New struct.
10504 (add_assert_info): New helper.
10505 (register_edge_assert_for_2): Refactor to add asserts to a vector
10507 (register_edge_assert_for_1): Likewise.
10508 (register_edge_assert_for): Likewise.
10509 (finish_register_edge_assert_for): New helper actually registering
10510 asserts where live on edge.
10511 (find_conditional_asserts): Adjust.
10512 (find_switch_asserts): Likewise.
10513 (evrp_dom_walker::try_find_new_range): Generalize.
10514 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10516 2017-04-27 Marek Polacek <polacek@redhat.com>
10519 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10520 arg10 and arg11 to itype.
10522 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10524 * doc/extend.texi (Object Size Checking): Improve grammar.
10526 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10529 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10530 that the logic for permitting reciprocal estimates matches that
10533 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10536 * tree.c (type_cache_hasher::equal): Only compare
10537 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10538 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10539 non-aggregate element types.
10540 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10541 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10543 2017-04-27 Richard Biener <rguenther@suse.de>
10545 PR middle-end/80533
10546 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10547 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10548 keeping a reference to a trailing array.
10550 2017-04-27 Richard Biener <rguenther@suse.de>
10552 PR middle-end/80539
10553 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10554 being in loop-closed SSA form conservatively.
10555 (chrec_fold_multiply_poly_poly): Likewise.
10557 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10559 PR middle-end/79665
10560 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10561 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10563 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10566 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10567 (aarch64_function_arg_alignment): Return unsigned int again, but still
10568 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10569 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10570 Don't emit -Wpsabi note.
10571 (aarch64_function_arg_boundary): Likewise.
10572 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10575 2017-04-26 Nathan Sidwell <nathan@acm.org>
10577 * tree.h (crc32_unsigned_n): Declare.
10578 (crc32_unsigned, crc32_unsigned): Make inline.
10579 * tree.c (crc32_unsigned_bits): Replace with ...
10580 (crc32_unsigned_n): ... this.
10581 (crc32_unsigned, crc32_byte): Remove.
10582 (crc32_string): Remove unnecessary braces.
10584 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10586 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10587 * ipa-inline-analysis.c (MAX_TIME): Remove.
10588 (account_size_time): Use sreal for time.
10589 (dump_inline_summary): Update.
10590 (estimate_function_body_sizes): Update.
10591 (estimate_edge_size_and_time): Update.
10592 (estimate_calls_size_and_time): Update.
10593 (estimate_node_size_and_time): Update.
10594 (inline_merge_summary): Update.
10595 (inline_update_overall_summary): Update.
10596 (estimate_time_after_inlining): Update.
10597 (inline_read_section): Update.
10598 (inline_write_summary): Update.
10599 * ipa-inline.c (compute_uninlined_call_time): Update.
10600 (compute_inlined_call_time): Update.
10601 (recursive_inlining): Update.
10602 (inline_small_functions): Update.
10603 (dump_overall_stats): Update.
10604 * ipa-inline.h: Include sreal.h.
10605 (size_time_entry): Turn time to sreal.
10606 (inline_summary): Turn self_time nad time to sreal.
10608 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10610 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10612 (sreal::stream_out, sreal::stream_in): New.
10613 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10615 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10617 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10620 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10623 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10625 (dimode_scalar_chain::compute_convert_gain): Ditto.
10626 (dimode_scalar_chain::make_vector_copies): Ditto.
10627 (dimode_scalar_chain::convert_reg): Ditto.
10628 (dimode_scalar_chain::convert_insn): Ditto.
10629 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10630 (VI248_AVX512BW_1): New mode iterator.
10631 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10632 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10635 2017-04-25 Martin Sebor <msebor@redhat.com>
10637 PR tree-optimization/80497
10638 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10639 constants are representable in HOST_WIDE_INT.
10640 (parse_directive): Ditto.
10642 2017-04-25 Martin Sebor <msebor@redhat.com>
10645 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10646 (new_zero_array): Adjust signature.
10647 (dom_info::dom_init): Used unsigned rather that size_t.
10648 (dom_info::dom_info): Same.
10650 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10651 Jakub Jelinek <jakub@redhat.com>
10654 * config/arm/arm.c: Include gimple.h.
10655 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10656 returns negative, increment ncrn only if it returned positive.
10657 (arm_needs_doubleword_align): Return int instead of bool,
10658 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10659 members, but if there is any such non-FIELD_DECL
10660 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10661 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10662 returns negative, increment nregs only if it returned positive.
10663 (arm_setup_incoming_varargs): Likewise.
10664 (arm_function_arg_boundary): Emit -Wpsabi note if
10665 arm_needs_doubleword_align returns negative, return
10666 DOUBLEWORD_ALIGNMENT only if it returned positive.
10668 2017-04-25 Marek Polacek <polacek@redhat.com>
10671 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10672 first argument to type.
10674 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10677 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10678 type checks to test for compatibility instead of equality.
10680 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10681 Jakub Jelinek <jakub@redhat.com>
10684 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10686 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10687 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10688 the alignment computation, but return their maximum in warn_alignment.
10689 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10690 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10692 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10693 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10696 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10698 * config/arc/simdext.md (dmpyh): Fix typo.
10700 2017-04-25 Richard Biener <rguenther@suse.de>
10702 PR tree-optimization/80492
10703 * alias.c (compare_base_decls): Handle registers with asm
10704 specification conservatively.
10705 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10706 compare_base_decls returning dont-know properly.
10708 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10710 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10711 (legitimate_offset_address_p): New function.
10712 (arc_legitimate_address_p): Use above function.
10714 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10716 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10718 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10720 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
10721 ACCH registers whenever they are available.
10723 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10725 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
10726 double regs fix when not used.
10728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10730 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
10732 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
10733 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
10735 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10737 * config/arc/arc.c (arc_output_addsi): Check for h-register class
10738 when emitting short ADD instructions.
10740 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10742 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
10744 (cmpsi_cc_c_insn): Likewise.
10745 (cbranchsi4_scratch): Compute proper instruction length using
10746 compact_hreg_operand.
10747 * config/arc/predicates.md (compact_hreg_operand): New predicate.
10749 2017-04-25 Richard Biener <rguenther@suse.de>
10751 PR middle-end/80509
10752 * passes.c (pass_manager::pass_manager): Initialize
10753 m_name_to_pass_map.
10755 2017-04-25 Richard Biener <rguenther@suse.de>
10757 PR tree-optimization/79201
10758 * tree-ssa-sink.c (statement_sink_location): Handle calls.
10760 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10763 * config/s390/vector.md: Split MEM->GPR vector moves for
10764 non-s_operand addresses.
10766 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10769 * config/s390/predicates.md (reload_const_wide_int_operand): New
10771 * config/s390/s390.md ("movti"): Remove d/P alternative.
10772 ("movti_bigconst"): New pattern definition.
10774 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10777 * s390-protos.h (s390_expand_cs_hqi): Removed.
10778 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
10779 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
10780 modes as well as CCZ1mode and CCZmode.
10781 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
10782 signature of s390_emit_compare_and_swap.
10783 (s390_expand_cs_hqi): Likewise, make static.
10784 (s390_expand_cs_tdsi): Generate an explicit compare before trying
10785 compare-and-swap, in some cases.
10786 (s390_expand_cs): Wrapper function.
10787 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
10789 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
10790 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
10791 patterns for small and large integers. Forbid symref memory operands.
10792 Move expander to s390.c. Require cc register.
10793 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
10794 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
10795 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
10796 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
10797 symref memory operands. Remove CC mode and call s390_match_ccmode
10799 ("atomic_exchange<mode>"): Allow and implement all integer modes.
10801 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10803 * config/s390/s390.md (define_peephole2): New peephole to help
10804 combining the load-and-test pattern with volatile memory.
10806 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10808 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
10809 with CCZmode for TARGET_Z196.
10811 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10813 PR rtl-optimization/80501
10814 * combine.c (make_compound_operation_int): Set subreg_code to SET
10815 even for AND with mask of the sign bit of mode.
10817 PR rtl-optimization/80500
10818 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
10819 sum's initial value.
10821 2017-04-25 Julian Brown <julian@codesourcery.com>
10822 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10824 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
10826 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
10828 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
10830 2017-04-25 Julian Brown <julian@codesourcery.com>
10831 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10833 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
10834 (thunderx2t99_sha): New Reservation.
10836 2017-04-25 Julian Brown <julian@codesourcery.com>
10837 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10839 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
10840 type for 1-element load.
10842 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
10844 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
10846 2017-04-24 Martin Jambor <mjambor@suse.cz>
10848 PR tree-optimization/80293
10849 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
10850 char arrays not totally scalarizable if it is false.
10851 (analyze_all_variable_accesses): Pass correct value in the new
10852 parameter. Add a statistics counter.
10854 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
10856 PR middle-end/79931
10857 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
10859 2017-04-24 Richard Biener <rguenther@suse.de>
10861 PR tree-optimization/80494
10862 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
10863 out for complex types.
10865 2017-04-24 Richard Biener <rguenther@suse.de>
10867 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
10868 * tree-ssa-sccvn.c (print_scc): Print SCC size.
10869 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
10870 (DFS): Adjust and never fail.
10871 (sccvn_dom_walker::fail): Remove.
10872 (sccvn_dom_walker::before_dom_children): Adjust.
10873 (run_scc_vn): Likewise and never fail.
10874 * tree-ssa-pre.c (pass_pre::execute): Adjust.
10875 (pass_fre::execute): Likewise.
10877 2017-04-24 Richard Biener <rguenther@suse.de>
10879 PR tree-optimization/79725
10880 * tree-ssa-sink.c (statement_sink_location): Return whether
10881 failure reason was zero uses. Move that check later.
10882 (sink_code_in_bb): Deal with zero uses by removing the stmt
10885 2017-04-24 Richard Biener <rguenther@suse.de>
10888 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
10889 pointer-based references.
10891 2017-04-24 Richard Biener <rguenther@suse.de>
10894 * pass_manager.h (pass_manager::operator new): Remove.
10895 (pass_manager::operator delete): Likewise.
10896 * passes.c (pass_manager::operator new): Remove.
10897 (pass_manager::operator delete): Likewise.
10898 (pass_manager::pass_manager): Zero individual pass members.
10900 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
10903 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
10904 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
10905 Check "XEXP (src, 1)" operand here.
10906 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
10907 Check "XEXP (src, 1)" operand here.
10908 (dimode_scalar_chain::make_vector_copies): Detect count register
10909 of a shift instruction. Zero extend count register from QImode
10910 to DImode to satisfy vector shift pattern count operand predicate.
10911 Substitute vector shift count operand with a DImode copy.
10912 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
10915 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
10917 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
10918 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10919 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
10920 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10921 (UNSPEC_NOREX_MEM): Remove definition.
10923 2017-04-21 Richard Biener <rguenther@suse.de>
10925 PR tree-optimization/79547
10926 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10927 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
10928 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
10929 without any constraints.
10931 2017-04-21 Richard Biener <rguenther@suse.de>
10933 PR tree-optimization/78847
10934 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
10936 2017-04-21 Richard Biener <rguenther@suse.de>
10938 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
10939 (build_distinct_type_copy): Likewise.
10940 (build_variant_type_copy): Likewise.
10941 * tree.c (build_qualified_type): Pass down mem-stat info.
10942 (build_distinct_type_copy): Likewise.
10943 (build_variant_type_copy): Likewise.
10945 2017-04-21 Richard Biener <rguenther@suse.de>
10947 PR tree-optimization/80237
10948 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
10950 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
10951 for a simplified result.
10953 2016-04-21 Richard Biener <rguenther@suse.de>
10955 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
10956 sth as strict as a simple_iv but a chrec without symbols and an
10957 operand defined in the loop we are peeling (and not some subloop).
10958 (propagate_constants_for_unrolling): Propagate all constants.
10960 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10963 * config/i386/i386.c (print_reg): Remove assert for disalowed
10964 regno values, call output_operand_lossage instead.
10966 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10969 * config/i386/constraints.md (Yc): New register constraint.
10970 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
10971 Use Yc constraint for alternative 2 of operand 0. Remove
10972 preferred_for_speed attribute.
10974 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10976 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
10977 lastprivate clauses in SIMT case.
10979 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
10981 * doc/invoke.texi (-Wextra-semi): Document new warning option.
10983 2017-04-20 Richard Biener <rguenther@suse.de>
10985 PR tree-optimization/57796
10986 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
10987 as N scalar stores.
10988 (vect_model_load_cost): Cost gathers as N scalar loads.
10990 2017-04-20 Richard Biener <rguenther@suse.de>
10992 * ggc-page.c (ggc_allocated_p): Rename to ...
10993 (safe_lookup_page_table_entry): ... this and return the lookup
10995 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
10997 2017-04-20 Richard Biener <rguenther@suse.de>
10999 PR tree-optimization/80453
11000 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11001 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11002 from the conditions.
11003 (vn_phi_eq): Pass them down.
11004 (vn_phi_lookup): Record them.
11005 (vn_phi_insert): Likewise.
11007 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11009 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11010 uninitialized variable warning to avoid buffer overrun.
11012 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11015 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11016 is suppressed for '{ 0 }' in C.
11018 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11020 * BASE-VER: Set to 8.0.0.
11022 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11024 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11025 priority .init_array and .fini_array section with SECTION_NOTYPE
11028 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11030 PR middle-end/80423
11031 * tree.h (build_array_type): Add typeless_storage default argument.
11032 * tree.c (type_cache_hasher::equal): Also compare
11033 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11034 (build_array_type): Add typeless_storage argument, set
11035 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11037 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11038 (build_array_type): Likewise. Add typeless_storage argument.
11040 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11041 Jakub Jelinek <jakub@redhat.com>
11043 PR tree-optimization/80426
11044 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11045 operation on symbolic operands, also compute the overflow for the
11046 invariant part when the operation degenerates into a negation.
11048 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11051 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11052 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11055 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11057 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11060 * config/avr/avr.c (tree.h): Include it.
11061 (cgraph.h): Include it.
11062 (avr_encode_section_info): Don't warn for uninitialized progmem
11063 variable if it's just an alias.
11065 2017-04-19 Richard Biener <rguenther@suse.de>
11068 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11069 when needed by AutoPGO.
11071 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11074 * doc/lto.texi: Remove an extra 'that'.
11076 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11078 PR rtl-optimization/80429
11079 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11080 are only used in debug insns.
11082 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11083 Vladimir Makarov <vmakarov@redhat.com>
11085 * config/sparc/predicates.md (input_operand): Add comment. Return
11086 true for any memory operand when LRA is in progress.
11087 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11089 2017-04-18 Jeff Law <law@redhat.com>
11092 * mips.md ({return,simple_return}_internal): Do not overwrite
11095 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11097 PR tree-optimization/80443
11098 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11099 instead of adding 1, subtract -1 and similarly instead of subtracting
11102 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11104 PR rtl-optimization/80357
11105 * haifa-sched.c (tmp_bitmap): New variable.
11106 (model_recompute): Handle duplicate use records.
11107 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11108 (free_global_sched_pressure_data): Free it.
11110 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11113 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11114 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11115 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11116 instead of SYSTEM_HEADER_DIR.
11118 2017-04-18 Jeff Law <law@redhat.com>
11120 PR middle-end/80422
11121 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11122 predecessors after walking up the insn chain.
11124 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11127 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11128 sizetype type into debug info.
11130 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11133 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11134 unneeded test for TARGET_UPPER_REGS_SF.
11135 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11137 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11140 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11141 instead of gsi_after_labels.
11143 2017-04-18 Jeff Law <law@redhat.com>
11145 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11149 2017-04-13 Jeff Law <law@redhat.com>
11150 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11151 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11153 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11156 * config/avr/avr.c (intl.h): Include it.
11157 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11159 2017-04-18 Martin Liska <mliska@suse.cz>
11161 PR gcov-profile/78783
11162 * gcov-tool.c (gcov_output_files): Validate that destination
11163 file is either removed by the tool or by a user.
11165 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11166 Guy Benyei <guybe@mellanox.com>
11168 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11169 block, and do not negate it, the stored id is already negative.
11171 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11173 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11175 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11178 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11179 masks of options that should be turned off if the VSX vector
11180 options are turned off.
11181 (OTHER_P8_VECTOR_MASKS): Likewise.
11182 (OTHER_VSX_VECTOR_MASKS): Likewise.
11183 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11184 rs6000_disable_incompatible_switches to validate no type switches
11186 (rs6000_incompatible_switch): New function to disallow turning on
11187 other vector options if -mno-vsx, -mno-power8-vector, or
11188 -mno-power9-vector are specified.
11190 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11192 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11194 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11196 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11197 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11198 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11199 (ARG_POINTER_CFA_OFFSET): Likewise.
11201 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11203 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11204 conditions to take advantage of various optimizations.
11206 2017-04-13 Jeff Law <law@redhat.com>
11208 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11209 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11210 (zero_extendsidi2_dext): Likewise.
11212 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11215 * fold-const.c (fold_ternary_loc): Revert
11216 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11219 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11221 PR rtl-optimization/80343
11222 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11223 new scratch pseudo.
11225 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11228 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11229 to ubsan_encode_value.
11231 2017-04-13 Jeff Law <law@redhat.com>
11233 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11234 appearing in DEBUG_INSNs.
11236 2017-04-13 Martin Liska <mliska@suse.cz>
11238 PR gcov-profile/80413
11239 * gcov-io.c (gcov_write_string): Copy to buffer just when
11240 allocated size is greater than zero.
11242 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11245 * dwarf2out.c (decls_for_scope): Ignore declarations of
11246 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11248 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11251 * ipa-visibility.c (non_local_p): Fix typos.
11252 (localize_node): When localizing symbol in same comdat group,
11253 dissolve the group only when we know external symbols are going
11255 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11257 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11259 PR tree-optimization/79390
11260 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11261 order does not result in usable sequence, retry with reversed operand
11267 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11268 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11269 op0 instead of fold_convert_loc (loc, type, arg0).
11271 2017-04-12 Jeff Law <law@redhat.com>
11273 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11274 has a delay slot in the generated code.
11276 * config/cris/cris.md (cris_preferred_reload_class): Return
11277 GENNONACR_REGS rather than GENERAL_REGS.
11279 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11282 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11283 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11284 signedness of the result type.
11286 2017-04-12 Richard Biener <rguenther@suse.de>
11287 Jeff Law <law@redhat.com>
11289 PR tree-optimization/80359
11290 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11291 trim stores to TARGET_MEM_REFs.
11293 2017-04-12 Richard Biener <rguenther@suse.de>
11295 PR tree-optimization/79390
11296 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11297 threading case even more.
11299 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11302 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11303 for quad_address_p for TImode, instead of just not indexed_address.
11305 2017-04-12 Richard Biener <rguenther@suse.de>
11306 Bernd Edlinger <bernd.edlinger@hotmail.de>
11308 PR middle-end/79671
11309 * alias.c (component_uses_parent_alias_set_from): Handle
11310 TYPE_TYPELESS_STORAGE.
11311 (get_alias_set): Likewise.
11312 * tree-core.h (tree_type_common): Add typeless_storage flag.
11313 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11314 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11315 for types containing members with TYPE_TYPELESS_STORAGE.
11316 (place_field): Likewise.
11317 (layout_type): Likewise for ARRAY_TYPE.
11318 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11319 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11320 TYPE_TYPELESS_STORAGE.
11321 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11323 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11326 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11327 first argument to type.
11329 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11333 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11334 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11335 (rs6000_expand_binop_builtin): Likewise.
11336 (rs6000_expand_ternop_builtin): Likewise; also add missing
11337 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11338 vshasigma built-ins.
11339 * doc/extend.texi: Document that vec_xxpermdi's third argument
11340 must be a constant.
11342 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11344 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11345 Use shift_const cost parameter when calculating gain of STV shifts.
11347 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11349 PR rtl-optimization/70478
11350 * lra-constraints.c (process_alt_operands): Check memory for
11351 disfavoring memory insn operand.
11353 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11355 PR middle-end/80100
11356 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11357 left shift in unsigned HOST_WIDE_INT type.
11359 PR rtl-optimization/80385
11360 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11361 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11364 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11365 if they have any depend clauses.
11367 2017-04-11 Martin Liska <mliska@suse.cz>
11370 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11371 * ipa-split.c (split_function): Create a local comdat symbol
11372 if caller is in a comdat group.
11374 2017-04-11 Martin Liska <mliska@suse.cz>
11377 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11380 2017-04-11 Martin Sebor <msebor@redhat.com>
11382 PR middle-end/80364
11383 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11384 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11386 (directive::set_width, directive::set_precision, format_character):
11388 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11391 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11394 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11395 conflict, set target->arch_name instead of target->cpu_name.
11397 2017-04-11 Richard Biener <rguenther@suse.de>
11399 PR tree-optimization/80374
11400 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11401 build_zero_cst, remove fold_convertible_p check again.
11403 2017-04-11 Martin Liska <mliska@suse.cz>
11406 * ubsan.c (instrument_object_size): Do not instrument register
11409 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11412 * config/i386/i386-builtin-types.def
11413 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11414 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11415 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11416 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11417 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11418 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11419 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11420 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11421 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11422 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11423 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11424 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11425 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11426 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11427 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11428 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11429 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11430 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11431 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11432 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11433 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11434 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11435 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11436 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11437 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11438 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11439 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11440 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11441 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11442 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11443 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11444 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11445 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11446 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11447 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11448 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11449 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11450 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11451 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11452 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11453 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11454 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11455 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11456 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11457 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11458 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11460 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11461 flag to second_arg_count, handle 4 argument function type _COUNT
11462 aliases, handle second_arg_count on second argument rather than last.
11464 2017-04-10 Jeff Law <law@redhat.com>
11466 PR tree-optimization/80374
11467 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11468 record anything if we can not convert integer_zero_node to the
11471 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11474 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11475 Enhance special handling given to the TARGET_P9_MINMAX option in
11476 relation to certain other options.
11478 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11480 PR tree-optimization/80153
11481 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11482 remove POINTER_PLUS_EXPR's base part directly, rather than through
11485 2017-04-10 Richard Biener <rguenther@suse.de>
11486 Bin Cheng <bin.cheng@arm.com>
11488 PR tree-optimization/80153
11489 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11490 the first element of pointer type aff_tree. Build result expr in
11492 (add_elt_to_tree): Convert to type unconditionally. Remove other
11493 fold_convert calls.
11494 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11495 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11497 2017-04-10 Richard Biener <rguenther@suse.de>
11499 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11502 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11504 PR rtl-optimization/70478
11505 * lra-constraints.c (curr_small_class_check): New.
11506 (update_and_check_small_class_inputs): New.
11507 (process_alt_operands): Update curr_small_class_check. Disfavor
11508 alternative insn memory operands. Check available regs for small
11511 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11514 * config/mips/mips.opt (-mvirt): Update description.
11515 * doc/invoke.texi (-mvirt): Likewise.
11517 2017-04-10 Richard Biener <rguenther@suse.de>
11519 PR middle-end/80362
11520 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11521 looking for NEGATE_EXPR in -A / -B to A / B folding.
11523 2017-04-10 Martin Liska <mliska@suse.cz>
11525 PR gcov-profile/80224
11526 * gcov.c (print_usage): Fix usage string.
11527 (get_gcov_intermediate_filename): Remove.
11528 (output_gcov_file): Use both for normal and intermediate format.
11529 (generate_results): Do not initialize special file for
11530 intermediate format.
11532 2017-04-10 Richard Biener <rguenther@suse.de>
11534 PR tree-optimization/80304
11535 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11538 2017-04-10 Nathan Sidwell <nathan@acm.org>
11541 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11542 (rs6000_init_builtins): Use it.
11544 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11546 * config/arm/arm.md (<mrc>): Add mode to SET source.
11547 (<mrrc>): Likewise.
11549 2017-04-10 Richard Biener <rguenther@suse.de>
11551 PR middle-end/80344
11552 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11554 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11557 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11558 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11559 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11560 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11561 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11562 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11563 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11564 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11565 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11566 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11567 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11568 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11569 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11570 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11571 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11572 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11573 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11574 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11575 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11576 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11577 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11578 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11579 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11581 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11583 PR rtl-optimization/70478
11584 * lra-constraints.c: Reverse the last patch.
11586 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11588 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11589 Add comment for WCHAR_T.
11591 2017-04-08 Martin Liska <mliska@suse.cz>
11594 2017-04-07 Martin Liska <mliska@suse.cz>
11597 * ipa-split.c (split_function): Add function part to a same comdat
11600 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11603 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11605 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11607 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11608 (vec_pairing): ...to this.
11609 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11610 (rs6000_sched_init): Adjust for name change.
11611 (struct rs6000_sched_context): Likewise.
11612 (rs6000_init_sched_context): Likewise.
11613 (rs6000_set_sched_context): Likewise.
11615 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11621 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11623 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11624 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11625 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11627 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11629 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11631 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11633 PR rtl-optimization/70703
11634 * ira-color.c (update_conflict_hard_regno_costs): Use
11635 int64_t instead of HOST_WIDE_INT.
11637 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11639 PR rtl-optimization/70478
11640 * lra-constraints.c (process_alt_operands): Disfavor alternative
11641 insn memory operands.
11643 2017-04-07 Jeff Law <law@redhat.com>
11645 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11646 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11648 2017-04-07 Martin Liska <mliska@suse.cz>
11651 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11652 Show error message instead of an ICE.
11654 2017-04-07 Martin Liska <mliska@suse.cz>
11657 * ipa-split.c (split_function): Add function part to a same comdat
11660 2017-04-07 Richard Biener <rguenther@suse.de>
11662 PR middle-end/80341
11663 * tree.c (get_unwidened): Also handle ! for_type case for
11665 * convert.c (do_narrow): Split out from ...
11666 (convert_to_integer_1): ... here. Do not pass final truncation
11667 type to get_unwidened for TRUNC_DIV_EXPR.
11669 2017-04-07 Richard Biener <rguenther@suse.de>
11671 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11673 (aff_combination_const): Adjust.
11674 (aff_combination_scale): Likewise.
11675 (aff_combination_add_elt): Likewise.
11676 (aff_combination_add_cst): Likewise.
11677 (aff_combination_convert): Likewise.
11678 (add_elt_to_tree): Likewise. Remove unused argument.
11679 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11681 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11683 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11685 * config/arm/arm.c (arm_default_short_enums): Use
11686 ARM_DEFAULT_SHORT_ENUMS.
11687 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11689 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11692 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11693 members with redundant out-of-class redeclaration.
11695 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11698 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11699 * config/i386/i386.md (*zero_extendsidi2):
11700 Add (?*x,*x) and (?*v,*v) alternatives.
11702 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11705 * config/i386/i386.c (ix86_expand_builtin)
11706 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11707 mode from insn data. Convert operands to insn operand mode.
11708 Copy operands that don't satisfy insn predicate to a register.
11710 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11712 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11715 2017-04-06 Richard Biener <rguenther@suse.de>
11717 PR tree-optimization/80334
11718 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11719 preserve alignment of accesses.
11721 2017-04-06 Richard Biener <rguenther@suse.de>
11723 PR tree-optimization/80262
11724 * tree-sra.c (build_ref_for_offset): Preserve address-space
11726 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
11727 Drop useless address-space information on MEM_REF offsets.
11729 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
11731 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
11733 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11735 PR rtl-optimization/70703
11736 * ira-color.c (update_conflict_hard_regno_costs): Use
11737 HOST_WIDE_INT instead of long.
11739 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
11742 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
11743 not defined for x86_64 target. Add -mmmx target option when __SSE2__
11745 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
11746 for x86_64 target. Handle -m3dnowa option.
11748 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11750 PR rtl-optimization/70703
11751 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
11752 (update_conflict_hard_regno_costs): Use long instead of unsigned
11753 arithmetic for cost calculation.
11755 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11756 Bernd Edlinger <bernd.edlinger@hotmail.de>
11759 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
11762 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
11765 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
11766 ptr_mode with Pmode throughout.
11767 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
11768 into probe_stack_range and use DImode.
11770 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
11773 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
11774 call_eh_return is true.
11776 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11778 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
11779 Initialize last_match_fntype_index.
11781 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11784 * tree-nvr.c: Include internal-fn.h.
11785 (pass_return_slot::execute): Ignore internal calls without
11788 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11789 Richard Biener <rguenther@suse.de>
11792 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
11793 captures used multiple times, except for the last use.
11794 * generic-match-head.c: Include gimplify.h.
11796 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11798 PR tree-optimization/79390
11799 * target.h (struct noce_if_info): Declare.
11800 * targhooks.h (default_noce_conversion_profitable_p): Declare.
11801 * target.def (noce_conversion_profitable_p): New target hook.
11802 * ifcvt.h (struct noce_if_info): New type, moved from ...
11803 * ifcvt.c (struct noce_if_info): ... here.
11804 (noce_conversion_profitable_p): Renamed to ...
11805 (default_noce_conversion_profitable_p): ... this. No longer
11807 (noce_try_store_flag_constants, noce_try_addcc,
11808 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
11809 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
11810 instead of noce_conversion_profitable_p.
11811 * config/i386/i386.c: Include ifcvt.h.
11812 (ix86_option_override_internal): Don't override
11813 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
11814 (ix86_noce_conversion_profitable_p): New function.
11815 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
11816 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
11817 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
11818 * doc/tm.texi: Regenerated.
11820 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11822 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
11825 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11828 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
11829 instructions for small multiply cores.
11831 2017-04-04 Jeff Law <law@redhat.com>
11833 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
11835 (mips_expand_vec_perm_const): Initialize elements in orig_perm
11836 that are not set by the loop over the elements.
11838 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11841 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
11842 int mode, convert_modes it to mode as unsigned, otherwise use
11843 lowpart_subreg to mode rather than SImode.
11844 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
11845 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
11846 Use DImode instead of SImode for the shift count operand.
11847 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
11850 2017-04-04 Richard Biener <rguenther@suse.de>
11852 PR middle-end/80281
11853 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
11854 arithmetic done for the negate or the plus. Simplify.
11855 (A - (-B) -> A + B): Likewise.
11856 * fold-const.c (split_tree): Make sure to not negate pointers.
11858 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
11860 PR rtl-optimization/60818
11861 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
11862 a compare of comparisons with the thing compared if this results
11863 in a different machine mode.
11865 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
11867 * alias.c (base_alias_check): Fix typo in comment.
11868 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
11869 * cgraphunit.c (symbol_table::compile): Likewise.
11870 * collect2.c (maybe_run_lto_and_relink): Likewise.
11871 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
11872 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
11873 * config/avr/avr.c (avr_map_op_t): Likewise.
11874 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
11875 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
11876 * config/epiphany/epiphany.md (movcc): Likewise.
11877 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
11878 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
11880 * config/mips/mips.c (mips_save_restore_reg): Likewise.
11881 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
11882 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
11883 * config/sh/sh.c (sh_rtx_costs): Likewise.
11884 * fold-const.c (fold_truth_andor): Likewise.
11885 * genautomata.c (collapse_flag): Likewise.
11886 * gengtype.h (struct type::u::s): Likewise.
11887 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
11888 * input.c (FORMAT_AMOUNT): Likewise.
11889 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
11890 (known_aggs_to_agg_replacement_list): Likewise.
11891 * ipa-inline-analysis.c: Likewise.
11892 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
11893 * ipa-polymorphic-call.c
11894 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
11895 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
11896 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
11898 * modulo-sched.c (apply_reg_moves): Likewise.
11899 * omp-expand.c (build_omp_regions_1): Likewise.
11900 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
11901 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
11902 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
11903 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
11904 * value-prof.c: Likewise.
11905 * var-tracking.c (val_reset): Likewise.
11907 2017-04-03 Richard Biener <rguenther@suse.de>
11909 PR tree-optimization/80275
11910 * fold-const.c (split_address_to_core_and_offset): Handle
11913 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
11915 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
11916 descriptors is at least equal to that of functions.
11918 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11920 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
11922 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11925 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
11926 (mov<IMOD4:mode>): New expander.
11927 (*mov<IMOD4:mode>_internal): New insn and split pattern.
11929 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11931 PR rtl-optimization/79405
11932 * fwprop.c (propagations_left): New variable.
11933 (forward_propagate_into): Decrement it.
11934 (fwprop_init): Initialize it.
11935 (fw_prop): If the variable has reached zero, stop propagating.
11936 (fwprop_addr): Ditto.
11938 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11941 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
11942 a FUNCTION_DECL, pass it as decl instead of origin to
11945 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
11947 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
11950 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
11953 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
11954 TARGET_VSX_SMALL_INTEGER.
11956 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11958 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11959 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
11961 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11963 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
11964 extraction from odd-numbered MSA register.
11966 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11968 PR middle-end/80173
11969 * expmed.c (store_bit_field_1): Don't attempt to create
11970 a word subreg out of hard registers wider than word if they
11971 have HARD_REGNO_NREGS of 1 for their mode.
11973 PR middle-end/80163
11974 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
11975 conversions to integer types wider than word and pointer.
11978 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
11979 (rtx_equal_for_cselib_p): Pass 0 to it.
11980 * cselib.c (cselib_hasher::equal): Likewise.
11981 (rtx_equal_for_cselib_1): Add depth argument. If depth
11982 is 128, don't look up VALUE locs and punt. Increment
11983 depth in recursive calls when walking VALUE locs.
11985 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
11987 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
11988 (make_gcov_file_name): Use the canonical path name for generating
11990 (read_line): Fix handling of files with ascii null bytes.
11992 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
11994 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
11995 to initialise a vector register instead
11996 of using a const_int.
11998 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12000 PR translation/80189
12001 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12002 diagnostic messages.
12004 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12007 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12008 (dfp_diex_<mode>): Update mode of operand 1.
12009 * doc/extend.texi (dxex, dxexq): Document change to return type.
12010 (diex, diexq): Document change to argument type.
12012 2017-03-30 Martin Jambor <mjambor@suse.cz>
12015 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12016 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12017 it reflects the signature changes performed at the callee side.
12018 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12019 to cgraph_build_function_type_skip_args.
12020 (build_function_decl_skip_args): Adjust call to the above function.
12022 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12025 * config/i386/sse.md
12026 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12027 register as dest whenever it is a MEM not rtx_equal_p to the
12028 corresponding dup operand, and when forcing into reg move the
12029 reg into the memory afterwards.
12030 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12031 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12032 for the force_reg mode.
12033 (avx512vl_vextractf128<mode>): Use register as dest either
12034 always when a MEM, or when it is a MEM not rtx_equal_p to the
12035 corresponding dup operand, or even not when it is a CONST_VECTOR
12036 depending on the mode and lo vs. hi.
12037 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12039 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12040 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12041 Likewise. Require that operands[2] is even.
12042 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12043 Remove extraneous parens. Require that operands[2] is a multiple
12045 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12046 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12047 disallow memory then.
12049 2017-03-30 Richard Biener <rguenther@suse.de>
12051 PR tree-optimization/77498
12052 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12053 to non-constants over backedges.
12055 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12057 PR rtl-optimization/80233
12058 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12059 as last_combined_insn. Do not test for BARRIER_P separately.
12061 2017-03-29 Andreas Schwab <schwab@suse.de>
12064 * calls.c (prepare_call_address): Convert funexp to Pmode before
12065 copying to temp reg.
12067 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12069 PR tree-optimization/80158
12070 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12071 Handle possible future case of more than one alternate
12073 (replace_rhs_if_not_dup): Likewise.
12074 (replace_one_candidate): Likewise.
12076 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12078 PR rtl-optimization/80193
12079 * ira.c (ira): Do not check allocation for LRA.
12081 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12083 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12084 (nvptx_output_simt_exit): Declare.
12085 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12086 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12087 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12088 (nvptx_declare_function_name): ...here. Emit declaration of local
12089 memory space buffer for omp_simt_enter insn.
12090 (nvptx_output_unisimt_switch): New.
12091 (nvptx_output_softstack_switch): New.
12092 (nvptx_output_simt_enter): New.
12093 (nvptx_output_simt_exit): New.
12094 * config/nvptx/nvptx.h (struct machine_function): New fields
12095 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12096 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12097 (UNSPECV_SIMT_EXIT): Ditto.
12098 (omp_simt_enter_insn): New insn.
12099 (omp_simt_enter): New expansion.
12100 (omp_simt_exit): New insn.
12101 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12103 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12104 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12105 (expand_GOMP_SIMT_EXIT): New.
12106 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12107 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12108 (GOMP_SIMT_EXIT): Ditto.
12109 * target-insns.def (omp_simt_enter): New insn.
12110 (omp_simt_exit): Ditto.
12111 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12113 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12114 (lower_rec_input_clauses): Likewise.
12115 (lower_lastprivate_clauses): Handle SIMT privatization.
12117 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12118 (ompdevlow_adjust_simt_enter): New.
12119 (find_simtpriv_var_op): New.
12120 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12121 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12123 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12124 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12125 (copy_decl_for_dup_finish): Ditto.
12127 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12129 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12132 * config/i386/i386.c (ix86_option_override_internal): Always
12133 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12135 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12137 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12139 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12141 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12142 mark new edge's irreducible flag accordign to it.
12143 (vect_do_peeling): Check loop preheader edge's irreducible flag
12144 and pass it to function slpeel_add_loop_guard.
12146 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12148 PR tree-optimization/80218
12149 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12150 Update block frequencies and counts.
12152 2017-03-28 Richard Biener <rguenther@suse.de>
12154 PR tree-optimization/78644
12155 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12156 of a simplification result we may not use it at all.
12158 2017-03-28 Richard Biener <rguenther@suse.de>
12161 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12162 without arguments, generate default definition of a SSA name.
12164 2017-03-28 Richard Biener <rguenther@suse.de>
12166 PR middle-end/80222
12167 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12168 TYPE_REF_CAN_ALIAS_ALL references.
12169 * fold-const.c (fold_indirect_ref_1): Likewise.
12171 2017-03-28 Martin Liska <mliska@suse.cz>
12174 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12175 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12177 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12178 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12180 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12181 (EXTRA_SPECS): Define.
12182 (SUBTARGET_EXTRA_SPECS): Likewise.
12183 (SUBTARGET_CPP_SPEC): Likewise.
12184 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12185 SUBTARGET_EXTRA_SPECS.
12186 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12188 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12190 * config/arc/simdext.md (vst64_insn): Update pattern.
12191 (vld32wh_insn): Likewise.
12192 (vld32wl_insn): Likewise.
12193 (vld64_insn): Likewise.
12194 (vld32_insn): Likewise.
12196 2017-03-28 Marek Polacek <polacek@redhat.com>
12199 * fold-const.c (fold_comparison): Use protected_set_expr_location
12200 instead of SET_EXPR_LOCATION.
12202 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12204 * tree.c (add_expr): Avoid name lookup warning.
12206 2017-03-27 Jeff Law <law@redhat.com>
12208 PR tree-optimization/80216
12209 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12210 function name. Limit recursion depth.
12211 (record_temporary_equivalences): Corresponding changes.
12213 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12215 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12218 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12221 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12223 * cfgcleanup.c (reg_note_cfa_p): New array.
12224 (insns_have_identical_cfa_notes): New function.
12225 (old_insns_match_p): Don't cross-jump in between /f
12226 and non-/f instructions. If both i1 and i2 are frame related,
12227 verify all CFA notes, their order and content.
12229 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12232 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12233 HImode and SImode with zero extend to DImode to one insn.
12234 (bswap<mode>2_extenddi): Likewise.
12235 (bswapsi2_extenddi): Likewise.
12236 (bswaphi2_extendsi): Likewise.
12237 (bswaphi2): Combine bswap HImode and SImode into one insn.
12238 Separate memory insns from swapping register.
12239 (bswapsi2): Likewise.
12240 (bswap<mode>2): Likewise.
12241 (bswaphi2_internal): Delete, no longer used.
12242 (bswapsi2_internal): Likewise.
12243 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12244 store, and gpr<-gpr swap insns.
12245 (bswap<mode>2_store): Likewise.
12246 (bswaphi2_reg): Register only splitter, combine with the splitter.
12247 (bswaphi2 splitter): Likewise.
12248 (bswapsi2_reg): Likewise.
12249 (bswapsi2 splitter): Likewise.
12250 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12251 the insns into load, store, and register/register insns.
12252 (bswapdi2_ldbrx): Likewise.
12253 (bswapdi2_load): Likewise.
12254 (bswapdi2_store): Likewise.
12255 (bswapdi2_reg): Likewise.
12257 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12259 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12260 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12262 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12265 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12267 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12268 special handling for target option conflicts between dform
12269 options (-mpower9-dform, -mpower9-dform-vector,
12270 -mpower9-dform-scalar) and -mno-direct-move.
12272 2017-03-27 Richard Biener <rguenther@suse.de>
12274 PR tree-optimization/80181
12275 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12277 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12279 * config/arc/predicates.md (move_double_src_operand): Replace the
12280 call to move_double_src_operand with a call to address_operand.
12282 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12284 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12285 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12286 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12288 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12290 * config/arc/predicates.md (long_immediate_loadstore_operand):
12291 Consider scaled addresses cases.
12293 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12295 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12296 restored when in interrupt.
12297 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12298 doesn't have delay slot.
12300 2017-03-27 Richard Biener <rguenther@suse.de>
12303 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12304 inlined thunk clones.
12306 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12309 * asan.c (instrument_derefs): Copy over last operand from
12310 original COMPONENT_REF to the new COMPONENT_REF with
12311 DECL_BIT_FIELD_REPRESENTATIVE.
12312 * ubsan.c (instrument_object_size): Likewise.
12314 2017-03-27 Richard Biener <rguenther@suse.de>
12316 PR tree-optimization/80170
12317 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12318 sure DR/SCEV didnt fold in constants we do not see when looking
12319 at the reference base alignment.
12321 2017-03-27 Richard Biener <rguenther@suse.de>
12323 PR middle-end/80171
12324 * gimple-fold.c (fold_ctor_reference): Properly guard against
12325 NULL return value from canonicalize_constructor_val.
12327 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12330 * config/i386/i386.c (ix86_expand_builtin)
12331 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12332 flags reg setting and flags reg using instructions.
12333 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12334 clobbering instructions to zero extend op2.
12336 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12338 * doc/install.texi (Configuration) <--with-aix-soname>:
12339 Update link to AIX ld.
12341 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12343 PR rtl-optimization/80160
12344 PR rtl-optimization/80159
12345 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12346 reg_alternate_class into account.
12348 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12351 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12352 to consider in curr_insn_transform.
12354 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12356 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12357 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12358 and emit_mode_inner.
12360 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12362 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12363 argument to the overloaded builtin variants. Use the new flag to
12364 deprecate certain builtin variants.
12365 * config/s390/s390-builtin-types.def: Add new builtin types.
12366 * config/s390/s390-builtins.h: Support new flags field for
12367 overloaded builtins.
12368 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12369 (s390_macro_to_expand): Enable vector float data type.
12370 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12371 builtins by incrementing the __VEC__ version number.
12372 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12374 (s390_resolve_overloaded_builtin): Emit error messages depending
12375 on the builtin flags.
12376 * config/s390/s390.c (s390_expand_builtin): Support additional
12377 flags argument. Change error message to match the messages
12378 emitted in s390-c.c.
12379 * config/s390/s390.md: New UNSPEC_* constants.
12380 (op_type): Add new instruction types.
12381 * config/s390/vecintrin.h: Add new builtins and test data class
12383 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12384 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12385 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12386 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12387 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12388 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12390 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12391 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12392 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12393 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12395 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12396 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12397 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12398 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12400 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12401 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12402 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12403 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12404 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12405 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12406 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12408 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12409 ("vec_scatter_element<V_HW_4:mode>_DI")
12410 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12411 ("vec_fpint<mode>", "vflls")
12412 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12413 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12414 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12415 ("*vec_cmphe<mode>_cc"): ... these.
12417 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12418 mode constant instead of magic value.
12420 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12422 * config/s390/s390.c (s390_expand_vec_compare): Support other
12423 vector floating point modes than just V2DF.
12424 (s390_expand_vcond): Likewise.
12425 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12426 (s390_cannot_change_mode_class): Prevent mode changes between TF
12427 and V1TF in vector registers.
12428 * config/s390/s390.md (DF, SF): New mode attributes.
12429 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12430 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12431 SFmode support for VRs.
12432 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12434 (VFT, VF_HW): New mode iterators.
12435 (vw, sdx): New mode attributes.
12436 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12437 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12438 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12439 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12440 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12441 also the new vector floating point modes. Renaming to ...
12443 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12444 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12445 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12446 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12447 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12448 ("vec_unordered<mode>"): ... these.
12450 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12451 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12452 ("*vec_extendv2df"): New insn definitions.
12454 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12456 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12457 ("mulditi3_2", "*muldi3_sign"): New patterns.
12458 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12459 rename the pattern definition.
12461 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12463 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12465 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12467 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12469 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12470 instruction if possible.
12471 * config/s390/vector.md (vec_halfnumelts): New mode
12473 ("*vec_vllezlf<mode>"): New pattern.
12475 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12477 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12478 ("popcountv4si2", "popcountv2di2"): Rename to ...
12479 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12480 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12482 ("popcount<mode>2_vxe"): New pattern.
12484 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12486 * common/config/s390/s390-common.c (processor_flags_table): Add
12488 * config.gcc: Add arch12.
12489 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12490 Default to arch12 for unknown CPU model numbers.
12491 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12492 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12493 PROCESSOR_max sanity check.
12494 * config/s390/s390-opts.h (enum processor_type): Add
12496 * config/s390/s390.c (processor_table): Add arch12.
12497 (s390_expand_builtin): Add check for B_VXE flag.
12498 (s390_issue_rate): Add PROCESSOR_ARCH12.
12499 (s390_get_sched_attrmask): Likewise.
12500 (s390_get_unit_mask): Likewise.
12501 (s390_sched_score): Enable z13 scheduling for arch12.
12502 (s390_sched_reorder): Likewise.
12503 (s390_sched_variable_issue): Likewise.
12504 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12506 (s390_tune_attr): Use z13 scheduling also for arch12.
12507 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12508 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12509 (TARGET_VXE_P): New macros.
12510 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12511 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12512 * config/s390/s390.opt: Add arch12 as processor_type.
12514 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12516 * config/s390/s390.md
12517 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12518 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12519 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12521 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12522 Rename expanders to ...
12524 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12525 ("fixuns_truncdddi2_emu"): ... these.
12527 ("fixuns_trunc<mode>si2_emu"): New expander.
12529 ("*fixuns_truncdfdi2_z13"): Rename to ...
12530 ("*fixuns_truncdfdi2_vx"): ... this.
12532 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12534 * config/s390/2964.md: Remove the single element vector compare
12535 instructions which are no longer used.
12536 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12538 (s390_canonicalize_comparison): Remove handling of DFmode
12540 (s390_expand_vec_compare_scalar): Remove function.
12541 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12542 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12544 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12546 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12548 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12550 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12551 will anyway by matched by mov<mode>_64dfp.
12553 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12555 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12556 vlef/vstef. Add missing operand to vleif.
12558 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12560 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12561 pair for all vector types with 64 bit elements.
12562 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12563 * config/s390/vector.md (V_HW_64): ... here.
12564 (V_128_NOSINGLE): New mode iterator.
12565 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12566 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12567 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12568 ("*vec_load_pairv2di"): Change to ...
12569 ("*vec_load_pair<mode>"): ... this one.
12571 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12573 * config/s390/constraints.md: Add comments.
12574 (jKK): Reject element sizes > 8 bytes.
12575 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12577 * config/s390/s390.md: Add the s_operand checks formerly in
12578 s390_split_ok_p to various splitters where they are still
12580 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12581 for 128 bit vectors. Plus two splitters.
12583 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12585 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12588 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12591 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12592 error if the boundary argument is not constant.
12594 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12596 PR rtl-optimization/80112
12597 * loop-doloop.c (doloop_condition_get): Don't check condition
12598 if cmp isn't SET with IF_THEN_ELSE src.
12600 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12602 PR tree-optimization/80158
12603 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12604 replacing a candidate statement, also replace it for the
12605 candidate's alternate interpretation.
12606 (replace_rhs_if_not_dup): Likewise.
12607 (replace_one_candidate): Likewise.
12609 2017-03-24 Richard Biener <rguenther@suse.de>
12611 PR tree-optimization/80167
12612 * graphite-isl-ast-to-gimple.c
12613 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12615 (translate_isl_ast_to_gimple::get_rename): Likewise.
12617 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12619 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12620 handling of certain combinations of target options, including the
12621 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12622 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12624 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12627 * config/arm/arm.md (*load_multiple): Add reload_completed to
12628 matching condition.
12630 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12631 Richard Biener <rguenth@suse.de>
12633 PR tree-optimization/79908
12634 PR tree-optimization/80136
12635 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12636 been cast away, gimplify_and_add suffices.
12638 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12640 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12642 2017-03-23 Richard Biener <rguenther@suse.de>
12644 PR tree-optimization/80032
12645 * gimplify.c (gimple_push_cleanup): Forced unconditional
12646 cleanups still have to go to the conditional_cleanups
12649 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12651 PR tree-optimization/80072
12652 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12654 (next_operand_entry_id): Change type to unsigned int.
12655 (sort_by_operand_rank): Make sure to return the right return value
12656 even if unsigned fields are bigger than INT_MAX.
12657 (struct oecount): Change cnt and id type to unsigned int.
12658 (oecount_hasher::equal): Formatting fix.
12659 (oecount_cmp): Make sure to return the right return value
12660 even if unsigned fields are bigger than INT_MAX.
12661 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12664 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12665 TREE_READONLY on result if writing it more than once.
12668 * doc/invoke.texi (-fsanitize=thread): Document that with
12669 -fnon-call-exceptions atomics are not able to throw
12673 * tsan.c: Include tree-eh.h.
12674 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12675 maybe_clean_or_replace_eh_stmt where needed.
12676 (instrument_memory_accesses): Add cfg_changed argument.
12677 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12678 if it returned true.
12679 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12681 PR rtl-optimization/63191
12682 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12683 wrapper function, moved the whole old content into ...
12684 (ix86_delegitimize_address_1): ... this. New inline function.
12685 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12686 true as last argument instead of ix86_delegitimize_address.
12688 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12690 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12691 cortexa57_branch_cost.
12693 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12695 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12697 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12700 * doc/md.texi (Constraints): Document wA constraint.
12701 * config/rs6000/constraints.md (wA): New.
12702 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12703 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12704 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12705 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12707 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12710 * gimplify.c (is_oacc_declared): New function.
12711 (oacc_default_clause): Use it to set default flags for acc declared
12712 variables inside parallel regions.
12713 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12714 declared variables.
12715 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12716 declare attribute to any decl as necessary.
12718 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12721 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
12722 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
12723 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
12724 (arm_arch_lpae): This.
12725 * config/arm/arm.c (arm_arch7ve): Rename into ...
12726 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
12727 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
12730 2017-03-22 Martin Liska <mliska@suse.cz>
12733 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
12734 error message instead of an ICE.
12736 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12738 * doc/extend.texi (6.11 Additional Floating Types): Revise.
12740 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
12742 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12744 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12747 2017-03-21 Martin Sebor <msebor@redhat.com>
12749 * doc/extend.texi: Use "cannot" instead of "can't."
12750 * doc/hostconfig.texi: Same.
12751 * doc/install.texi: Same.
12752 * doc/invoke.texi: Same.
12753 * doc/loop.texi: Same.
12754 * doc/md.texi: Same.
12755 * doc/objc.texi: Same.
12756 * doc/rtl.texi: Same.
12757 * doc/tm.texi: Same.
12758 * doc/tm.texi.in: Same.
12759 * doc/trouble.texi: Same.
12761 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
12764 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
12765 (collect_checksum_attributes): Set it.
12766 (die_checksum_ordered): Use it.
12768 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12770 PR tree-optimization/79908
12771 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
12772 change: For a VA_ARG whose LHS has been cast away, use
12773 force_gimple_operand to construct the side effects.
12775 2017-03-21 David Malcolm <dmalcolm@redhat.com>
12777 PR translation/80001
12778 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
12779 more amenable to translation.
12780 (oacc_loop_auto_partitions): Likewise.
12782 2017-03-21 Marek Polacek <polacek@redhat.com>
12783 Martin Sebor <msebor@redhat.com>
12785 PR tree-optimization/80109
12786 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
12787 on INTEGRAL_TYPE_P.
12789 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12790 Segher Boessenkool <segher@kernel.crashing.org>
12793 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
12794 check reg_used_between_p between insn and one of succ or succ2
12795 depending on if succ is artificial insn not inserted into insn
12798 2017-03-21 Martin Liska <mliska@suse.cz>
12800 PR gcov-profile/80081
12801 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
12802 * doc/gcc.texi: Include gcov-dump stuff.
12803 * doc/gcov-dump.texi: New file.
12805 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
12807 PR rtl-optimization/79150
12808 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
12809 conditional jump, if the jump is the last insn of the loop.
12811 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12812 Richard Biener <rguenth@suse.de>
12814 PR tree-optimization/79908
12815 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12816 been cast away, use force_gimple_operand to construct the side
12819 2017-03-21 Martin Liska <mliska@suse.cz>
12821 PR libfortran/79956
12822 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
12825 2017-03-21 Brad Spengler <spender@grsecurity.net>
12828 * plugin.c (htab_hash_plugin): New function.
12829 (add_new_plugin): Use it and adjust.
12830 (parse_plugin_arg_opt): Adjust.
12831 (init_one_plugin): Likewise.
12833 2017-03-21 Richard Biener <rguenther@suse.de>
12835 PR tree-optimization/80032
12836 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
12837 if set force the cleanup to happen unconditionally.
12838 (gimplify_target_expr): Push inserted clobbers with force_uncond
12839 to avoid them being removed by control-dependent DCE.
12841 2017-03-21 Richard Biener <rguenther@suse.de>
12843 PR tree-optimization/80122
12844 * tree-inline.c (copy_bb): Do not expans va-arg packs or
12845 va_arg_pack_len when the inlined call stmt requires pack
12847 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
12849 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12852 * tsan.c (instrument_builtin_call): If the memory model argument
12853 is not a constant, assume it is valid.
12856 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
12859 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12861 PR rtl-optimization/79910
12862 * combine.c (can_combine_p): Do not allow combining an I0 or I1
12863 if its dest is used by an insn before I2 (other than the combined
12864 insns themselves, which are properly handled already).
12866 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12869 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12871 * combine.c (record_used_regs): New static function.
12872 (try_combine): Handle situations where there is an additional
12873 instruction between I2 and I3 which needs to have a LOG_LINK
12877 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12879 * combine.c (try_combine): Delete redundant i1 test. Call
12880 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12882 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12885 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
12886 alternatives 13/14.
12888 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12890 PR tree-optimization/80054
12891 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
12892 the optimization if a PHI or any of its arguments is not dominated
12893 by the candidate's basis. Use gphi* rather than gimple* as
12895 (replace_profitable_candidates): Clean up a gimple* variable that
12896 should be a gphi* variable.
12898 2017-03-20 Martin Sebor <msebor@redhat.com>
12901 * doc/extend.texi (attribute constructor): Document present limitation.
12903 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12906 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
12907 __POWER9_VECTOR__ #ifdef control, change template definition to
12908 use Power9-specific built-in function.
12909 (vec_any_eq): Likewise.
12910 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
12911 to control outcomes from this test.
12912 (vector_ae_<mode>p): For VEC_F modes, likewise.
12914 2017-03-20 Ian Lance Taylor <iant@google.com>
12916 * config/i386/i386.c (ix86_function_regparm): Save an extra
12917 register for -fsplit-stack with DECL_STATIC_CHAIN.
12919 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12922 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
12923 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
12925 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12927 * config/riscv/riscv.c (riscv_print_operand): Use "fence
12929 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
12932 2017-03-20 Marek Polacek <polacek@redhat.com>
12935 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
12937 2017-03-20 Richard Biener <rguenther@suse.de>
12939 PR tree-optimization/80113
12940 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
12941 allocate extra SSA name for PHI def.
12942 (add_close_phis_to_outer_loops): Likewise.
12943 (add_close_phis_to_merge_points): Likewise.
12944 (copy_loop_close_phi_args): Likewise.
12945 (copy_cond_phi_nodes): Likewise.
12947 2017-03-20 Martin Liska <mliska@suse.cz>
12949 PR middle-end/79753
12950 * tree-chkp.c (chkp_build_returned_bound): Do not build
12951 returned bounds for a LHS that's not a BOUNDED_P type.
12953 2017-03-20 Martin Liska <mliska@suse.cz>
12957 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
12958 COMPLEX_CST and VECTOR_CST.
12960 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12963 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
12964 target operand. A new splitter adds the clobber statement in case
12965 the target operand is dead anyway.
12967 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
12969 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
12970 to age-old versions of binutils and glibc.
12972 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
12974 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
12976 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12978 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
12980 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12982 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
12983 requirement for binutils 2.13.
12985 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12987 * combine.c (try_combine): Delete redundant i1 test. Call
12988 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12990 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
12992 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
12993 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
12995 <riscv64-*-elf>: Re-arrange section
12996 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
12997 <riscv32-*-linux>: Likewise.
12998 <riscv64-*-elf>: Likewise
12999 <riscv64-*-linux>: Likewise.
13001 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13004 * aarch64.opt(verbose-cost-dump): Fix typo.
13006 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13009 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13010 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13012 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13014 * reload.c (find_reloads): When reloading a nonoffsettable address,
13015 use RELOAD_OTHER for it and its address reloads.
13017 PR rtl-optimization/79910
13018 * combine.c (record_used_regs): New static function.
13019 (try_combine): Handle situations where there is an additional
13020 instruction between I2 and I3 which needs to have a LOG_LINK
13023 2017-03-17 Jeff Law <law@redhat.com>
13025 PR tree-optimization/71437
13026 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13027 conditional in the hash table first.
13028 (vrp_dom_walker::before_dom_children): Extract condition from
13029 ASSERT_EXPR. Record condition, its inverion and any implied
13030 conditions as well.
13032 2017-03-17 Marek Polacek <polacek@redhat.com>
13033 Markus Trippelsdorf <markus@trippelsdorf.de>
13035 PR tree-optimization/80079
13036 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13039 2017-03-17 Richard Biener <rguenther@suse.de>
13041 PR middle-end/80075
13042 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13043 Properly verify the LHS before the RHS possibly claims to be
13045 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13048 2017-03-17 Martin Jambor <mjambor@suse.cz>
13050 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13051 (List of -O2 options): Likewise.
13052 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13055 2017-03-17 Tom de Vries <tom@codesourcery.com>
13057 * gcov-dump.c (print_usage): Print bug_report_url.
13059 2017-03-17 Richard Biener <rguenther@suse.de>
13061 PR middle-end/80050
13062 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13063 (parser::peek): Likewise.
13065 2017-03-17 Richard Biener <rguenther@suse.de>
13067 PR tree-optimization/80048
13068 * sese.c (free_sese_info): Properly release rename_map and
13069 copied_bb_map elements.
13071 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13073 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13074 Add linked-list forward and backlinks. Insert on
13075 construction, remove on destruction.
13076 (class pass_store_merging): Add m_stores_head field.
13077 (pass_store_merging::terminate_and_process_all_chains):
13078 Iterate over m_stores_head list.
13079 (pass_store_merging::terminate_all_aliasing_chains):
13081 (pass_store_merging::execute): Check for debug stmts first.
13082 Push new chains onto the m_stores_head stack.
13084 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13087 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13088 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13089 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13091 2017-03-16 Jeff Law <law@redhat.com>
13093 PR tree-optimization/71437
13094 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13095 member function. Implementation moved into after_dom_children
13096 member function and into the threader's thread_outgoing_edges
13098 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13099 some code into new thread_outgoing_edges.
13100 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13101 definition. Simplify marker handling (do it here). Assume we always
13102 have the available expression and the const/copies tables.
13103 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13105 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13106 * tree-vrp.c (equiv_stack): No longer file scoped.
13107 (vrp_dom_walker): New class.
13108 (vrp_dom_walker::before_dom_children): New member function.
13109 (vrp_dom_walker::after_dom_children): Likewise.
13110 (identify_jump_threads): Setup domwalker. Use it rather than
13111 walking edges in a random order by hand. Simplify setup/finalization.
13112 (finalize_jump_threads): Remove.
13113 (vrp_finalize): Do not call identify_jump_threads here.
13114 (execute_vrp): Do it here instead and call thread_through_all_blocks
13117 PR tree-optimization/71437
13118 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13120 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13122 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13123 (dom_opt_dom_walker::thread_across_edge): Remove
13124 handle_dominating_asserts argument. All callers changed.
13125 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13126 changes. Remove calls to lhs_of_dominating_assert. Other
13127 uses of handle_dominating_asserts turn into unconditional code
13128 (simplify_control_stmt_condition_1): Likewise.
13129 (simplify_control_stmt_condition): Likewise.
13130 (thread_through_normal_block, thread_across_edge): Likewise.
13131 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13132 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13133 object if it is not an SSA_NAME.
13134 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13135 before calling into the VRP specific simplifiers.
13136 (identify_jump_threads): Remove handle_dominating_asserts
13139 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13142 * tree-diagnostic.c (default_tree_printer): No longer static.
13143 * tree-diagnostic.h (default_tree_printer): New prototype.
13145 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13147 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13148 Change ins into fmov.
13150 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13152 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13153 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13154 Use h_con constraint for operand 1.
13155 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13156 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13158 2017-03-15 Jeff Law <law@redhat.com>
13160 PR tree-optimization/71437
13161 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13162 (record_temporary_equivalences): Use it.
13164 PR tree-optimization/71437
13165 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13166 tree-ssa-scopedtables.
13167 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13168 (record_conditions, record_cond, vuse_eq): Likewise.
13169 (record_edge_info): Adjust to API tweak of record_conditions.
13170 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13171 (record_temporary_equivalences, optimize_stmt): Likewise.
13172 (eliminate_redundant_computations): Likewise.
13173 (record_equivalences_from_stmt): Likewise.
13174 * tree-ssa-scopedtables.c: Include options.h and params.h.
13175 (vuse_eq): New function, moved from tree-ssa-dom.c
13176 (build_and_record_new_cond): Likewise.
13177 (record_conditions): Likewise. Accept vector of conditions rather
13178 than edge_equivalence structure for first argument.
13179 for the first argument.
13180 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13181 from tree-ssa-dom.c.
13182 (avail_exprs_stack::record_cond): Likewise.
13183 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13184 from tree-ssa-dom.c.
13185 (avail_exprs_stack): Add new member functions lookup_avail_expr
13187 (record_conditions): Declare.
13189 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13192 * lra-constraints.c (process_alt_operands): Increase reject for
13193 reloading an input/output operand.
13195 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13198 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13199 insns to convert from signed/unsigned char/short to IEEE 128-bit
13201 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13203 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13206 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13207 subreg of inner mode for values already in registers.
13209 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13211 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13212 iteration reg is used after the loop.
13214 2017-03-14 Martin Sebor <msebor@redhat.com>
13216 PR tree-optimization/79800
13217 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13218 precision in negative-positive range.
13219 (format_floating): Call non-const overload with adjusted precision.
13221 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13224 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13227 2017-03-14 Martin Sebor <msebor@redhat.com>
13229 PR middle-end/80020
13230 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13231 * builtins.def (aligned_alloc): Use it.
13234 * Makefile.in (GTFILES): Add calls.c.
13235 * calls.c: Include "gt-calls.h".
13237 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13239 PR rtl-optimization/79728
13240 * regs.h (struct target_regs): New field
13241 x_contains_allocatable_regs_of_mode.
13242 (contains_allocatable_regs_of_mode): New macro.
13243 * reginfo.c (init_reg_sets_1): Initialize it, and change
13244 contains_reg_of_mode so it includes global regs as well.
13245 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13246 rather than contains_regs_of_mode.
13248 2017-03-14 Martin Liska <mliska@suse.cz>
13250 * doc/invoke.texi: Document options that can't be combined with
13251 -fcheck-pointer-bounds.
13253 2017-03-14 Martin Liska <mliska@suse.cz>
13255 PR middle-end/79831
13256 * doc/invoke.texi (-Wchkp): Document the option.
13258 2017-03-14 Martin Liska <mliska@suse.cz>
13260 * Makefile.in: Install gcov-dump.
13262 2017-03-14 Martin Liska <mliska@suse.cz>
13264 * multiple_target.c (expand_target_clones): Bail out for
13265 an invalid attribute.
13267 2017-03-14 Richard Biener <rguenther@suse.de>
13269 * alias.c (struct alias_set_entry): Pack properly.
13270 * cfgloop.h (struct loop): Likewise.
13271 * cse.c (struct set): Likewise.
13272 * ipa-utils.c (struct searchc_env): Likewise.
13273 * loop-invariant.c (struct invariant): Likewise.
13274 * lra-remat.c (struct cand): Likewise.
13275 * recog.c (struct change_t): Likewise.
13276 * rtl.h (struct address_info): Likewise.
13277 * symbol-summary.h (function_summary): Likewise.
13278 * tree-loop-distribution.c (struct partition): Likewise.
13279 * tree-object-size.c (struct object_size_info): Likewise.
13280 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13281 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13282 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13283 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13284 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13285 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13286 (struct _stmt_vec_info): Likewise.
13288 2017-03-14 Martin Liska <mliska@suse.cz>
13291 * multiple_target.c (create_dispatcher_calls): Check that
13292 a target can create a function dispatcher.
13294 2017-03-14 Martin Liska <mliska@suse.cz>
13297 * multiple_target.c (expand_target_clones): Drop local.local
13298 flag for default implementation.
13300 2017-03-14 Richard Biener <rguenther@suse.de>
13302 PR tree-optimization/80030
13303 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13305 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13307 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13308 gcc_fallthrough() instead of __attribute__((fallthrough));
13310 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13312 * doc/gcc.texi: Remove "up" link to (DIR).
13313 * doc/gccint.texi: Ditto.
13315 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13317 * doc/install.texi (Specific) <avr>: Remove reference to
13320 2017-03-13 Jeff Law <law@redhat.com>
13322 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13323 attribute rather than comments.
13325 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13326 match_scratch operand is highest.
13328 2017-03-13 Martin Liska <mliska@suse.cz>
13330 PR middle-end/78339
13331 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13332 is a CHKP clone, use original declaration.
13334 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13336 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13337 (arc_conditional_register_usage): Use a different allocation order
13338 when optimizing for size.
13339 * common/config/arc/arc-common.c (arc_option_optimization_table):
13340 Section anchors default on when optimizing for size.
13342 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13344 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13346 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13348 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13349 * config/arc/arc.md (cpu_facility): Add cd variant.
13350 (*movqi_insn): Add code density variant.
13351 (*movhi_insn): Likewise.
13352 (*movqi_insn): Likewise.
13353 (*addsi3_mixed): Likewise.
13354 (subsi3_insn): Likewise.
13356 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13358 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13360 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13362 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13363 expressions with MINUS and UNARY ops.
13365 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13368 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13370 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13371 between vec_select and vector argument.
13372 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13373 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13374 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13375 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13376 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13377 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13379 2017-03-13 Richard Biener <rguenther@suse.de>
13382 * params.def (vect-max-peeling-for-alignment): Fix typo.
13384 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13386 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13387 issue that only occurred with binutils below 2.18.
13389 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13391 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13392 refer to binutils 2.11/2.12 minimum.
13394 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13396 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13397 ftp.kernel.org and simplify binutils requirement.
13399 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13401 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13403 (Optimize Options): Ditto. Also remove redundancy.
13405 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13407 PR translation/79848
13408 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13410 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13411 to G_ to avoid double translation.
13413 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13415 PR translation/79923
13416 * auto-profile.c (get_combined_location): Convert leading
13417 character of diagnostics to lower case and remove trailing period.
13418 (read_profile): Likewise for various diagnostics.
13419 * config/arm/arm.c (arm_option_override): Remove trailing period
13420 from various diagnostics.
13421 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13422 (msp430_expand_delay_cycles): Likewise.
13424 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13427 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13428 full command-line argument, rather than just "str".
13429 (aarch64_validate_march): Likewise.
13430 (aarch64_validate_mtune): Likewise.
13432 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13434 PR rtl-optimization/78911
13435 * lra-assigns.c (must_not_spill_p): New function.
13436 (spill_for): Use it.
13438 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13440 PR tree-optimization/79981
13441 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13442 ATOMIC_COMPARE_EXCHANGE ifn result.
13443 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13444 IFN_ATOMIC_COMPARE_EXCHANGE.
13446 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13449 * opts.c (parse_sanitizer_options): Add missing question mark to
13450 "did you mean" message.
13452 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13454 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13456 (VMULEUH_UNS): Likewise.
13457 (VMULOUB_UNS): Likewise.
13458 (VMULOUH_UNS): Likewise.
13459 * config/rs6000/rs6000.c (builtin_function_type): Remove
13460 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13462 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13465 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13466 x with result of extra_parsing_for_operand_code_0.
13467 (function_reader::extra_parsing_for_operand_code_0): Convert
13468 return type from void to rtx, returning x. When reading
13469 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13470 larger size containing struct block_symbol.
13472 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13474 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13475 -mfloat128-hardware without -m64.
13477 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13480 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13481 entries to the case statement that marks unsigned arguments to
13482 overloaded functions.
13484 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13486 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13487 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13489 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13492 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13493 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13495 2017-03-10 Martin Liska <mliska@suse.cz>
13499 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13500 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13503 2017-03-10 Olivier Hainque <hainque@adacore.com>
13505 * tree-switch-conversion (array_value_type): Start by resetting
13506 candidate type to it's main variant.
13508 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13510 PR rtl-optimization/79909
13511 * combine.c (try_combine): Use simplify_replace_rtx on individual
13512 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13513 of the whole CALL_INSN_FUNCTION_USAGE.
13515 PR tree-optimization/79972
13516 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13517 get_range_info on SSA_NAMEs. Formatting fixes.
13519 2017-03-10 Richard Biener <rguenther@suse.de>
13520 Jakub Jelinek <jakub@redhat.com>
13522 PR tree-optimization/77975
13523 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13524 edge to be constant.
13525 (get_val_for): For constant x return it. Formatting fix.
13526 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13527 would use the same bases as the current one.
13529 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13531 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13532 instead of vec_select for V1TImode.
13533 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13535 (VSX_LE_128): Add V1TI to this mode iterator.
13536 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13537 (*vsx_le_perm_store_<mode>): Likewise.
13538 (pre-reload splitter for VSX stores): Likewise.
13539 (post-reload splitter for VSX stores): Likewise.
13540 (*vsx_xxpermdi2_le_<mode>): Likewise.
13541 (*vsx_lxvd2x2_le_<mode>): Likewise.
13542 (*vsx_stxvd2x2_le_<mode>): Likewise.
13544 2017-03-09 Michael Eager <eager@eagercon.com>
13546 Correct failures with --enable-checking=yes,rtl.
13548 * config/microblaze/microblaze.c (microblaze_expand_shift):
13549 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13550 test for const0_rtx.
13551 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13552 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13554 2017-03-09 Richard Biener <rguenther@suse.de>
13556 PR tree-optimization/79977
13557 * graphite-scop-detection.c (scop_detection::merge_sese):
13558 Handle the case of extra exits to blocks dominating the entry.
13560 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13562 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13565 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13567 PR rtl-optimization/79949
13568 * lra-constraints.c (process_alt_operands): Check memory when
13569 trying to predict a cycle. Print about the overall increase.
13571 2017-03-09 Richard Biener <rguenther@suse.de>
13573 PR middle-end/79971
13574 * gimple-expr.c (useless_type_conversion_p): Preserve
13575 TYPE_SATURATING for fixed-point types.
13577 2017-03-09 Richard Biener <rguenther@suse.de>
13580 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13581 alignment of BLKmode params.
13583 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13586 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13587 (VALL_NO_V2Q): Likewise.
13589 * config/aarch64/aarch64-simd.md
13590 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13592 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13593 VALL_NO_V2Q mode iterator.
13594 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13596 2017-03-09 Martin Liska <mliska@suse.cz>
13598 PR tree-optimization/79631
13599 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13600 tree_int_cst_sign_bit just for INTEGER constants.
13602 2017-03-09 Martin Liska <mliska@suse.cz>
13606 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13609 2017-03-09 Marek Polacek <polacek@redhat.com>
13612 * tree.c (inchash::add_expr): Handle TREE_VEC.
13614 2017-03-09 Martin Liska <mliska@suse.cz>
13617 (chkp_narrow_size_and_offset): New function.
13618 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13619 (void chkp_parse_bit_field_ref): New function.
13620 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13621 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13623 2017-03-09 Martin Liska <mliska@suse.cz>
13626 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13627 (chkp_find_bounds_1): Remove gcc_unreachable.
13629 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13632 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13633 BUILT_IN_SYNC*, determine the access type from the size suffix and
13634 always build a MEM_REF with that type. Handle forgotten
13635 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13638 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13639 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13640 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13641 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13642 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13643 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13644 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13645 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13646 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13647 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13648 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13649 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13650 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13651 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13652 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13653 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13654 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13655 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13656 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13657 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13658 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13659 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13660 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13661 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13662 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13663 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13664 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13665 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13666 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13667 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13668 definitions outside of __OPTIMIZE__ guarded section.
13671 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13672 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13673 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13674 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13677 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13679 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13680 ("vfenez<mode>"): Add missing constraints.
13682 2017-03-08 Martin Sebor <msebor@redhat.com>
13685 * config/nds32/nds32.c (nds32_option_override):
13686 Fix misspelled diagnostic.
13688 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13691 * gimplify.c (gimplify_omp_for): Replace index var in outer
13692 taskloop statement with an artificial variable and add
13693 OMP_CLAUSE_PRIVATE clause for it.
13695 2017-03-08 Richard Biener <rguenther@suse.de>
13697 PR tree-optimization/79955
13698 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13699 for accesses that are completely outside of the variable.
13701 2017-03-08 Andrew Haley <aph@redhat.com>
13703 PR tree-optimization/79943
13704 * tree-ssa-loop-split.c (compute_new_first_bound): When
13705 calculating the new upper bound, (END-BEG) should be added, not
13708 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13710 * config/avr/avr.md (setmemhi): Make sure match_dup
13711 operand number comes before match_scratch.
13713 2017-03-08 Richard Biener <rguenther@suse.de>
13715 PR tree-optimization/79920
13716 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13717 with ncopies == 1 to ...
13718 (vect_transform_slp_perm_load): ... here. Properly compute
13719 all element loads by iterating VF times over the group. Do
13720 not handle ncopies (computed in a broken way) in
13721 vect_create_mask_and_perm.
13723 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13726 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
13727 is a uniform vector, use uniform_vector_p return value instead of
13728 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
13730 2017-03-07 Marek Polacek <polacek@redhat.com>
13732 PR middle-end/79809
13733 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
13734 (alloca_call_type): Likewise.
13736 2017-03-07 Martin Liska <mliska@suse.cz>
13738 * gcov.c (process_args): Put comment to correct location.
13740 2017-03-07 Martin Liska <mliska@suse.cz>
13742 PR middle-end/68270
13743 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
13744 Use array_at_struct_end_p instead of DECL_CHAIN (field).
13745 (chkp_narrow_bounds_for_field): Likewise.
13746 (chkp_parse_array_and_component_ref): Pass one more argument to
13749 2017-03-07 Richard Biener <rguenther@suse.de>
13751 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
13754 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
13756 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
13757 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
13759 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13762 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
13763 to end of description.
13764 (PARAM_MAX_STORES_TO_MERGE): Likewise.
13766 2017-03-07 Jakub Jelinek <jakub@redhat.com>
13768 PR rtl-optimization/79901
13769 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
13771 (*avx512f_<code><mode>3<mask_name>): ... this.
13772 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
13773 iterator instead of VI8_AVX2_AVX512BW.
13775 PR rtl-optimization/79901
13776 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
13777 min/max expander, expand it using expand_vec_cond_expr.
13780 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
13783 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13786 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
13787 to void * for PCH reasons.
13788 * dwarf2out.c (output_loc_operands, output_die): Cast
13789 v.val_vec.array to unsigned char *.
13791 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
13794 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
13797 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
13799 PR rtl-optimization/79571
13800 * lra-constraints.c (process_alt_operands): Calculate static
13801 reject and subtract it from overall when only addresses will be
13804 2017-03-06 Julia Koval <julia.koval@intel.com>
13807 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
13808 incoming stack boundary to 128 for 64-bit targets.
13810 2017-03-06 Richard Biener <rguenther@suse.de>
13812 PR tree-optimization/79894
13813 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
13814 to NULL after folding it.
13816 2017-03-06 Richard Biener <rguenther@suse.de>
13818 PR tree-optimization/79824
13819 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
13820 check disabling peeling for gaps.
13822 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
13824 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13825 attributes): Document gettimeofday.
13827 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13829 * config/s390/s390.c (s390_option_override_internal): Set
13830 PARAM_MIN_VECT_LOOP_BOUND
13832 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13834 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
13835 * config/s390/s390.md: Likewise.
13837 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13840 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
13841 (<avx2_avx512>_perm<mode>): Rename to ...
13842 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
13844 (<avx512>_perm<mode>_mask): Rename to ...
13845 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
13847 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
13848 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
13849 instead of VI8F_256_512.
13850 (avx512f_perm<mode>): New define_expand.
13851 (avx512f_perm<mode>_mask): Likewise.
13852 (avx512f_perm<mode>_1<mask_name>): New define_insn.
13853 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
13855 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13857 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
13858 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
13860 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
13862 2017-03-06 Martin Liska <mliska@suse.cz>
13865 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
13866 when having a SSA NAME w/o VAR_DECL assigned to it.
13868 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13870 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
13871 msa_dpsub_<su>_d): Fix MODE for vec_select.
13873 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13875 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
13877 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
13879 2017-03-06 Richard Biener <rguenther@suse.de>
13881 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
13882 * plugin.c (register_plugin_info): Likewise.
13883 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
13885 2017-03-05 Jakub Jelinek <jakub@redhat.com>
13887 * config/i386/sse.md (sse_storehps, sse_storelps,
13888 avx_<castmode><avxsizesuffix>_<castmode>,
13889 avx512f_<castmode><avxsizesuffix>_<castmode>,
13890 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
13891 in condition that at least one operand is not a MEM.
13893 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13895 PR middle-end/79805
13896 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
13897 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
13899 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
13900 gimple_call_nothrow_p flag based on whether original builtin can throw.
13901 If it can, emit following stmts on the fallthrough edge.
13902 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
13903 don't create new bb if inserting just debug stmts on the edge, try to
13904 insert them on the fallthru bb or just reset debug stmts.
13906 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
13909 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
13910 restore recog_data (including the operand rtxes inside it) around
13911 the call to get_insn_template.
13913 2017-03-03 Martin Sebor <msebor@redhat.com>
13915 PR tree-optimization/79699
13916 * context.c (context::~context): Free MPFR caches to avoid
13917 a memory leak on program exit.
13919 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13921 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
13922 Use wide_int::ulow () instead of .elt (0).
13924 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13926 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
13927 (*pushxf): Limit oF constraint to 32bit targets and add oC
13928 constraint for 64bit targets.
13929 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
13930 (*pushdf): Change rmF constraint to rmC.
13932 2017-03-03 Martin Liska <mliska@suse.cz>
13934 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
13935 Remove unused variable.
13937 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13940 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
13941 is a memory operand, increase num_memory.
13942 (ix86_expand_args_builtin): Likewise.
13944 2017-03-03 Jan Hubicka <jh@suse.cz>
13947 * ipa-devirt.c (maybe_record_node): Properly handle
13948 __cxa_pure_virtual visibility.
13950 2017-03-03 Martin Liska <mliska@suse.cz>
13952 PR tree-optimization/79803
13953 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
13955 (pass_loop_prefetch::execute): Disabled optimization if an
13956 assumption about L1 cache size is not met.
13958 2017-03-03 Martin Liska <mliska@suse.cz>
13960 PR rtl-optimization/79574
13961 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
13962 (hash_scan_set): Likewise.
13963 (dump_hash_table): Likewise.
13964 (hoist_code): Likewise.
13966 2017-03-03 Richard Biener <rguenther@suse.de>
13968 * fixed-value.c (fixed_from_string): Restore use of elt (1)
13969 in place of uhigh ().
13970 (fixed_convert_from_real): Likewise.
13972 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13975 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
13977 2017-03-03 Richard Biener <rguenther@suse.de>
13979 PR middle-end/79818
13980 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
13981 TYPE_OVERFLOW_UNDEFINED check.
13983 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13985 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
13987 (vector_ae_<mode>_p): Likewise.
13988 (vector_nez_<mode>_p): Likewise.
13989 (vector_ne_v2di_p): Likewise.
13990 (vector_ae_v2di_p): Likewise.
13991 (vector_ne_<mode>_p): Likewise.
13992 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
13994 (vsx_tsqrt<mode>2_fe): Likewise.
13996 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
13999 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14001 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14003 PR rtl-optimization/79780
14004 * cprop.c (one_cprop_pass): When second and further conditional trap
14005 in a single basic block is turned into an unconditional trap, turn it
14006 into a deleted note to avoid RTL verification failures.
14008 2017-03-02 Richard Biener <rguenther@suse.de>
14010 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14012 2017-03-02 Richard Biener <rguenther@suse.de>
14014 PR tree-optimization/79345
14016 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14017 param and abort the walk, returning -1 if it is hit.
14018 (walk_aliased_vdefs): Take a limit param and pass it on.
14019 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14020 defaulting to 0 and return a signed int.
14021 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14022 (check_defs): New helper.
14023 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14024 about uninitialized memory.
14025 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14026 bogus uninitialized warning.
14027 (fixed_convert_from_real): Likewise.
14029 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14031 PR tree-optimization/66768
14032 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14033 iv_use if base object can't be determined.
14035 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14037 PR tree-optimization/79345
14038 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14039 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14040 (get_pattern_stats): Initialize it.
14041 * genemit.c (gen_expand): Verify match_scratch numbers come after
14042 match_operand/match_dup numbers.
14043 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14044 match_scratch numbers.
14045 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14047 * config/s390/s390.md (trunctdsd2): Likewise.
14049 2017-03-02 Richard Biener <rguenther@suse.de>
14051 * wide-int.h (wide_int_storage::operator=): Implement in terms
14054 2017-03-02 Richard Biener <rguenther@suse.de>
14056 PR tree-optimization/79777
14057 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14058 the to insert expression to sth existing.
14060 2017-03-01 Martin Sebor <msebor@redhat.com>
14062 PR middle-end/79692
14063 * gimple-ssa-sprintf.c
14064 (directive::known_width_and_precision): New function.
14065 (format_integer): Use it.
14066 (get_mpfr_format_length): Consider the full range of precision
14067 when computing %g output with the # flag. Set the likely byte
14068 count to 3 rather than 1 when precision is indeterminate.
14069 (format_floating): Correct the lower bound of precision.
14071 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14073 * doc/invoke.texi: Document default code model for 64-bit Linux.
14075 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14078 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14079 udiv rather than div since input pattern is unsigned.
14081 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14083 * config/i386/i386.c (print_reg): Warn for values of
14084 unsupported size in integer register.
14086 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14089 * config/rs6000/predicates.md (current_file_function_operand): Do
14090 not allow self calls to be local if the function is replaceable.
14092 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14095 * config/rs6000/altivec.h (vec_ctz and others): Change the
14096 preprocessor macro that controls conditional compilation from
14097 _ARCH_PWR9 to __POWER9_VECTOR__.
14098 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14099 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14100 control (instead of _ARCH_PWR9 control) so that template
14101 definition uses power9-specific function.
14102 (vec_any_eq): Likewise.
14103 (vec_all_ne): Change macro definition to use a power9-specific
14104 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14105 _ARCH_PWR9 control).
14106 (vec_any_eq) Likewise.
14107 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14108 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14109 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14110 support for xvcmpnedp instruction.
14111 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14112 macro expansion so that Power9 implementation of vec_all_ne does
14113 not use the AltiVec predicate framework.
14114 (VCMPNEH_P): Likewise.
14115 (VCMPNEW_P): Likewise.
14116 (VCMPNED_P): Likewise.
14117 (VCMPNEFP_P): Likewise.
14118 (VCMPNEDP_P): Likewise.
14119 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14120 implementation of vec_any_eq to not use AltiVec predicate
14122 (VCMPAEH_P): Likewise.
14123 (VCMPAEW_P): Likewise.
14124 (VCMPAED_P): Likewise.
14125 (VCMPAEFP_P): Likewise.
14126 (VCMPAEDP_P): Likewise.
14127 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14128 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14129 not use the AltiVec predicate framework.
14130 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14131 of vec_any_eq to not use AltiVec predicate framework.
14132 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14133 support for predefined __POWER9_VECTOR__ macro to indicate that
14134 Power9 instruction selection is enabled.
14135 (altivec_overloaded_builtins): Remove extraneous
14136 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14137 function argument types RS6000_BTI_bool_V16QI and
14138 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14139 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14140 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14141 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14142 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14143 Power9 for implementations of vec_cmpne. Change the signature for
14144 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14145 (representing vec_all_ne) to remove the previously described first
14146 argument of type RS6000_BTI_INTSI, as this was an artifact of
14147 reliance on the AltiVec predicate framework, which is no longer
14148 used in the implementation of these functions. Add
14149 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14150 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14151 since, unlike the AltiVec predicate framework implementation, we
14152 do not share function descriptors between vec_alle and vec_anyeq.
14153 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14154 set of modes that receive special treatment even when
14155 TARGET_P9_VECTOR is true. The special treatment emits code that
14156 does not depend on Power9 instructions.
14157 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14158 define_expand to not rely on AltiVec predicate framework.
14159 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14161 (vector_ne_v2di_p): Change this define_expand to not rely on
14162 AltiVec predicate framework.
14163 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14165 (vector_ne_<mode>_p): Change this define_expand to not rely on
14166 AltiVec predicate framework.
14167 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14169 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14170 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14171 define_insn pattern.
14172 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14173 define_insn pattern because the xvcmpne<VSs>. instruction is not
14175 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14176 instruction is not supported.
14178 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14180 * config/nvptx/nvptx.c: Include intl.h.
14182 2017-03-01 Martin Jambor <mjambor@suse.cz>
14185 * ipa-prop.h (ipa_bits): Removed field known.
14186 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14187 to pointers. Adjusted their comments to warn about their sharing.
14188 (ipcp_transformation_summary): Change bits to a vector of pointers.
14189 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14190 (ipa_get_ipa_bits_for_value): Declare.
14191 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14192 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14193 (ipa_bits_hash_table): Likewise.
14194 (ipa_vr_ggc_hash_traits): Likewise.
14195 (ipa_vr_hash_table): Likewise.
14196 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14197 being pointers and vr_known being removed.
14198 (ipa_set_jf_unknown): Likewise.
14199 (ipa_get_ipa_bits_for_value): New function.
14200 (ipa_set_jfunc_bits): Likewise.
14201 (ipa_get_value_range): New overloaded functions.
14202 (ipa_set_jfunc_vr): Likewise.
14203 (ipa_compute_jump_functions_for_edge): Use the above functions to
14204 construct bits and vr parts of jump functions.
14205 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14206 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14208 (ipcp_grow_transformations_if_necessary): Also allocate
14209 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14211 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14212 them. Fix too long lines.
14213 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14214 vr_known being removed.
14215 (ipa_read_jump_function): Use new setter functions to construct bits
14216 and vr parts of jump functions or set them to NULL.
14217 (write_ipcp_transformation_info): Adjust for bits being pointers.
14218 (read_ipcp_transformation_info): Likewise.
14219 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14221 Include gt-ipa-prop.h.
14222 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14224 (ipcp_store_bits_results): Likewise.
14225 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14226 Do not write to existing jump functions but use a temporary instead.
14228 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14231 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14232 attempt to use its first operand as BIT_FIELD_REF base.
14234 2017-03-01 Richard Biener <rguenther@suse.de>
14236 PR middle-end/79721
14237 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14238 interpolating formula in wrapping arithmetic.
14239 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14241 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14243 PR tree-optimization/79734
14244 * tree-vect-generic.c (expand_vector_condition): Optimize
14245 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14246 Handle VEC_COND_EXPR where comparison has different inner width from
14247 type's inner width.
14249 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14251 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14252 markup, and similar issues. Remove @opindex entries for things
14253 that aren't options. Add missing -mmpy-option entries.
14255 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14257 PR tree-optimization/79737
14258 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14259 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14260 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14261 instead of byte_size. Formatting fix.
14262 (shift_bytes_in_array_right): Formatting fix.
14264 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14267 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14268 condition on optimize for the leaf function test.
14270 2017-02-28 Martin Liska <mliska@suse.cz>
14273 * read-rtl-function.c (function_reader::handle_unknown_directive):
14274 Bail out when one uses -flto.
14276 2017-02-28 Martin Liska <mliska@suse.cz>
14278 * common.opt: Replace space with tabular for options of <number>
14280 * config/i386/i386.opt: Show <number> value for
14281 -mlarge-data-threshold.
14282 * opts.c (print_filtered_help): Do not display number in hexadecimal
14285 2017-02-28 Martin Liska <mliska@suse.cz>
14287 * common.opt: Fix --help=option -Q for options which are of
14290 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14292 * config/i386/i386.c (print_reg): Error out for values
14293 of 8-bit size in invalid integer register.
14295 2017-02-28 Martin Sebor <msebor@redhat.com>
14297 PR tree-optimization/79691
14298 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14300 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14303 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14304 gcc_unreachable with output_operand_lossage.
14306 2017-02-28 Richard Biener <rguenther@suse.de>
14308 PR tree-optimization/79740
14309 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14311 (visit_nary_op): Insert the nary into the hashtable if we
14312 pattern-matched sth.
14313 * tree-ssa-pre.c (eliminate_insert): Robustify.
14315 2017-02-28 Richard Biener <rguenther@suse.de>
14317 PR middle-end/79731
14318 * fold-const.c (decode_field_reference): Reject out-of-bound
14321 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14323 * config/i386/i386.c: Include intl.h.
14324 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14325 instead of just cond ? "..." : "...".
14326 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14327 * coverage.c (read_counts_file): Likewise.
14328 * omp-offload.c: Include intl.h.
14329 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14330 of just cond ? "..." : "...".
14331 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14332 of just cond ? "..." : "...".
14334 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14337 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14339 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14340 'tune for' CPU name.
14341 * config/arm/arm-cpu-data.h: Regenerated.
14343 2017-02-28 Richard Biener <rguenther@suse.de>
14345 PR tree-optimization/79732
14346 * tree-inline.c (expand_call_inline): Do not shadow var.
14348 2017-02-28 Richard Biener <rguenther@suse.de>
14350 PR tree-optimization/79723
14351 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14352 address-space properly.
14354 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14356 * doc/optinfo.texi (Optimization groups): Fix option used for
14358 * doc/invoke.texi (-fopt-info): Document "omp".
14359 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14360 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14361 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14362 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14363 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14365 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14367 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14368 OPTGROUP_OMP with "omp".
14370 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14373 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14374 for arithmetic shift of unsigned V2DI.
14376 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14378 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14379 arc/linux.h headers.
14380 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14381 (LINK_SPEC): Likewise.
14382 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14383 (EXTRA_SPECS): Likewise.
14384 (STARTFILE_SPEC): Likewise.
14385 (ENDFILE_SPEC): Likewise.
14386 (LIB_SPEC): Likewise.
14387 (TARGET_SDATA_DEFAULT): Likewise.
14388 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14389 (MULTILIB_DEFAULTS): Likewise.
14390 (DWARF2_UNWIND_INFO): Likewise.
14391 * config/arc/big.h: New file.
14392 * config/arc/elf.h: Likewise.
14393 * config/arc/linux.h: Likewise.
14394 * config/arc/t-uClibc: Remove.
14396 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14398 PR tree-optimization/77536
14399 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14400 (tree_transform_and_unroll_loop): Use above function to compute the
14401 estimated niter of unrolled loop and use it when scaling profile.
14402 Also use count info rather than frequency if it's non-zero.
14403 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14404 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14405 (vect_transform_loop): Call above function.
14407 2017-02-27 Richard Biener <rguenther@suse.de>
14409 PR tree-optimization/45397
14410 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14411 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14412 (visit_nary_op): Add pattern matching for CSEing sign-changed
14413 or truncated operations with wider ones.
14415 2017-02-27 Richard Biener <rguenther@suse.de>
14417 PR tree-optimization/79690
14418 * tree-vect-stmts.c (vectorizable_store): Use vector type
14419 built from the DR with address-space.
14421 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14423 * doc/invoke.texi (Optimize Options): Refine the description
14424 of asan-use-after-return.
14426 2017-02-25 Alan Modra <amodra@gmail.com>
14428 PR rtl-optimization/79584
14429 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14430 base, not ad->base_term, the reg within base. Remove assertion
14431 that ad->base == ad->base_term. Replace gen_int_mode using
14432 bogus mode with const0_rtx.
14434 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14436 PR middle-end/79396
14437 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14438 FMA_EXPR like tcc_binary or tcc_unary.
14440 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14443 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14445 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14446 (output_loc_operands): Handle DW_OP_call_ref and
14447 DW_OP_GNU_variable_value.
14448 (struct variable_value_struct): New type.
14449 (struct variable_value_hasher): Likewise.
14450 (variable_value_hash): New variable.
14451 (string_types): Remove.
14452 (copy_loc_descr): New function.
14453 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14454 (prepend_loc_descr_to_each): New function.
14455 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14456 instead of add_loc_descr_to_each if the first argument is single
14457 location list and the second has multiple.
14458 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14459 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14460 when looking for variable value which doesn't have other location info.
14461 (loc_list_from_tree): Formatting fix.
14462 (gen_array_type_die): Simplify DW_AT_string_length handling.
14463 (adjust_string_types): Remove.
14464 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14465 string_types. Call resolve_variable_values.
14466 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14467 (resolve_addr_in_expr): Likewise. Add A argument.
14468 (copy_deref_exprloc): Remove deref argument. Adjust for the
14469 original expression being DW_OP_GNU_variable_value with optionally
14470 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14471 optionally after it.
14472 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14473 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14474 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14475 (variable_value_hasher::hash, variable_value_hasher::equal): New
14477 (resolve_variable_value_in_expr, resolve_variable_value,
14478 resolve_variable_values, note_variable_value_in_expr,
14479 note_variable_value): New functions.
14480 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14483 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14486 * opts.h (handle_generated_option): Add GENERATED_P argument.
14487 * opts-common.c (handle_option): Adjust function comment.
14488 (handle_generated_option): Add GENERATED_P argument, pass it to
14490 (control_warning_option): Pass false to handle_generated_option
14492 * opts.c (maybe_default_option): Pass true to handle_generated_option
14494 * optc-gen.awk: Likewise.
14496 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14498 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14499 a REG, look at the REG it is a SUBREG of.
14500 (splitter for cmpeqsi_t): Ditto.
14502 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14504 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14505 the special USEs with the pattern of the insn, not the insn itself.
14507 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14510 * doc/invoke.texi: Document -mload-store-pairs.
14512 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14513 Sandra Loosemore <sandra@codesourcery.com>
14515 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14516 argument isn't a CONST_INT.
14517 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14518 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14519 (nios2_validate_compare): Bypass alternate compare logic if *op2
14520 is not a CONST_INT.
14521 (ldstwm_operation_p): Return false if first_base is not a REG or
14522 if first_offset is not a CONST_INT.
14524 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14526 * config/cris/cris.md: Use correct operand in a define_peephole2.
14528 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14530 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14532 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14534 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14535 this_insn if it is an INSN or JUMP_INSN.
14536 (force_offsettable): Look at base, not at addr.
14537 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14538 on things that aren't necessarily CONST_INTs.
14540 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14542 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14543 -mfpmath=sse is the default also for x86-32 targets with SSE2
14544 instruction set when @option{-ffast-math} is enabled
14546 2017-02-24 Jeff Law <law@redhat.com>
14548 PR rtl-optimizatoin/79286
14549 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14552 2017-02-24 Richard Biener <rguenther@suse.de>
14554 PR tree-optimization/79389
14555 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14558 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14560 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14561 function comment to reflect reality.
14562 (loop_exits_before_overflow): Fix typo in function description.
14564 2017-02-24 Richard Biener <rguenther@suse.de>
14566 PR tree-optimization/79389
14567 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14568 properly that a threading opportunity exists. Detect conditional
14569 copy/constant propagation opportunities.
14571 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14573 * config/visium/visium.md (type): Add trap.
14574 (b): New mode attribute.
14575 (*btst): Rename into...
14576 (*btst<mode>): ...this and adjust.
14577 (*cbranchsi4_btst_insn): Rename into...
14578 (*cbranch<mode>4_btst_insn): ...this and adjust.
14579 (trap): New define_insn.
14581 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14583 PR tree-optimization/79389
14584 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14585 (noce_reversed_cond_code): New function.
14586 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14587 reversed_comparison_code. Formatting fix.
14588 (noce_try_store_flag): Test rev_cond != NULL in addition to
14589 reversed_comparison_code.
14590 (noce_try_store_flag_constants): Likewise.
14591 (noce_try_store_flag_mask): Likewise.
14592 (noce_try_addcc): Use rev_cond if non-NULL instead of
14593 reversed_comparison_code.
14594 (noce_try_cmove_arith): Likewise. Formatting fixes.
14595 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14596 (noce_find_if_block): Initialize rev_cond.
14597 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14598 instead of false as last argument never attempt to reverse it
14601 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14603 PR tree-optimization/79663
14604 * tree-predcom.c (combine_chains): Process refs in reverse order
14605 only for ZERO length chains, and add explaining comment.
14607 2017-02-23 Jeff Law <law@redhat.com>
14609 PR tree-optimization/79578
14610 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14611 in call to operand_equal_p.
14613 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14616 * config/i386/cpuid.h: Fix another undefined behavior.
14618 2017-02-23 Richard Biener <rguenther@suse.de>
14620 PR tree-optimization/79683
14621 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14622 vector types for data-refs.
14624 2017-02-23 Martin Liska <mliska@suse.cz>
14626 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14628 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14630 PR middle-end/79665
14631 * internal-fn.c (get_range_pos_neg): Moved to ...
14632 * tree.c (get_range_pos_neg): ... here. No longer static.
14633 * tree.h (get_range_pos_neg): New prototype.
14634 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14635 are known to be in between 0 and signed maximum inclusive, try to
14636 expand both unsigned and signed divmod and use the cheaper one from
14639 2017-02-22 Jeff Law <law@redhat.com>
14641 PR tree-optimization/79578
14642 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14643 to compare base operands.
14645 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14648 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14649 gpc_reg_operand instead of fpr_reg_operand.
14651 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14653 * config/mips/mips.c (mips_return_in_memory): Force FP
14654 vector types to be returned in memory for o32 ABI.
14656 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14658 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14659 instead of DW_TAG_member for static data member declarations and don't
14660 set no_linkage_name for static inline data members.
14661 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14664 2017-02-22 Martin Liska <mliska@suse.cz>
14666 * doc/invoke.texi: Replace inequality signs with square brackets
14669 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14671 PR tree-optimization/68644
14672 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14674 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14677 * lra-constraints.c (simplify_operand_subreg): Handle
14678 WORD_REGISTER_OPERATIONS targets.
14680 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14683 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14684 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14685 elimination by swapping fld*.
14687 2017-02-22 Richard Biener <rguenther@suse.de>
14689 PR tree-optimization/79673
14690 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14691 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14692 irrelevant address-space qualifiers and avoiding a
14693 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14695 2017-02-22 Richard Biener <rguenther@suse.de>
14697 PR tree-optimization/79666
14698 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14699 to not symbolically negate if that may introduce undefined
14702 2017-02-22 Martin Liska <mliska@suse.cz>
14705 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14706 * data-streamer-out.c (streamer_write_gcov_count_stream):
14708 * value-prof.c (stream_out_histogram_value): Make assert more
14709 precise based on type of counter.
14711 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14714 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14715 Use nonimmediate_operand instead of memory_operand for operand 1.
14716 (float-extend standard_x87sse_constant_load splitter): Ditto.
14718 2017-02-21 Jeff Law <law@redhat.com>
14720 PR tree-optimization/79621
14721 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
14722 blocks with edges to themselves.
14724 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14727 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
14728 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
14729 Use gimple_call_builtin_p.
14732 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
14733 on temporarily removed DEBUG_INSNs.
14735 PR tree-optimization/79649
14736 * tree-loop-distribution.c (classify_partition): Give up on
14737 non-generic address space loads/stores.
14739 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
14741 * doc/loop.texi (Loop manipulation): Remove nonexistent
14742 tree_ssa_loop_version from the documentation.
14743 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
14745 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14748 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
14749 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
14750 * config/rs6000/rs6000.c: Include except.h.
14751 (rs6000_expand_split_stack_prologue): Call
14752 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
14754 2017-02-21 Martin Jambor <mjambor@suse.cz>
14757 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
14758 have been analyzed.
14760 2017-02-21 Martin Jambor <mjambor@suse.cz>
14762 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
14763 for backward compatibility only.
14764 * doc/invoke.texi (Option Summary): Remove all references to
14765 -fipa-cp-alignment.
14767 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
14771 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14773 * lra-constraints.c (curr_insn_transform): Handle
14774 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14776 2017-02-21 Martin Liska <mliska@suse.cz>
14778 * config/i386/i386.opt: Replace -masm-dialect with -masm.
14780 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
14782 PR translation/79638
14783 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
14785 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
14788 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
14789 (arm_function_ok_for_sibcall): Return false for an indirect call by
14790 descriptor if all the argument registers are used.
14791 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
14792 alignment of the function.
14794 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14796 PR tree-optimization/61441
14797 * simplify-rtx.c (simplify_const_unary_operation): For
14798 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
14799 the sNaN unmodified.
14801 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14803 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14804 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14805 instead of SYSTEM_HEADER_DIR.
14807 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
14808 Martin Liška <mliska@suse.cz>
14810 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
14811 Fix typos and grammar, use active voice, and clarify.
14813 2017-02-20 Marek Polacek <polacek@redhat.com>
14815 PR middle-end/79537
14816 * gimplify.c (gimplify_expr): Handle unused *&&L;.
14819 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
14821 2017-02-20 Jakub Jelinek <jakub@redhat.com>
14824 * config/i386/i386.c (ix86_expand_builtin): Handle
14825 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
14826 ix86_builtins_isa[fcode].isa as a requirement of those
14827 flags and any other flag in the bitmask.
14828 (ix86_init_mmx_sse_builtins): Use 0 instead of
14829 ~OPTION_MASK_ISA_64BIT as mask.
14830 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
14831 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
14832 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
14833 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
14835 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14838 * lra-constraints.c (split_reg): Check requested split mode
14839 is supported by the register.
14841 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14843 * lra-constraints.c (simplify_operand_subreg): Remove early
14846 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14849 * lra-constraints.c (curr_insn_transform): Tighten condition
14850 for converting SUBREG reloads from OP_OUT to OP_INOUT.
14852 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14855 * lra-constraints.c (curr_insn_transform): Handle
14856 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14858 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
14861 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
14863 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
14865 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
14868 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14871 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14873 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
14876 2017-02-18 Jakub Jelinek <jakub@redhat.com>
14878 * final.c (last_columnnum, override_columnnum): New variables.
14879 (final_start_function): Set last_columnnum, pass it to begin_prologue
14880 hook and pass 0 to dwarf2out_begin_prologue.
14881 (final_scan_insn): Update override_columnnum. Pass last_columnnum
14882 to source_line debug hook.
14883 (notice_source_line): Compute last_columnnum and for debug_column_info
14884 return true on column changes.
14885 * debug.h (struct gcc_debug_hooks): Add column argument to
14886 source_line and begin_prologue hooks.
14887 (debug_nothing_int_charstar_int_bool): Remove prototype.
14888 (debug_nothing_int_int_charstar,
14889 debug_nothing_int_int_charstar_int_bool): New prototypes.
14890 (dwarf2out_begin_prologue): Add column argument.
14891 * debug.c (do_nothing_debug_hooks): Adjust source_line and
14892 begin_prologue hooks.
14893 (debug_nothing_int_charstar_int_bool): Remove.
14894 (debug_nothing_int_int_charstar,
14895 debug_nothing_int_int_charstar_int_bool): New functions.
14896 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
14897 through to dwarf2out_source_line.
14898 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
14899 (dwarf2out_source_line): Add column argument, emit it if requested.
14900 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
14902 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14903 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14904 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
14905 through to dwarf2out_begin_prologue.
14906 (vmsdbgout_source_line): Add column argument, pass it through to
14907 dwarf2out_source_line.
14908 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
14909 dbxout_source_line caller.
14910 (dbxout_source_line): Add column argument.
14912 * common.opt (gno-column-info, gcolumn-info): New options.
14913 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
14914 (check_die): Also test for multiple DW_AT_decl_column attributes.
14915 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
14916 DW_AT_decl_column if requested.
14917 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
14919 (gen_variable_die): Likewise.
14920 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
14921 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
14924 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
14925 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
14926 (ix86_handle_option): Handle OPT_m3dnowa.
14927 * doc/invoke.texi (-m3dnowa): Document.
14928 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
14929 -m3dnowa instead of -m3dnow -march=athlon.
14932 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
14933 instead of gcc_assert for K, r and R code checks. Formatting fixes.
14935 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14938 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
14939 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
14940 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
14941 generator for vsx_xxpermdi_<mode>_be.
14942 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
14943 force big-endian semantics.
14944 (vsx_xxpermdi_<mode>_be): New define_expand with same
14945 implementation as previous version of vsx_xxpermdi_<mode>.
14947 2017-02-17 Jakub Jelinek <jakub@redhat.com>
14949 PR tree-optimization/79327
14950 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
14951 variable, its initialization and use.
14953 2017-02-17 Julia Koval <julia.koval@intel.com>
14955 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
14956 (OPTION_MASK_ISA_PKU_UNSET): New.
14957 (ix86_handle_option): Handle -mrdpid.
14958 * config/i386/cpuid.h (bit_RDPID): New.
14959 * config/i386/driver-i386.c (host_detect_local_cpu):
14960 Detect RDPID feature.
14961 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
14962 * config/i386/i386-c.c (ix86_target_macros_internal):
14964 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
14965 (ix86_valid_target_attribute_inner_p): Add "rdpid".
14966 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14967 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
14968 * config/i386/i386.md (define_insn "rdpid"): New.
14969 * config/i386/i386.opt Add -mrdpid.
14970 * config/i386/immintrin.h (_rdpid_u32): New.
14972 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
14974 PR rtl-optimization/79541
14975 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
14976 instead of transforming it into USE.
14978 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
14980 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
14981 If HONOR_SNANS (SFmode) force the input to a register.
14982 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
14983 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
14984 an frsp or similar insn.
14986 2017-02-17 Martin Liska <mliska@suse.cz>
14988 PR rtl-optimization/79577
14989 * params.def (selsched-max-sched-times): Increase minimum to 1.
14991 2017-02-17 Martin Liska <mliska@suse.cz>
14993 PR rtl-optimization/79574
14994 * gcse.c (want_to_gcse_p): Prevent integer overflow.
14996 2017-02-17 Martin Liska <mliska@suse.cz>
14998 PR tree-optimization/79529
14999 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15000 ssa_defined_default_def_p to handle cases which are implicitly
15002 * tree-ssa.c (ssa_defined_default_def_p): New function.
15003 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15004 which are implicitly defined.
15005 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15007 2017-02-17 Richard Biener <rguenther@suse.de>
15009 PR middle-end/79576
15010 * params.def (max-ssa-name-query-depth): Limit to 10.
15012 2017-02-17 Richard Biener <rguenther@suse.de>
15014 PR tree-optimization/79552
15015 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15018 2017-02-17 Richard Biener <rguenther@suse.de>
15021 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15023 2017-02-17 Marek Polacek <polacek@redhat.com>
15025 PR middle-end/79536
15026 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15027 (fold_negate_expr): New wrapper.
15029 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15031 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15032 Correct terminology and de-emphasize pre-standard behavior.
15034 2017-02-16 Alan Modra <amodra@gmail.com>
15036 PR rtl-optimization/79286
15037 * ira.c (def_dominates_uses): New function.
15038 (update_equiv_regs): Don't create an equivalence for insns that
15039 may trap where the register def does not dominate the use.
15041 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15043 PR rtl-optimization/78127
15044 * lra.c (lra): Call lra_eliminate before finish the loop after
15047 2017-02-16 Richard Biener <rguenther@suse.de>
15049 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15051 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15052 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15053 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15054 (isl_val_int_from_wi): New function.
15055 (extract_affine_gmp): Rename to ...
15056 (extract_affine_wi): ... this, take a widest_int.
15057 (extract_affine_int): Just wrap extract_affine_wi.
15058 (add_param_constraints): Use isl_val_int_from_wi.
15059 (add_loop_constraints): Likewise, and extract_affine_wi.
15061 2017-02-15 Jeff Law <law@redhat.com>
15063 PR middle-end/79521
15064 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15065 ira_init_register_move_cost_if_necessary.
15067 2017-02-15 Martin Sebor <msebor@redhat.com>
15069 PR middle-end/32003
15070 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15071 removed in a prior commit.
15073 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15075 PR tree-optimization/79347
15076 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15077 counters during peeling.
15079 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15081 * Makefile.in (site.exp): Remove "set ISLVER".
15083 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15086 * real.c (real_from_integer): Call real_convert even for decimal.
15088 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15091 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15093 2017-02-14 Andrew Pinski <apinski@cavium.com>
15095 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15096 cores and change the partno/implementer to be correct.
15097 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15098 the 'B" as the implementer.
15099 * config/aarch64/aarch64-tune.md: Regenerate.
15101 2017-02-14 Carl Love <cel@us.ibm.com>
15103 * config/rs6000/rs6000.c: Add case statement entry to make the
15104 xvcvuxdsp built-in argument unsigned.
15105 * config/rs6000/vsx.md: Fix the source and return operand types so they
15106 match the instruction definitions from the ISA document. Fix typo
15107 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15110 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15113 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15114 member early_clobber_alts.
15115 * lra-lives.c (reg_early_clobber_p): New.
15116 (process_bb_lives): Use it.
15117 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15118 (debug_operand_data): Initialize early_clobber_alts.
15119 (setup_operand_alternative): Set up early_clobber_alts.
15120 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15121 alternatives to new_insn_reg.
15122 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15124 (lra_update_insn_regno_info): Pass the new arg.
15126 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15128 PR middle-end/79505
15129 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15130 (new_oacc_loop_raw): Don't clear already cleared fields.
15133 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15134 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15135 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15137 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15140 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15142 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15145 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15146 the extra instruction to the right place to store 128-bit constant
15149 2017-02-14 Martin Sebor <msebor@redhat.com>
15151 PR middle-end/79448
15152 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15153 warning for strings of unknown length.
15155 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15157 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15159 2017-02-14 Jeff Law <law@redhat.com>
15162 * ira-costs.c (scan_one_insn): Initialize register move costs
15163 for pseudos seen in USE/CLOBBER insns.
15165 PR tree-optimization/79095
15166 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15167 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15168 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15169 if the operands are known to be not equal, then the resulting range
15171 (intersect_ranges): If the new range is ~[0,0] and the old range is
15172 wide, then prefer ~[0,0].
15173 * tree-vrp.c (overflow_comparison_p_1): New function.
15174 (overflow_comparison_p): New function.
15175 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15176 if NAME is used in an overflow test.
15177 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15178 overflow check that can be expressed as an equality test, then adjust
15179 ops to be that equality test.
15181 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15183 * config/s390/s390-builtin-types.def: Remove flags argument.
15184 * config/s390/s390.c (s390_init_builtins): Likewise.
15186 2017-02-14 Martin Liska <mliska@suse.cz>
15188 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15189 vector. Fix trailing white spaces.
15191 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15193 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15196 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15198 PR rtl-optimization/68664
15199 * config/arm/arm.c (arm_sched_can_speculate_insn):
15200 New function. Declare prototype.
15201 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15203 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15205 PR rtl-optimization/68664
15206 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15208 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15210 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15212 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15213 max skip bytes for function, loop and jump.
15215 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15217 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15218 ABS_EXPR for gimple dump.
15220 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15223 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15225 PR tree-optimization/79408
15226 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15227 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15228 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15229 also if rhs1 is INTEGER_CST.
15231 2017-02-14 Richard Biener <rguenther@suse.de>
15233 PR middle-end/79432
15234 * tree-into-ssa.c (insert_phi_nodes): When the function can
15235 have abnormal edges rewrite SSA names with broken use-def
15236 dominance out of SSA and register them for PHI insertion.
15238 2017-02-13 Martin Sebor <msebor@redhat.com>
15240 PR middle-end/79496
15241 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15242 clearing info.nowrite flag when snprintf size argument is a range.
15244 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15246 * cprop.c (cprop_jump): Add missing space in string literal.
15247 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15248 (get_constraint_for_component_ref): Likewise.
15249 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15250 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15251 * lra-constraints.c (process_alt_operands): Likewise.
15252 * ipa-inline.c (inline_small_functions): Likewise.
15253 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15254 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15255 * trans-mem.c (diagnose_tm_1_op): Likewise.
15256 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15257 (grid_parallel_clauses_gridifiable): Likewise.
15259 * config/nvptx/mkoffload.c (process): Add space in between
15262 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15263 "MOD4_SSE_REGS" and "ALL_REGS".
15265 * spellcheck.c (test_data): Add , in between "foo" and "food".
15267 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15270 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15271 boundary crossing check and subsequent code generation agree.
15273 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15275 * config/aarch64/aarch64.c (has_memory_op): Delete.
15276 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15279 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15281 PR rtl-optimization/79388
15282 PR rtl-optimization/79450
15283 * combine.c (distribute_notes): When removing TEM_INSN for which
15284 corresponding dest has last value recorded, invalidate that last
15287 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15289 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15290 of explicit '@'. Add missing assembly comment marker on branch costs
15293 2017-02-13 Nathan Sidwell <nathan@acm.org>
15295 * gengtype-lex.l (<in_struct>): Add '/'.
15297 2017-02-13 Martin Liska <mliska@suse.cz>
15300 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15302 2017-02-13 Richard Biener <rguenther@suse.de>
15304 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15306 * configure: Re-generate.
15307 * config.in: Likewise.
15308 * graphite-dependences.c: Simplify as if
15309 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15310 * graphite-isl-ast-to-gimple.c: Likewise.
15311 * graphite-optimize-isl.c: Likewise.
15312 * graphite-poly.c: Likewise.
15313 * graphite-sese-to-poly.c: Likewise.
15314 * graphite.h: Likewise.
15315 * toplev.c: Include isl/version.h and use isl_version () for
15316 printing the ISL version.
15317 * doc/install.texi: Update ISL requirement.
15319 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15321 * doc/standards.texi (Standards): Update reference to
15324 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15326 * doc/extend.texi (Named Address Spaces): sourceware.org now
15328 * doc/install.texi (Binaries): Ditto.
15331 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15333 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15334 standard terminology "stringize"/"stringizing" throughout.
15335 * doc/cppinternals.texi: Likewise.
15337 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15339 * doc/extend.texi: Fix some spelling mistakes and typos.
15340 * doc/invoke.texi: Likewise.
15342 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15345 * params.def (inline-min-speedup) Change from 10 to 8.
15347 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15349 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15352 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15355 * ipa-inline-analysis.c (get_minimal_bb): New function.
15356 (record_modified): Use it.
15357 (remap_edge_change_prob): Handle also ancestor functions.
15359 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15361 * doc/contrib.texi (Contributors): Remove broken link into
15362 the Mauve CVS repository.
15364 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15366 PR middle-end/79454
15367 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15368 result computation whenever lhs doesn't have vector mode, not
15369 just when it has BLKmode.
15371 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15373 * doc/makefile.texi (profiledbootstrap): Refer to the
15374 installation instructions only in textual form.
15376 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15379 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15381 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15383 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15384 (Specific): Update mingw-w64 reference.
15386 (Specific): Remove broken link to Renesas RX processor.
15388 2017-02-10 Richard Biener <rguenther@suse.de>
15390 * toplev.c (process_options): Do not mention obsolete graphite
15391 options when printing sorry message about missing graphite support.
15392 Mention -floop-nest-optimize.
15394 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15396 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15397 (vtst_p16): Likewise.
15398 (vtstq_p8): Likewise.
15399 (vtstq_p16): Likewise.
15401 (vtstq_p64): Likewise.
15402 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15403 (vset_lane_p64): New.
15404 (vsetq_lane_p64): New.
15406 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15408 PR tree-optimization/79411
15409 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15410 stmt operands are SSA_NAMEs used in abnormal phis.
15411 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15414 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15417 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15420 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15422 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15424 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15426 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15430 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15431 not arbitrary TREE_CONSTANT.
15434 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15435 "omp declare target link" attribute unless is_global_var.
15436 * omp-offload.c (find_link_var_op): Likewise.
15438 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15439 Chung-Lin Tang <cltang@codesourcery.com>
15441 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15443 (gimplify_adjust_omp_clauses): Don't delete TILE.
15444 (gimplify_omp_for): Deal with TILE.
15445 * internal-fn.c (expand_GOACC_TILE): New function.
15446 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15448 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15449 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15451 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15452 avoid DIV for outermost collapse var.
15453 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15454 Remove out of date comments, fix whitespace.
15455 * omp-general.c (omp_extract_for_data): Deal with tiling.
15456 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15457 adjust OLF_DIM_BASE value.
15458 (struct omp_for_data): Add tiling field.
15459 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15460 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15461 for auto loops. Remove default auto determining, moved to
15462 oacc_loop_fixed_partitions.
15463 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15464 stmts, add e_mask field.
15465 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15466 (oacc_thread_numbers): Use oacc_dim_call.
15467 (oacc_xform_tile): New.
15468 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15469 (finish_oacc_loop): Adjust for ifns vector.
15470 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15471 add case for GOACC_TILE fns.
15472 (oacc_loop_xform_loop): Delete.
15473 (oacc_loop_process): Iterate over call list directly, and add
15474 handling for GOACC_TILE fns.
15475 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15477 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15478 vector partitioning to outer loops. Assign 2 partitions to loops
15479 when available. Add TILE handling.
15480 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15481 (execite_oacc_device_lower): Process GOACC_TILE fns,
15482 ignore unknown specs.
15483 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15484 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15485 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15487 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15489 * configure.ac (ACX_BUGURL): Update.
15490 * configure: Regenerate.
15492 2017-02-09 Richard Biener <rguenther@suse.de>
15494 PR tree-optimization/69823
15495 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15496 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15498 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15500 * config/arc/arc-c.def: Add __NPS400__ definition.
15501 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15502 (TARGET_NPS400): Define.
15504 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15506 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15508 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15509 pointer, arch_info.
15510 (arc_cpu_types): Fill the arch_info field with a pointer into the
15511 arc_arch_types table.
15512 (arc_selected_cpu): Declare.
15513 * config/arc/arc.c (arc_selected_cpu): Make global.
15514 (arc_selected_arch): Delete.
15515 (arc_base_cpu): Delete.
15516 (arc_override_options): Remove references to deleted variables,
15517 update access to arch information.
15518 (ARC_OPT): Update access to arch information.
15519 (ARC_OPTX): Likewise.
15520 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15521 (TARGET_ARC600): Update access to arch information.
15522 (TARGET_ARC601): Likewise.
15523 (TARGET_ARC700): Likewise.
15524 (TARGET_EM): Likewise.
15525 (TARGET_HS): Likewise.
15526 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15529 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15532 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15533 condition/operands for integer GE/LE/GEU/LEU operations.
15535 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15537 PR translation/79397
15538 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15541 2017-02-08 Martin Jambor <mjambor@suse.cz>
15544 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15545 whether allocation happened.
15546 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15547 nothing was allocated.
15549 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15551 PR tree-optimization/79408
15552 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15553 constant, but SSA_NAME with a known integer range, use the minimum
15554 of that range instead of op1 to determine if modulo can be replaced
15555 with its first operand.
15557 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15559 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15561 2017-02-08 Richard Biener <rguenther@suse.de>
15563 PR tree-optimization/71824
15564 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15565 Check all loops contained in the merged region.
15567 2017-02-07 Andrew Pinski <apinski@cavium.com>
15569 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15571 2017-02-07 Andrew Pinski <apinski@cavium.com>
15573 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15574 (thunderxt88): Likewise.
15575 (thunderxt81): Disable LSE and change v8.1 to v8.
15576 (thunderxt83): Likewise.
15578 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15579 Richard Biener <rguenther@suse.de>
15581 PR middle-end/79399
15582 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15583 type from int to size_t.
15584 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15586 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15588 PR rtl-optimization/79386
15589 * cprop.c (bypass_conditional_jumps): Initialize
15590 bypass_last_basic_block already before splitting bbs after
15591 unconditional traps...
15592 (bypass_conditional_jumps): ... rather than here.
15595 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15596 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15597 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15598 fix -masm=intel patterns.
15600 2017-02-07 Richard Biener <rguenther@suse.de>
15602 PR tree-optimization/79256
15603 PR middle-end/79278
15604 * builtins.c (get_object_alignment_2): Use min_align_of_type
15605 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15606 and ADJUST_FIELD_ALIGN.
15608 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15610 * doc/tm.texi: Regenerate.
15611 * stor-layout.c (layout_decl): Adjust.
15612 (update_alignment_for_field): Likewise.
15613 (place_field): Likewise.
15614 (min_align_of_type): Likewise.
15615 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15616 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15617 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15618 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15619 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15620 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15621 * config/i386/i386.c (x86_field_alignment): Likewise.
15622 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15623 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15624 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15625 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15626 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15627 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15631 2017-01-30 Richard Biener <rguenther@suse.de>
15633 PR tree-optimization/79256
15634 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15635 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15638 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15640 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15641 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15642 builtins to SImode and emit a zero-extend, if necessary.
15644 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15646 * docs/invoke.texi (RISC-V Options): Alphabetize.
15648 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15650 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15653 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15655 * config/riscv/riscv.c: New file.
15656 * common/config/riscv/riscv-common.c: Likewise.
15657 * config.gcc: Likewise.
15658 * config/riscv/constraints.md: Likewise.
15659 * config/riscv/elf.h: Likewise.
15660 * config/riscv/generic.md: Likewise.
15661 * config/riscv/linux.h: Likewise.
15662 * config/riscv/multilib-generator: Likewise.
15663 * config/riscv/peephole.md: Likewise.
15664 * config/riscv/pic.md: Likewise.
15665 * config/riscv/predicates.md: Likewise.
15666 * config/riscv/riscv-builtins.c: Likewise.
15667 * config/riscv/riscv-c.c: Likewise.
15668 * config/riscv/riscv-ftypes.def: Likewise.
15669 * config/riscv/riscv-modes.def: Likewise.
15670 * config/riscv/riscv-opts.h: Likewise.
15671 * config/riscv/riscv-protos.h: Likewise.
15672 * config/riscv/riscv.h: Likewise.
15673 * config/riscv/riscv.md: Likewise.
15674 * config/riscv/riscv.opt: Likewise.
15675 * config/riscv/sync.md: Likewise.
15676 * config/riscv/t-elf-multilib: Likewise.
15677 * config/riscv/t-linux: Likewise.
15678 * config/riscv/t-linux-multilib: Likewise.
15679 * config/riscv/t-riscv: Likewise.
15680 * configure.ac: Likewise.
15681 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15682 Waterman as RISC-V maintainers.
15683 * doc/install.texi: Add RISC-V entries.
15684 * doc/invoke.texi: Add RISC-V options section.
15685 * doc/md.texi: Add RISC-V constraints section.
15686 * configure: Regenerated.
15688 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15691 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15692 false values to be constant vectors with all 0 or all 1 bits set.
15693 (vcondu<mode><mode>): Likewise.
15694 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15696 (fpmask_comparison_operator): Update comment.
15697 (vecint_comparison_operator): New predicate.
15698 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15699 vector conditionals when the true and false values are constant
15700 vectors with all 0 bits or all 1 bits set.
15702 2017-02-06 Martin Sebor <msebor@redhat.com>
15704 PR tree-optimization/79376
15705 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15707 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15709 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15710 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15711 to simplify split condition.
15713 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15715 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15716 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15719 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
15721 PR rtl-optimization/68664
15722 * target.def (can_speculate_insn): New hook.
15723 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
15724 * doc/tm.texi: Regenerate.
15725 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
15726 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
15727 (rs6000_sched_can_speculate_insn): New function.
15729 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15731 PR tree-optimization/79284
15732 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
15733 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
15734 vectorizable_mask_load_store, vectorizable_operation,
15735 vect_is_simple_cond, get_same_sized_vectype): Use it instead
15736 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
15737 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
15738 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
15739 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15740 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15741 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
15742 is_gimple_assign (stmt). Replace another such test with
15743 is_gimple_assign (stmt).
15745 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15748 * config/avr/avr.c (rtl-iter.h): Include it.
15749 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
15750 (avr_legitimate_combined_insn): ...and implementation.
15752 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
15754 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
15755 * config/s390/s390.c (s390_const_operand_ok)
15756 (s390_canonicalize_comparison, s390_extract_part)
15757 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
15758 (s390_contiguous_bitmask_p, s390_rtx_costs)
15759 (legitimize_pic_address): Likewise.
15760 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
15761 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
15762 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
15763 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
15764 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
15766 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15768 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
15769 REGNO($0) == REGNO($1).
15771 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15773 * config/s390/linux.h(SIZE_TYPE): Add comment.
15775 2017-02-06 Julian Brown <julian@codesourcery.com>
15776 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15777 Virendra Pathak <virendra.pathak@broadcom.com>
15779 * config/aarch64/aarch64-cores.def: Change the scheduler
15781 * config/aarch64/aarch64.md: Include thunderx2t99.md.
15782 * config/aarch64/thunderx2t99.md: New file.
15784 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15786 * doc/standards.texi (Go Language): Update link to language
15789 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15791 * tree-eh.c (lower_resx): Sanitize profile.
15792 (cleanup_empty_eh_move_lp): Likewise.
15794 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15797 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
15799 * cfgloopmanip.h (loop_version): Update prototype.
15800 * modulo-sched.c (sms_schedule): Update call of loop_version.
15801 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
15802 * tree-parloops.c (gen_parallel_loop): Likewise.
15803 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
15804 * tree-ssa-loop-split.c (split_loop): Likewise.
15805 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
15806 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15808 2017-02-05 Martin Liska <mliska@suse.cz>
15811 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
15813 (print_operand_address): Initialize a struct to zero.
15815 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15817 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
15818 garbage collector only in textual form.
15820 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15822 * doc/extend.texi (x86 specific memory model extensions for
15823 transactional memory): Simplify a phrase.
15825 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
15828 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
15829 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
15830 (atomic_storedi_1): Likewise.
15832 2017-02-04 Jakub Jelinek <jakub@redhat.com>
15834 PR tree-optimization/79338
15835 * tree-parloops.c (gather_scalar_reductions): Don't call
15836 vect_analyze_loop_form for loop->inner before destroying loop's
15839 2017-02-03 Martin Sebor <msebor@redhat.com>
15841 PR tree-optimization/79327
15842 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
15843 when precision has resulted in leading zeros.
15844 (format_integer): Adjust the likely counter to assume an unknown
15845 argument that may be zero is non-zero.
15847 2017-02-03 Jason Merrill <jason@redhat.com>
15850 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
15851 avoid copying non-taken branch.
15853 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15855 PR tree-optimization/79340
15856 * tree-vect-loop.c (vectorizable_reduction): Release
15857 vec_defs elements after safe_splicing them into other vectors.
15860 PR tree-optimization/79327
15861 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
15862 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
15864 (format_integer): Use wide_int_to_tree instead of build_int_cst
15865 + to_?hwi. If argmin is NULL, just set argmin and argmax to
15866 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
15867 of shortest and longest sequence.
15869 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
15871 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
15872 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
15874 2017-02-03 Walter Lee <walt@tilera.com>
15877 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
15878 after initial stackframe link reg save.
15879 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
15881 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15884 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
15885 wu for stxssp alternative.
15887 2017-02-03 Martin Sebor <msebor@redhat.com>
15889 PR tree-optimization/79352
15890 * gimple-fold.c (get_range_strlen): Add argument.
15891 (get_range_strlen): Change return type to bool.
15892 (get_maxval_strlen): Pass in a dummy argument.
15893 * gimple-fold.h (get_range_strlen): Change return type to bool.
15894 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
15895 * tree.h (array_at_struct_end_p): Add argument.
15896 * tree.c (array_at_struct_end_p): Handle it.
15898 2017-02-03 Martin Liska <mliska@suse.cz>
15901 * multiple_target.c (create_dispatcher_calls): Redirect edge
15902 from a caller of a dispatcher.
15903 (expand_target_clones): Make the clones local.
15904 (ipa_target_clone): Do both target clones and resolvers.
15905 (ipa_dispatcher_calls): Remove the pass.
15906 (pass_dispatcher_calls::gate): Likewise.
15907 (make_pass_dispatcher_calls): Likewise.
15908 * passes.def (pass_target_clone): Put as very first IPA early
15911 2017-02-03 Martin Liska <mliska@suse.cz>
15913 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
15914 in case of a function with ifunc attribute.
15916 2017-02-03 Martin Liska <mliska@suse.cz>
15918 * cgraph.c (cgraph_node::dump): Dump function version info.
15919 * symtab.c (symtab_node::dump_base): Add missing new line.
15921 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15923 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
15924 (ifcombine_ifandif): Use it.
15926 2017-02-03 Martin Liska <mliska@suse.cz>
15928 * doc/invoke.texi: Document default value for
15929 use-after-scope-direct-emission-threshold.
15931 2017-02-03 Martin Liska <mliska@suse.cz>
15933 PR tree-optimization/79339
15934 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
15935 (format_floating): Likewise.
15937 2017-02-03 Martin Liska <mliska@suse.cz>
15940 * ipa-prop.c (ipa_node_params_t::insert): Remove current
15942 (ipa_node_params_t::remove): Likewise.
15943 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
15944 initialization from removed ipa_node_params_t::insert.
15945 (ipa_node_params::~ipa_node_params): Move from removed
15946 ipa_node_params_t::release.
15947 * symbol-summary.h (symbol_summary::m_released): New member.
15948 Do not release a summary twice. Do not allow to call finalizer
15949 for types of a summary that live in GGC memory.
15951 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15953 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
15956 2017-02-02 Martin Sebor <msebor@redhat.com>
15958 PR middle-end/79275
15959 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
15960 (format_string): Tighten up the range of output for non-constant
15961 strings and correct the expected range for wide non-constant strings.
15963 2017-02-02 Martin Sebor <msebor@redhat.com>
15965 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
15967 PR middle-end/32003
15968 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
15970 (-fdump-tree-@var): Add to index and document how to come up
15971 with pass-specific option and dump file names.
15972 (-fdump-passes): Clarify where to look for output.
15974 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15976 PR middle-end/77445
15977 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
15978 statistics of the analyzed path; allow threading for speed when
15979 any of BBs along the path are optimized for speed.
15981 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
15983 PR middle-end/78468
15984 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
15985 settings of the virtual registers.
15988 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
15990 * explow.c (get_dynamic_stack_size): Take known alignment of stack
15991 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
15994 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15996 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
15997 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
15999 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16001 * config/s390/s390.md: Add missing comments with the expanded
16003 * config/s390/vector.md: Likewise.
16004 * config/s390/vx-builtins.md: Likewise.
16006 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16009 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16010 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16011 conditions on a single line.
16013 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16015 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16016 __S390_VX__ to __VX__.
16018 2017-02-01 Andrew Pinski <apinski@cavium.com>
16020 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16021 stmt_info to record_stmt_cost.
16022 (vect_get_known_peeling_cost): Pass stmt_info if known to
16024 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16025 cpu_vector_cost field into
16026 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16027 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16028 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16029 splitting of scalar_stmt_cost and vec_stmt_cost.
16030 (thunderx_vector_cost): Likewise.
16031 (cortexa57_vector_cost): LIkewise.
16032 (exynosm1_vector_cost): Likewise.
16033 (xgene1_vector_cost): Likewise.
16034 (thunderx2t99_vector_cost): Improve after the splitting of the two
16036 (aarch64_builtin_vectorization_cost): Update for the splitting of
16037 scalar_stmt_cost and vec_stmt_cost.
16039 2017-02-01 Torvald Riegel <triegel@redhat.com>
16040 Richard Henderson <rth@redhat.com>
16042 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16043 conditional on existance of a fast atomic load.
16044 * optabs-query.c (can_atomic_load_p): New function.
16045 * optabs-query.h (can_atomic_load_p): Declare it.
16046 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16047 no fast atomic load is available for the particular size of access.
16048 (expand_atomic_compare_and_swap): Likewise.
16049 (expand_atomic_load): Likewise.
16050 (expand_atomic_store): Likewise.
16051 (expand_atomic_fetch_op): Likewise.
16052 * testsuite/lib/target-supports.exp
16053 (check_effective_target_sync_int_128): Remove x86 because it provides
16054 no fast atomic load.
16055 (check_effective_target_sync_int_128_runtime): Likewise.
16057 2017-02-01 Richard Biener <rguenther@suse.de>
16059 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16060 (graphite_transform_loops): Provide opt-info for optimized nests.
16061 * tree-parloop.c (parallelize_loops): Provide opt-info for
16062 parallelized loops.
16064 2017-02-01 Richard Biener <rguenther@suse.de>
16066 PR middle-end/79315
16067 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16068 was not set before.
16070 2017-02-01 Richard Biener <rguenther@suse.de>
16072 PR tree-optimization/71824
16073 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16074 Verify the loops are valid in the merged SESE region.
16075 (scop_detection::can_represent_loop_1): Check analyzing the
16076 evolution of the number of iterations in the region succeeds.
16078 2017-01-31 Ian Lance Taylor <iant@golang.org>
16080 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16081 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16083 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16085 PR preprocessor/79210
16086 * input.c (get_substring_ranges_for_loc): Replace line_width
16087 assertion with error-handling.
16089 2017-01-31 Richard Biener <rguenther@suse.de>
16091 PR tree-optimization/77318
16092 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16093 (create_pw_aff_from_tree): Take loop parameter.
16094 (add_condition_to_pbb): Pass loop of the condition to
16095 create_pw_aff_from_tree.
16097 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16099 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16100 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16102 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16106 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16108 (convert_int_to_float128): Likewise.
16109 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16110 (convert_int_to_float128): Likewise.
16111 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16112 (UNSPEC_IEEE128_CONVERT): Likewise.
16113 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16114 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16115 Use local variables for IBM extended format.
16116 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16117 (fix_trunc<mode>si2_fprs): Likewise.
16118 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16119 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16120 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16121 to know that we can now have integers of all sizes in vector
16123 (fix<uns>_<mode>di2_hw): Likewise.
16124 (float<uns>_<mode>si2_hw): Likewise.
16125 (fix_<mode>si2_hw): Likewise.
16126 (fixuns_<mode>si2_hw): Likewise.
16127 (float<uns>_<mode>di2_hw): Likewise.
16128 (float_<mode>di2_hw): Likewise.
16129 (float_<mode>si2_hw): Likewise.
16130 (floatuns_<mode>di2_hw): Likewise.
16131 (floatuns_<mode>si2_hw): Likewise.
16132 (xscvqp<su>wz_<mode>): Delete, no longer used.
16133 (xscvqp<su>dz_<mode>): Likewise.
16134 (xscv<su>dqp_<mode>): Likewise.
16135 (ieee128_mfvsrd_64bit): Likewise.
16136 (ieee128_mfvsrd_32bit): Likewise.
16137 (ieee128_mfvsrwz): Likewise.
16138 (ieee128_mtvsrw): Likewise.
16139 (ieee128_mtvsrd_64bit): Likewise.
16140 (ieee128_mtvsrd_32bit): Likewise.
16142 2017-01-31 Martin Liska <mliska@suse.cz>
16145 * ipa-prop.c (ipa_free_all_node_params): Call release method
16146 instead of ~sumbol_summary to not to trigger double times
16149 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16151 PR tree-optimization/71691
16152 * bitmap.h (class auto_bitmap): New.
16153 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16154 is_maybe_undefined instead of ssa_undefined_value_p.
16156 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16158 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16159 __S390_ARCH_LEVEL__ to __ARCH__.
16161 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16163 PR tree-optimization/79267
16164 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16165 if should_remove_lhs_p is true.
16167 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16170 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16171 (add_alignment_attribute): New.
16172 (base_type_die): Add alignment attribute.
16173 (subrange_type_die): Likewise.
16174 (modified_type_die): Likewise.
16175 (gen_array_type_die): Likewise.
16176 (gen_descr_array_type_die: Likewise.
16177 (gen_enumeration_type_die): Likewise.
16178 (gen_subprogram_die): Likewise.
16179 (gen_variable_die): Likewise.
16180 (gen_field_die): Likewise.
16181 (gen_ptr_to_mbr_type_die): Likewise.
16182 (gen_struct_or_union_type_die): Likewise.
16183 (gen_subroutine_type_die): Likewise.
16184 (gen_typedef_die): Likewise.
16185 (base_type_cmp): Compare alignment attribute.
16187 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16190 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16191 (setb_unsigned) New pattern for setb with CCUNS.
16192 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16193 subfc./subfe sequence to avoid overflow problems. Generate a
16194 shorter sequence with cmpld/setb for power9.
16195 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16196 for generating subfc. instruction.
16197 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16198 now uses this instruction.
16200 2017-01-30 Ian Lance Taylor <iant@google.com>
16203 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16204 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16206 2017-01-30 Martin Sebor <msebor@redhat.com>
16208 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16209 Move constant to the right of a relational operator.
16210 (get_mpfr_format_length, format_character, format_string): Ditto.
16211 (should_warn_p, maybe_warn): Same.
16213 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16215 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16218 * asan.c (get_translation_unit_decl): Remove function.
16219 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16221 2017-01-30 Martin Liska <mliska@suse.cz>
16223 PR gcov-profile/79259
16224 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16225 -fprofile-generate.
16227 2017-01-30 Martin Liska <mliska@suse.cz>
16230 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16231 Initialize variables with NULL value.
16233 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16236 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16238 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16240 2017-01-30 Richard Biener <rguenther@suse.de>
16242 PR tree-optimization/79276
16243 * tree-vrp.c (process_assert_insertions): Properly adjust common
16244 when removing a duplicate.
16246 * gcc.dg/torture/pr79276.c: New testcase.
16248 2017-01-30 Richard Biener <rguenther@suse.de>
16250 PR tree-optimization/79256
16251 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16252 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16254 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16256 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16259 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16260 ("*r<noxa>sbg_<mode>_sll_bitmask")
16261 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16262 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16263 Use contiguous_bitmask_nowrap_operand.
16265 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16268 * config/rs6000/altivec.h (vec_xl): Revise #define.
16269 (vec_xst): Likewise.
16271 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16273 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16275 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16277 PR rtl-optimization/79194
16278 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16279 traps before call to bypass_conditional_jumps.
16281 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16283 PR tree-optimization/71374
16284 * lra-constraints.c (check_conflict_input_operands): New.
16285 (match_reload): Use it.
16287 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16290 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16291 account to calculate conflict_set.
16293 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16295 PR rtl-optimization/78559
16296 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16297 other_insn in combine.
16299 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16301 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16302 uint16_type_node for BT_UINT16.
16304 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16306 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16307 "RTL Tests" to menu.
16308 (GIMPLE Tests): New node.
16309 (RTL Tests): New node.
16311 2017-01-27 Richard Biener <rguenther@suse.de>
16313 PR tree-optimization/79245
16314 * tree-loop-distribution.c (distribute_loop): Apply cost
16315 modeling also to detected patterns.
16317 2017-01-27 Richard Biener <rguenther@suse.de>
16319 PR tree-optimization/71433
16320 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16321 (compare_assert_loc): New function.
16322 (process_assert_insertions): Sort and optimize assert locations
16323 to remove duplicates and push down identical assertions on
16324 edges to their destination block.
16326 2017-01-27 Richard Biener <rguenther@suse.de>
16328 PR tree-optimization/79244
16329 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16330 out SSA names even if abnormal.
16332 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16334 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16335 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16336 instead of MPFR_RNDN.
16338 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16341 * arm.c (arm_option_override): Don't call build_target_option_node
16342 until after doing all option overrides.
16343 (arm_valid_target_attribute_tree): Likewise.
16345 2017-01-27 Martin Liska <mliska@suse.cz>
16347 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16348 for {cd}tors and C++ {cd}tors.
16350 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16352 * config/s390/s390.md ("*setmem_long_and")
16353 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16355 2017-01-26 Martin Sebor <msebor@redhat.com>
16357 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16360 2017-01-26 Martin Sebor <msebor@redhat.com>
16362 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16363 HAVE_DFmode before using XFmode or DFmode.
16364 (parse_directive): Avoid using the z length modifier to avoid
16365 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16367 PR middle-end/78703
16368 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16369 to accept adjustment as an array.
16370 (get_int_range): New function.
16371 (struct directive): Make width and prec arrays.
16372 (directive::set_width, directive::set_precision): Call get_int_range.
16373 (format_integer, format_floating): Handle width and precision ranges.
16374 (format_string, parse_directive): Same.
16376 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16379 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16380 comdat_type_p set, just clone them, but keep the children in the
16384 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16385 which have direct callers with -fvar-tracking-assignments enabled
16387 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16388 inside of type units.
16390 2017-01-26 Martin Sebor <msebor@redhat.com>
16392 PR middle-end/78703
16393 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16395 (struct format_result): Replace number_chars, number_chars_min,
16396 and number_chars_max with a single member of struct result_range.
16398 (format_result::operator+=): Adjust.
16399 (struct fmtresult): Remove bounded. Handle likely and unlikely
16401 (fmtresult::adjust_for_width_or_precision): New function.
16402 (fmtresult:type_max_digits): New function.
16403 (bytes_remaining): Handle likely and unlikely counters.
16404 (min_bytes_remaining): Remove.
16405 (format_percent): Simplify.
16406 (format_integer, format_floating): Set likely and unlikely counters.
16407 (get_string_length, format_character, format_string): Same.
16408 (format_plain, should_warn_p): New function.
16409 (maybe_warn): Call should_warn_p. Update diagnostic messages
16410 and handle those for all directives, including plain strings.
16411 (format_directive): Handle likely and unlikely counters.
16412 Remove unnecessary quoting from diagnostics. Add an informational
16414 (add_bytes): Remove.
16415 (pass_sprintf_length::compute_format_length): Simplify.
16416 (try_substitute_return_value): Handle likely and unlikely counters.
16418 2017-01-26 Carl Love <cel@us.ibm.com>
16420 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16421 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16423 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16426 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16427 endianess for subregs into account.
16428 * lra-constraints.c (lra_constraints): Do risky transformations
16429 always on the first iteration.
16430 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16431 last_call_used_reg_set.
16432 (process_bb_lives): Define and use last_call_used_reg_set.
16433 * lra.c (lra): Always continue after lra_constraints on the first
16436 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16438 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16440 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16442 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16444 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16445 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16446 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16447 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16448 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16449 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16450 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16451 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16452 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16454 2017-01-26 Marek Polacek <polacek@redhat.com>
16457 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16458 for the third operand.
16460 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16462 PR middle-end/79236
16463 * omp-low.c (struct omp_context): Add simt_stmt field.
16464 (scan_omp_for): Return omp_context *.
16465 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16466 context to the _simt_ SIMD stmt.
16467 (lower_omp_for): For combined SIMD with sibling _simt_
16468 SIMD, make sure to use the same decls in _looptemp_
16469 clauses as in the sibling.
16471 2017-01-26 David Sherwood <david.sherwood@arm.com>
16473 PR middle-end/79212
16474 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16477 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16480 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16481 emit fld b; fld a; if possible.
16483 * brig-builtins.def: Update copyright years.
16484 * config/arm/arm_acle_builtins.def: Update copyright years.
16486 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16489 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16490 constraint instead of o for the stxsd instruction.
16492 2017-01-25 Carl Love <cel@us.ibm.com>
16494 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16495 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16497 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16499 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16501 2017-01-25 Richard Biener <rguenther@suse.de>
16503 PR tree-optimization/69264
16504 * target.def (vector_alignment_reachable): Improve documentation.
16505 * doc/tm.texi: Regenerate.
16506 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16508 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16509 earlier changes with respect to TYPE_USER_ALIGN.
16510 (vector_alignment_reachable_p): Likewise. Improve dumping.
16512 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16515 * config/arm/arm.md (xordi3): Force constant operand into a register
16518 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16520 * doc/invoke.texi (-fstore-merging): Correct default optimization
16521 levels at which it is enabled.
16522 (-O): Move -fstore-merging from list to...
16525 2017-01-25 Richard Biener <rguenther@suse.de>
16528 * omp-expand.c: Include debug.h.
16529 (expand_omp_taskreg): Make sure to generate early debug before
16530 outlining anything from a function.
16531 (expand_omp_target): Likewise.
16532 (grid_expand_target_grid_body): Likewise.
16534 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16537 * asan.c (get_translation_unit_decl): New function.
16538 (asan_add_global): Extract modules file name from globals
16539 TRANSLATION_UNIT_DECL name.
16541 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16544 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16545 for long calls with APCS frame and VFP.
16547 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16549 * cfg.c (original_copy_tables_initialized_p): New function.
16550 * cfg.h (original_copy_tables_initialized_p): New decl.
16551 * cfgrtl.c (relink_block_chain): Guard the call to
16552 free_original_copy_tables with a call to
16553 original_copy_tables_initialized_p.
16554 * cgraph.h (symtab_node::native_rtl_p): New decl.
16555 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16556 (symtab_node::needed_p): Don't assert for early assembly output
16557 for __RTL functions.
16558 (cgraph_node::finalize_function): Set "force_output" for __RTL
16560 (cgraph_node::analyze): Bail out early for __RTL functions.
16561 (analyze_functions): Update assertion to support __RTL functions.
16562 (cgraph_node::expand): Bail out early for __RTL functions.
16563 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16565 * function.h (struct function): Update comment for field
16567 * gimple-expr.c: Include "tree-pass.h".
16568 (gimple_has_body_p): Return false for __RTL functions.
16569 * Makefile.in (OBJS): Add run-rtl-passes.o.
16570 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16572 (gcc::pass_manager::get_clean_slate): New accessor.
16573 * passes.c: Include "insn-addr.h".
16574 (should_skip_pass_p): Add logging. Update logic for running
16575 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16576 property-provider override so it is only done for gimple passes.
16578 (skip_pass): New function.
16579 (execute_one_pass): Call skip_pass when skipping passes.
16580 * read-md.c (md_reader::read_char): Support filtering
16581 the input to a subset of line numbers.
16582 (md_reader::md_reader): Initialize fields
16583 m_first_line and m_last_line.
16584 (md_reader::read_file_fragment): New function.
16585 * read-md.h (md_reader::read_file_fragment): New decl.
16586 (md_reader::m_first_line): New field.
16587 (md_reader::m_last_line): New field.
16588 * read-rtl-function.c (function_reader::create_function): Only
16589 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16590 curr_properties. Set DECL_INITIAL to a dummy block.
16591 (read_rtl_function_body_from_file_range): New function.
16592 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16594 * run-rtl-passes.c: New file.
16595 * run-rtl-passes.h: New file.
16597 2017-01-24 Jeff Law <law@redhat.com>
16599 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16602 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16604 PR tree-optimization/79159
16605 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16606 (record_nonwrapping_iv): Improve boundary using above function if no
16607 value range information.
16609 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16610 Martin Jambor <mjambor@suse.cz>
16612 * brig-builtins.def: New file.
16613 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16614 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16615 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16616 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16617 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16618 * builtin-types.def (BT_INT8): New.
16619 (BT_INT16): Likewise.
16620 (BT_UINT8): Likewise.
16621 (BT_UINT16): Likewise.
16622 (BT_FN_ULONG): Likewise.
16623 (BT_FN_UINT_INT): Likewise.
16624 (BT_FN_UINT_ULONG): Likewise.
16625 (BT_FN_UINT_LONG): Likewise.
16626 (BT_FN_UINT_PTR): Likewise.
16627 (BT_FN_ULONG_PTR): Likewise.
16628 (BT_FN_INT8_FLOAT): Likewise.
16629 (BT_FN_INT16_FLOAT): Likewise.
16630 (BT_FN_UINT32_FLOAT): Likewise.
16631 (BT_FN_UINT16_FLOAT): Likewise.
16632 (BT_FN_UINT8_FLOAT): Likewise.
16633 (BT_FN_UINT64_FLOAT): Likewise.
16634 (BT_FN_UINT16_UINT32): Likewise.
16635 (BT_FN_UINT32_UINT16): Likewise.
16636 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16637 (BT_FN_INT_PTR_INT): Likewise.
16638 (BT_FN_UINT_PTR_UINT): Likewise.
16639 (BT_FN_LONG_PTR_LONG): Likewise.
16640 (BT_FN_ULONG_PTR_ULONG): Likewise.
16641 (BT_FN_VOID_UINT64_UINT64): Likewise.
16642 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16643 (BT_FN_INT8_INT8_INT8): Likewise.
16644 (BT_FN_INT16_INT16_INT16): Likewise.
16645 (BT_FN_INT_INT_INT): Likewise.
16646 (BT_FN_UINT_FLOAT_UINT): Likewise.
16647 (BT_FN_FLOAT_UINT_UINT): Likewise.
16648 (BT_FN_ULONG_UINT_UINT): Likewise.
16649 (BT_FN_ULONG_UINT_PTR): Likewise.
16650 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16651 (BT_FN_UINT_UINT_UINT): Likewise.
16652 (BT_FN_VOID_UINT_PTR): Likewise.
16653 (BT_FN_UINT_UINT_PTR: Likewise.
16654 (BT_FN_UINT32_UINT64_PTR): Likewise.
16655 (BT_FN_INT_INT_UINT_UINT): Likewise.
16656 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16657 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16658 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16659 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16660 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16661 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16662 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16663 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16664 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16665 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16666 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16667 * doc/frontends.texi: List BRIG FE.
16668 * doc/install.texi (Testing): Add BRIG tesring requirements.
16669 * doc/invoke.texi (Overall Options): Mention BRIG.
16670 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16672 2017-01-24 Richard Biener <rguenther@suse.de>
16674 PR translation/79208
16675 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16677 2017-01-24 Martin Jambor <mjambor@suse.cz>
16680 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16681 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16682 and known_contexts.
16684 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16686 PR middle-end/79123
16687 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16688 casts from signed to unsigned really don't have a range.
16690 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16692 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16693 GMP_RNDx for compatiblity.
16695 2017-01-24 Martin Liska <mliska@suse.cz>
16698 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16699 that would prevent us to call alloca with -1 as argument.
16701 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16703 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16704 Avoid -Wformat-security warning.
16706 2017-01-23 Andrew Pinski <apinski@cavium.com>
16708 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16711 2017-01-23 Martin Sebor <msebor@redhat.com>
16713 PR middle-end/78703
16714 * gimple-ssa-sprintf.c (warn_level): New global.
16715 (format_integer): Use it here and throughout the rest of the file.
16716 Use the same switch to compute sign as base.
16717 (maybe_warn): New function.
16718 (format_directive): Factor out warnings into maybe_warn.
16719 Add debugging output. Use warn_level.
16720 (add_bytes): Use warn_level.
16721 (pass_sprintf_length::compute_format_length): Add debugging output.
16722 (try_substitute_return_value): Same.
16723 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
16725 PR middle-end/78703
16726 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
16727 (struct fmtresult, format_integer, format_floating): Adjust.
16728 (fmtresult::fmtresult): Set max correctly in two argument ctor.
16729 (get_string_length, format_string,format_directive): Same.
16730 (pass_sprintf_length::compute_format_length): Same.
16731 (try_substitute_return_value): Simplify slightly.
16733 PR middle-end/78703
16734 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
16735 (fmtresult::operator+=): Outlined.
16736 (struct fmtresult): Add ctors.
16737 (struct conversion_spec): Rename...
16738 (struct directive): ...to this. Add and remove data members.
16739 (directive::set_width, directive::set_precision): New functions.
16740 (format_percent): Use fmtresult ctor.
16741 (get_width_and_precision): Remove.
16742 (format_integer): Make naming changes. Avoid computing width and
16744 (format_floating): Same. Adjust indentation.
16745 (format_character, format_none): New functions.
16746 (format_string): Moved character handling to format_character.
16747 (format_directive): Remove arguments, change return type.
16748 (parse_directive): New function.
16749 (pass_sprintf_length::compute_format_length): Move directive
16750 parsing to parse_directive.
16752 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16754 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
16755 (assign_assembler_name_if_needed): ... this.
16756 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
16757 (assign_assembler_name_if_needed): ... this.
16758 (free_lang_data_in_cgraph): Adjust callers.
16759 * cgraphunit.c (cgraph_node::analyze): Likewise.
16760 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
16763 2017-01-23 Richard Biener <rguenther@suse.de>
16765 PR tree-optimization/79088
16766 PR tree-optimization/79188
16767 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
16768 resetting loop bounds after last path deletion. Reset loop
16769 bounds of the target loop, make code match the comments.
16770 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
16771 Make sure loops need no fixups.
16773 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16775 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
16776 exponent support with double type for first argument.
16777 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
16778 type returned by __builtin_vec_extract_sig,
16779 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
16780 functions from "vector int" to "vector unsigned int" or from
16781 "vector long long int" to "vector unsigned long long int".
16782 Changed type returned by __builtin_vec_extract_exp,
16783 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
16784 functions from "vector int" to "vector unsigned int" or from
16785 "vector long long int" to "vector unsigned long long int".
16786 Changed return type of __builtin_vec_test_data_class,
16787 __builtin_vec_test_data_class_sp, and
16788 __builtin_vec_test_data_class_dp from "vector int" to
16789 "vector bool int" or from "vector long long int" to "vector bool
16790 long long int" and changed second argument type from "unsigned
16791 int" to "int". Added new overloaded function forms "vector float
16792 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
16793 "vector float __builtin_vec_insert_exp_sp (vector float, vector
16794 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
16795 double, vector unsigned long long int)" and "vector double
16796 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
16797 long int)". Changed return type of
16798 __builtin_scalar_test_data_class and
16799 __builtin_scalar_test_data_class_sp and
16800 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
16801 int" and changed second argument from "unsigned int" to "int".
16802 Changed type returned by __builtin_scalar_test_neg,
16803 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
16804 from "int" to "bool int". Added new overloaded function form
16805 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
16806 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
16807 exponent double-precision with floating point first argument.
16808 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16809 documentation of scalar_test_data_class, scalar_test_neg,
16810 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
16811 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
16812 vec_test_data_class built-in functions to reflect refinements in
16813 their type signatures.
16815 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
16817 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
16819 (aarch64_elf_asm_destructor): Likewise.
16821 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
16823 PR rtl-optimization/78634
16824 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
16825 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
16826 * ifcvt.c (noce_try_cmove): Add missing cost check.
16828 PR rtl-optimization/71724
16829 * combine.c (if_then_else_cond): Look for situations where it is
16830 beneficial to undo the work of one of the recursive calls.
16832 2017-01-23 Bin Cheng <bin.cheng@arm.com>
16834 PR tree-optimization/70754
16835 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
16836 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
16837 combined stmt before it if not NULL.
16838 (combine_chains): Process refs reversely and compute dominance point
16841 2017-01-23 Martin Liska <mliska@suse.cz>
16843 PR tree-optimization/79196
16844 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
16845 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
16847 (strlen_optimize_stmt): Call the renamed function.
16849 2017-01-23 Michael Matz <matz@suse.de>
16851 PR tree-optimization/78384
16852 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
16854 2017-01-23 Richard Biener <rguenther@suse.de>
16856 PR tree-optimization/79186
16857 * tree-vrp.c (register_new_assert_for): Make sure we've seen
16858 both incoming edges before moving an assert.
16860 2017-01-23 Martin Jambor <mjambor@suse.cz>
16862 * ipa-prop.c (load_from_param_1): Removed.
16863 (load_from_unmodified_param): Bits from load_from_param_1 put back
16865 (load_from_param): Removed.
16866 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
16867 with stmt. Reverted back to use of load_from_unmodified_param.
16869 2017-01-23 Martin Jambor <mjambor@suse.cz>
16872 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
16873 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
16874 field a pointer to garbage collected vector, mark lattices and
16875 ipcp_orig_node with GTY((skip)).
16876 (ipa_get_param_count): Adjust to descriptors being a pointer.
16877 (ipa_get_param): Likewise.
16878 (ipa_get_type): Likewise.
16879 (ipa_get_param_move_cost): Likewise.
16880 (ipa_set_param_used): Likewise.
16881 (ipa_get_controlled_uses): Likewise.
16882 (ipa_set_controlled_uses): Likewise.
16883 (ipa_is_param_used): Likewise.
16884 (ipa_node_params_t): Move into garbage collector. New methods insert
16886 (ipa_node_params_sum): Annotate wth GTY(()).
16887 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
16889 (ipa_load_from_parm_agg): Adjust declaration.
16890 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
16891 * ipa-profile.c (ipa_profile): Likewise.
16892 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
16893 (ipa_populate_param_decls): Make descriptors parameter garbage
16895 (ipa_dump_param): Adjust to descriptors being a pointer.
16896 (ipa_alloc_node_params): Likewise.
16897 (ipa_initialize_node_params): Likewise.
16898 (load_from_param_1): Make descriptors parameter garbage collected.
16899 (load_from_unmodified_param): Likewise.
16900 (load_from_param): Likewise.
16901 (ipa_load_from_parm_agg): Likewise.
16902 (ipa_node_params::~ipa_node_params): Removed.
16903 (ipa_free_all_node_params): Remove call to delete operator.
16904 (ipa_node_params_t::insert): New.
16905 (ipa_node_params_t::remove): Likewise.
16906 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
16907 copy known_csts and known_contexts vectors.
16908 (ipa_read_node_info): Adjust to descriptors being a pointer.
16909 (ipcp_modif_dom_walker): Make m_descriptors field garbage
16911 (ipcp_transform_function): Make descriptors variable garbage
16914 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
16916 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
16917 * config/i386/avx512dqintrin.h: Ditto.
16918 * config/i386/avx512fintrin.h: Ditto.
16919 * config/i386/i386.c: Handle new builtins.
16920 * config/i386/i386-builtin.def: Add new builtins.
16921 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
16922 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
16924 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16925 Martin Liska <mliska@suse.cz>
16927 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
16928 * asan.c (asan_expand_poison_ifn): Support stores and use
16929 appropriate ASAN report function.
16930 * internal-fn.c (expand_ASAN_POISON_USE): New function.
16931 * internal-fn.def (ASAN_POISON_USE): Declare.
16932 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
16933 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
16934 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
16935 ASAN_POISON calls w/o LHS.
16936 * tree-ssa.c (execute_update_addresses_taken): Create clobber
16937 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
16938 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
16939 * gimplify.c (asan_poison_variables): Add attribute
16940 use_after_scope_memory to variables that really needs to live
16942 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
16943 having the attribute.
16945 2017-01-23 Martin Liska <mliska@suse.cz>
16947 * asan.c (create_asan_shadow_var): New function.
16948 (asan_expand_poison_ifn): Likewise.
16949 * asan.h (asan_expand_poison_ifn): New declaration.
16950 * internal-fn.c (expand_ASAN_POISON): Likewise.
16951 * internal-fn.def (ASAN_POISON): New builtin.
16952 * sanopt.c (pass_sanopt::execute): Expand
16953 asan_expand_poison_ifn.
16954 * tree-inline.c (copy_decl_for_dup_finish): Make function
16956 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
16957 * tree-ssa.c (is_asan_mark_p): New function.
16958 (execute_update_addresses_taken): Rewrite local variables
16959 (identified just by use-after-scope as addressable) into SSA.
16961 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16963 * doc/install.texi (Specific): opensource.apple.com uses https
16964 now. Remove trailing slash.
16966 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16968 * README.Portability: Remove note on an Irix compatibility issue.
16970 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
16972 * gcov.c (INCLUDE_ALGORITHM): Define.
16973 (INCLUDE_VECTOR): Define.
16974 No longer include <vector> and <algorithm> directly.
16976 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
16978 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
16980 * doc/invoke.texi (Code Gen Options): Ditto.
16982 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
16985 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
16987 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
16989 rtl-optimization/79125
16990 * cprop.c (local_cprop_pass): Handle cases where we make an
16991 unconditional trap.
16993 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
16997 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
16998 read from, for big endian.
17000 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17002 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17003 register pauth builtins for LP64 only.
17005 2017-01-20 Marek Polacek <polacek@redhat.com>
17008 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17011 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17013 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17015 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17016 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17017 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17019 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17022 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17023 in big-endian mode when they are not a single duplicated value.
17025 2017-01-20 Richard Biener <rguenther@suse.de>
17027 * BASE-VER: Bump to 7.0.1.
17029 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17031 * omp-low.c (omplow_simd_context): New struct. Use it...
17032 (lower_rec_simd_input_clauses): ...here and...
17033 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17034 references to idx, lane, max_vf, is_simt.
17036 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17038 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17041 2017-01-20 Martin Jambor <mjambor@suse.cz>
17043 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17044 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17046 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17047 (GTFILES): Rename hsa.c to hsa-common.c.
17048 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17049 * hsa-dump.c: Likewise.
17050 * hsa-gen.c: Likewise.
17051 * hsa-regalloc.c: Likewise.
17052 * ipa-hsa.c: Likewise.
17053 * omp-expand.c: Likewise.
17054 * omp-low.c: Likewise.
17055 * toplev.c: Likewise.
17057 2017-01-20 Marek Polacek <polacek@redhat.com>
17060 * doc/invoke.texi: Document -Wduplicated-branches.
17061 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17062 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17063 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17064 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17065 return 0 only when not OEP_LEXICOGRAPHIC.
17066 (fold_build_cleanup_point_expr): Use the expression
17067 location when building CLEANUP_POINT_EXPR.
17068 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17069 * tree.c (add_expr): Handle error_mark_node.
17071 2017-01-20 Martin Liska <mliska@suse.cz>
17074 * tree-profile.c (init_ic_make_global_vars): Do not call
17076 (gimple_init_gcov_profiler): Likewise.
17078 2017-01-20 Martin Liska <mliska@suse.cz>
17081 * cgraph.h (maybe_create_reference): Remove argument and
17083 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17085 * ipa-cp.c (create_specialized_node): Likewise.
17086 * symtab.c (symtab_node::maybe_create_reference): Handle
17087 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17089 2017-01-20 Martin Liska <mliska@suse.cz>
17091 * read-rtl-function.c (function_reader::create_function): Use
17092 build_decl instread of build_decl_stat.
17094 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17096 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17097 * config/i386/avx512dqintrin.h: Ditto.
17098 * config/i386/avx512fintrin.h: Ditto.
17099 * config/i386/i386-builtin-types.def: Add new types.
17100 * config/i386/i386.c: Handle new types.
17101 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17102 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17103 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17104 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17105 (__builtin_ia32_kshiftridi): New.
17106 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17108 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17112 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17113 define to rs6000_init_stack_protect_guard.
17114 (rs6000_init_stack_protect_guard): New function.
17116 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17117 Yunqiang Su <yunqiang.su@imgtec.com>
17119 * config.gcc (supported_defaults): Add madd4.
17120 (with_madd4): Add validation.
17121 (all_defaults): Add madd4.
17122 * config/mips/mips.opt (mmadd4): New option.
17123 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17125 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17127 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17128 (ISA_HAS_FUSED_MADD4): Likewise.
17129 * doc/invoke.texi (-mmadd4): Document the new option.
17130 * doc/install.texi (--with-madd4): Document the new option.
17132 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17134 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17135 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17136 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17137 (aarch64_init_pauth_hint_builtins): New.
17138 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17139 (aarch64_expand_builtin): Expand new builtins.
17141 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17143 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17144 * combine-stack-adj.c (no_unhandled_cfa): Handle
17145 REG_CFA_TOGGLE_RA_MANGLE.
17146 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17147 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17148 info for return address signing.
17149 (aarch64_expand_epilogue): Likewise.
17151 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17153 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17154 * config/aarch64/aarch64-protos.h
17155 (aarch64_return_address_signing_enabled): New declaration.
17156 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17158 (aarch64_expand_prologue): Sign return address before it's pushed onto
17160 (aarch64_expand_epilogue): Authenticate return address fetched from
17162 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17163 (aarch64_attributes): New function attributes for "sign-return-address".
17164 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17165 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17166 ("*do_return"): Generate combined instructions according to key index.
17167 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17168 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17170 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17171 * config/aarch64/aarch64.opt (msign-return-address=): New.
17172 * doc/extend.texi (AArch64 Function Attributes): Documents
17173 "sign-return-address=".
17174 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17176 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17178 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17179 overall option summary.
17181 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17183 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17184 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17185 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17186 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17188 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17190 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17191 -mpower9-minmax by default for -mcpu=power9.
17192 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17193 128-bit floating point.
17195 2017-01-20 Alan Modra <amodra@gmail.com>
17197 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17198 optimizing for size.
17200 2017-01-20 Alan Modra <amodra@gmail.com>
17203 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17204 for strcmp and strncmp from corresponding builtin decl.
17206 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17208 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17209 instead of i386/rtems-64.h.
17210 * config/i386/rtems-64.h: Remove.
17212 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17216 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17218 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17220 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17222 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17223 Change int to HOST_WIDE_INT.
17224 * config/aarch64/aarch64-protos.h
17225 (aarch64_simd_gen_const_vector_dup): Likewise.
17226 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17228 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17230 * langhooks-def.h (lhd_type_for_size): New decl.
17231 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17232 * langhooks.c (lhd_type_for_size): New function, taken from
17235 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17237 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17238 define_bypass for CR latency.
17239 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17240 (power9-alu2): Add define_bypass for CR latency.
17242 (power9-mul): Update insn latency.
17243 (power9-mul-compare): Update insn latency, bypass latency and remove
17246 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17248 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17250 * config/aarch64/aarch64.md
17251 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17252 aarch64_nopcrelative_literal_loads.
17253 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17255 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17257 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17258 TARGET_LOONGSON_3A.
17259 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17261 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17264 * config.gcc (supported_defaults): Add lxc1-sxc1.
17265 (with_lxc1_sxc1): Add validation.
17266 (all_defaults): Add lxc1-sxc1.
17267 * config/mips/mips.opt (mlxc1-sxc1): New option.
17268 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17270 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17271 __mips_no_lxc1_sxc1.
17272 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17273 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17274 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17276 2017-01-19 Richard Biener <rguenther@suse.de>
17278 PR tree-optimization/72488
17279 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17280 sure to restore SSA info.
17281 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17283 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17285 PR rtl-optimization/79121
17286 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17287 of the inner type when shifting an extended value.
17289 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17292 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17293 interposable aliases.
17295 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17298 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17299 Use the evmergelohi instruction.
17300 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17301 (mov_sitf_e500_subreg8_2_be): Likewise.
17302 (mov_sitf_e500_subreg12_2_le): Likewise.
17303 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17304 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17305 (mov_sitf_e500_subreg8_2_le): Likewise.
17306 (mov_sitf_e500_subreg12_2_be): Likewise.
17308 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17310 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17311 attribute from vecsimple to vecperm.
17312 (altivec_vbpermq2): Likewise.
17314 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17317 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17319 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17320 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17321 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17322 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17323 case where N arg is SIZE_MAX.
17324 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17325 (cmpstrsi): Add pattern.
17327 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17329 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17330 __builtin_vec_revb builtins.
17331 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17332 built-in functions to support generation of the ISA 3.0 XXBR<x>
17333 vector byte reverse instructions.
17334 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17335 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17336 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17337 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17338 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17339 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17340 (P9V_BUILTIN_VEC_REVB): Likewise.
17341 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17342 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17343 (p9_xxbrq_v16qi): Likewise.
17344 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17345 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17346 (p9_xxbrh_v8hi): Likewise.
17347 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17348 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17349 vec_revb built-in functions.
17351 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17353 PR rtl-optimization/78952
17354 * config/i386/i386.md (any_extract): New code iterator.
17355 (*insvqi_2): Use any_extract for source operand.
17356 (*insvqi_3): Use any_shiftrt for source operand.
17358 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17360 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17362 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17364 2017-01-18 Matthias Klose <doko@ubuntu.com>
17366 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17368 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17370 * config/rs6000/altivec.h (vec_bperm): Change #define.
17371 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17372 (altivec_vbpermq2): New define_insn.
17373 (altivec_vbpermd): Likewise.
17374 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17375 function interface.
17376 (VBPERMD): Likewise.
17377 (VBPERM): New polymorphic function interface.
17378 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17379 Add entries for P9V_BUILTIN_VEC_VBPERM.
17380 * doc/extend.texi: Add interfaces for vec_bperm.
17382 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17384 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17385 first letter of error messages.
17386 (s390_resolve_overloaded_builtin): Likewise.
17387 * config/s390/s390.c (s390_expand_builtin): Likewise.
17388 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17389 (s390_valid_target_attribute_inner_p): Likewise.
17390 * config/s390/s390.md ("tabort"): Likewise.
17392 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17394 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17395 (ISA_AVOID_DIV_HILO): New macro.
17396 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17397 (ISA_HAS_DDIV): Likewise.
17399 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17401 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17403 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17405 * doc/invoke.texi (fabi-version): Spelling fix.
17407 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17410 * doc/invoke.texi (fabi-version): Mention mangling fix for
17413 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17416 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17418 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17421 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17422 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17424 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17426 (stack_protect_test): Ditto.
17427 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17428 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17430 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17431 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17432 -mstack-protector-guard-offset=.
17433 (RS/6000 and PowerPC Options): Ditto.
17435 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17437 * config/i386/i386.h (MASK_CLASS_P): New define.
17438 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17439 there are no registers from different register sets also when
17440 mask registers are used. Update function comment.
17441 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17442 to (*k/*r) and (*k/*km) alternatives.
17444 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17446 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17447 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17448 (EH_RETURN_HANDLER_RTX): New define.
17449 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17450 Force frame pointer in EH return functions.
17451 (aarch64_expand_epilogue): Add barrier for eh_return.
17452 (aarch64_final_eh_return_addr): Remove.
17453 (aarch64_eh_return_handler_rtx): New function.
17454 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17456 (aarch64_eh_return_handler_rtx): New prototype.
17458 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17460 * config/rs6000/altivec.h (vec_rlmi): New #define.
17461 (vec_vrlnm): Likewise.
17462 (vec_rlnm): Likewise.
17463 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17464 (UNSPEC_VRLNM): Likewise.
17465 (VIlong): New mode iterator.
17466 (altivec_vrl<VI_char>mi): New define_insn.
17467 (altivec_vrl<VI_char>nm): Likewise.
17468 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17470 (VRLDNM): Likewise.
17471 (RLNM): New polymorphic function entry.
17472 (VRLWMI): New monomorphic function entry.
17473 (VRLDMI): Likewise.
17474 (RLMI): New polymorphic function entry.
17475 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17476 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17477 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17480 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17483 * dwarf2out.c (field_byte_offset): Restore the
17484 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17485 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17486 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17489 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17492 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17494 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17497 * dwarf2out.c (add_data_member_location_attribute): For constant
17498 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17499 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17500 DW_AT_byte_size attributes.
17502 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17504 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17505 after forcing to constant memory when the code model is medium.
17507 2017-01-17 Julia Koval <julia.koval@intel.com>
17510 * config/i386/avx512fintrin.h
17511 (_mm512_i32gather_ps): Change __addr type to void const*.
17512 (_mm512_mask_i32gather_ps): Ditto.
17513 (_mm512_i32gather_pd): Ditto.
17514 (_mm512_mask_i32gather_pd): Ditto.
17515 (_mm512_i64gather_ps): Ditto.
17516 (_mm512_mask_i64gather_ps): Ditto.
17517 (_mm512_i64gather_pd): Ditto.
17518 (_mm512_mask_i64gather_pd): Ditto.
17519 (_mm512_i32gather_epi32): Ditto.
17520 (_mm512_mask_i32gather_epi32): Ditto.
17521 (_mm512_i32gather_epi64): Ditto.
17522 (_mm512_mask_i32gather_epi64): Ditto.
17523 (_mm512_i64gather_epi32): Ditto.
17524 (_mm512_mask_i64gather_epi32): Ditto.
17525 (_mm512_i64gather_epi64): Ditto.
17526 (_mm512_mask_i64gather_epi64): Ditto.
17527 (_mm512_i32scatter_ps): Change __addr type to void*.
17528 (_mm512_mask_i32scatter_ps): Ditto.
17529 (_mm512_i32scatter_pd): Ditto.
17530 (_mm512_mask_i32scatter_pd): Ditto.
17531 (_mm512_i64scatter_ps): Ditto.
17532 (_mm512_mask_i64scatter_ps): Ditto.
17533 (_mm512_i64scatter_pd): Ditto.
17534 (_mm512_mask_i64scatter_pd): Ditto.
17535 (_mm512_i32scatter_epi32): Ditto.
17536 (_mm512_mask_i32scatter_epi32): Ditto.
17537 (_mm512_i32scatter_epi64): Ditto.
17538 (_mm512_mask_i32scatter_epi64): Ditto.
17539 (_mm512_i64scatter_epi32): Ditto.
17540 (_mm512_mask_i64scatter_epi32): Ditto.
17541 (_mm512_i64scatter_epi64): Ditto.
17542 (_mm512_mask_i64scatter_epi64): Ditto.
17543 * config/i386/avx512pfintrin.h
17544 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17545 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17546 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17547 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17548 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17549 (_mm512_prefetch_i32scatter_ps): Ditto.
17550 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17551 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17552 (_mm512_prefetch_i64scatter_pd): Ditto.
17553 (_mm512_prefetch_i64scatter_ps): Ditto.
17554 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17555 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17556 * config/i386/avx512vlintrin.h
17557 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17558 (_mm_mmask_i32gather_ps): Ditto.
17559 (_mm256_mmask_i32gather_pd): Ditto.
17560 (_mm_mmask_i32gather_pd): Ditto.
17561 (_mm256_mmask_i64gather_ps): Ditto.
17562 (_mm_mmask_i64gather_ps): Ditto.
17563 (_mm256_mmask_i64gather_pd): Ditto.
17564 (_mm_mmask_i64gather_pd): Ditto.
17565 (_mm256_mmask_i32gather_epi32): Ditto.
17566 (_mm_mmask_i32gather_epi32): Ditto.
17567 (_mm256_mmask_i32gather_epi64): Ditto.
17568 (_mm_mmask_i32gather_epi64): Ditto.
17569 (_mm256_mmask_i64gather_epi32): Ditto.
17570 (_mm_mmask_i64gather_epi32): Ditto.
17571 (_mm256_mmask_i64gather_epi64): Ditto.
17572 (_mm_mmask_i64gather_epi64): Ditto.
17573 (_mm256_i32scatter_ps): Change __addr type to void*.
17574 (_mm256_mask_i32scatter_ps): Ditto.
17575 (_mm_i32scatter_ps): Ditto.
17576 (_mm_mask_i32scatter_ps): Ditto.
17577 (_mm256_i32scatter_pd): Ditto.
17578 (_mm256_mask_i32scatter_pd): Ditto.
17579 (_mm_i32scatter_pd): Ditto.
17580 (_mm_mask_i32scatter_pd): Ditto.
17581 (_mm256_i64scatter_ps): Ditto.
17582 (_mm256_mask_i64scatter_ps): Ditto.
17583 (_mm_i64scatter_ps): Ditto.
17584 (_mm_mask_i64scatter_ps): Ditto.
17585 (_mm256_i64scatter_pd): Ditto.
17586 (_mm256_mask_i64scatter_pd): Ditto.
17587 (_mm_i64scatter_pd): Ditto.
17588 (_mm_mask_i64scatter_pd): Ditto.
17589 (_mm256_i32scatter_epi32): Ditto.
17590 (_mm256_mask_i32scatter_epi32): Ditto.
17591 (_mm_i32scatter_epi32): Ditto.
17592 (_mm_mask_i32scatter_epi32): Ditto.
17593 (_mm256_i32scatter_epi64): Ditto.
17594 (_mm256_mask_i32scatter_epi64): Ditto.
17595 (_mm_i32scatter_epi64): Ditto.
17596 (_mm_mask_i32scatter_epi64): Ditto.
17597 (_mm256_i64scatter_epi32): Ditto.
17598 (_mm256_mask_i64scatter_epi32): Ditto.
17599 (_mm_i64scatter_epi32): Ditto.
17600 (_mm_mask_i64scatter_epi32): Ditto.
17601 (_mm256_i64scatter_epi64): Ditto.
17602 (_mm256_mask_i64scatter_epi64): Ditto.
17603 (_mm_i64scatter_epi64): Ditto.
17604 (_mm_mask_i64scatter_epi64): Ditto.
17605 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17606 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17607 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17608 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17609 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17610 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17611 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17612 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17613 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17614 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17615 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17616 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17617 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17618 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17619 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17620 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17621 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17622 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17623 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17624 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17625 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17626 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17627 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17628 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17629 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17630 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17631 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17632 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17633 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17634 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17635 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17636 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17637 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17638 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17639 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17640 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17641 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17642 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17643 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17644 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17645 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17646 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17647 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17648 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17649 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17650 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17651 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17652 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17653 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17654 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17655 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17656 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17657 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17658 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17659 definitions accordingly.
17661 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17662 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17665 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17668 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17671 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17672 hard regs for inner regno.
17674 2017-01-17 Martin Liska <mliska@suse.cz>
17677 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17678 assumption and add comment.
17680 2017-01-17 Nathan Sidwell <nathan@acm.org>
17682 * ipa-visibility.c (localize_node): New function, broken out of ...
17683 (function_and_variable_visibility): ... here. Call it.
17685 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17687 PR middle-end/77445
17688 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17689 correctly set frequency of oudgoing edge.
17690 (duplicate_thread_path): Fix profile updating.
17692 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17695 * configure.ac: Add GCC_BASE_VER.
17696 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17697 version from BASE-VER file.
17698 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17699 (gcc.o): Depend on $(BASEVER).
17700 * common.opt (dumpfullversion): New option.
17701 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17702 * doc/invoke.texi: Document -dumpfullversion.
17703 * doc/install.texi: Document --with-gcc-major-version-only.
17704 * configure: Regenerated.
17706 2017-01-17 Richard Biener <rguenther@suse.de>
17708 PR tree-optimization/71433
17709 * tree-vrp.c (register_new_assert_for): Merge same asserts
17710 on all incoming edges.
17711 (process_assert_insertions_for): Handle insertions at the
17714 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17716 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17717 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17719 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
17722 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
17725 2017-01-17 Alan Modra <amodra@gmail.com>
17728 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
17729 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
17730 symbolic stack limit when pic.
17732 2017-01-16 Martin Sebor <msebor@redhat.com>
17734 PR tree-optimization/78608
17735 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
17737 2017-01-16 Jeff Law <law@redhat.com>
17740 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
17741 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
17742 for several include directories that may be relative to sysroot.
17743 * config/i386/x-mingw32 (gplus_includedir): Define.
17744 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
17745 (native_system_includedir): Likewise.
17746 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
17747 override if TARGET_SYSTEM_ROOT is defined.
17748 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
17750 PR tree-optimization/79090
17751 PR tree-optimization/33562
17752 PR tree-optimization/61912
17753 PR tree-optimization/77485
17754 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
17755 and computed trims into the dump file.
17757 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17759 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
17761 2017-01-16 Jakub Jelinek <jakub@redhat.com>
17764 * gimplify.c (gimplify_init_constructor): If want_value and
17765 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
17769 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
17770 sequence. Formatting fixes.
17771 (doloop_optimize): Formatting fixes.
17774 * gcc.c (debug_level_greater_than_spec_func): New function.
17775 (static_spec_functions): Add debug-level-gt spec function.
17776 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
17778 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17779 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17780 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
17781 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
17782 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
17783 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
17785 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
17787 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
17788 QImode fixups to general and mask registers only.
17790 2017-01-16 Carl Love <cel@us.ibm.com>
17792 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
17793 for built-in functions
17794 vector signed char vec_nabs (vector signed char)
17795 vector signed short vec_nabs (vector signed short)
17796 vector signed int vec_nabs (vector signed int)
17797 vector signed long long vec_nabs (vector signed long long)
17798 vector float vec_nabs (vector float)
17799 vector double vec_nabs (vector double)
17800 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
17802 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
17803 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
17804 * doc/extend.texi: Update the documentation file for the new built-in
17807 2017-01-16 Martin Sebor <msebor@redhat.com>
17809 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
17812 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17814 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
17815 UNSPEC_VSX__XXSPLTD to require special splat handling.
17817 2017-01-16 David Malcolm <dmalcolm@redhat.com>
17820 * system.h: Poison strndup.
17822 2017-01-16 Alan Modra <amodra@gmail.com>
17825 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
17828 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
17830 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
17832 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
17834 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
17835 call recog here. Assert that INSN_CODE (insn) is non-negative.
17837 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17840 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
17842 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
17843 in the currently scheduled RTL fragment.
17845 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17847 PR rtl-optimization/78751
17848 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
17851 2017-01-14 Jeff Law <law@redhat.com>
17853 PR tree-optimization/79090
17854 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
17855 variable length stores.
17856 (compute_trims): Delete dead assignment to *trim_tail.
17857 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
17860 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
17862 PR rtl-optimization/78626
17863 PR rtl-optimization/78727
17864 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
17865 of a block, and split such blocks after everything else is finished.
17867 2017-01-14 Alan Modra <amodra@gmail.com>
17870 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
17871 target legitimate_combined_insn.
17872 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
17873 (rs6000_legitimate_combined_insn): New function.
17874 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
17876 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
17877 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
17878 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
17880 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
17882 * doc/frontends.texi (G++ and GCC): Remove references to Java.
17884 2017-01-13 Jeff Law <law@redhat.com>
17886 PR tree-optimization/33562
17887 PR tree-optimization/61912
17888 PR tree-optimization/77485
17889 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
17891 (delete_dead_assignment): Likewise.
17892 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
17893 statement to delete_dead_call and delete_dead_assignment.
17895 2017-01-13 David Malcolm <dmalcolm@redhat.com>
17898 * substring-locations.c (format_warning_va): Strengthen case 1 so
17899 that both endpoints of the substring must be within the format
17900 range for just the substring to be printed.
17902 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
17904 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
17905 * config/i386/i386.c (ix86_target_string): Add missing options
17906 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
17907 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
17908 flags_other and ix86_target_other to flags2_other. Display unknown
17910 (ix86_valid_target_attribute_inner_p): Add missing options and
17911 reorder options by implied ISAs, as in ix86_target_string.
17913 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17915 * hash-table.h (hash_table::too_empty_p): New function.
17916 (hash_table::expand): Use it.
17917 (hash_table::traverse): Likewise.
17918 (hash_table::empty_slot): Use sizeof (value_type) instead of
17919 sizeof (PTR) to convert bytes to elements. Shrink the table
17920 if the current size is excessive for the current number of
17923 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17925 * ira-costs.c (record_reg_classes): Break from the inner loop
17926 early once alt_fail is known to be true. Update outer loop
17927 handling accordingly.
17929 2017-01-13 Jeff Law <law@redhat.com>
17931 * tree-ssa-dse.c (decrement_count): New function.
17932 (increment_start_addr, maybe_trim_memstar_call): Likewise.
17933 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
17934 when we know the partially dead statement is a mem* function.
17936 PR tree-optimization/61912
17937 PR tree-optimization/77485
17938 * tree-ssa-dse.c: Include expr.h.
17939 (maybe_trim_constructor_store): New function.
17940 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
17942 PR tree-optimization/33562
17943 PR tree-optimization/61912
17944 PR tree-optimization/77485
17945 * doc/invoke.texi: Document new dse-max-object-size param.
17946 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
17947 * tree-ssa-dse.c: Include params.h.
17948 (dse_store_status): New enum.
17949 (initialize_ao_ref_for_dse): New, partially extracted from
17951 (valid_ao_ref_for_dse, normalize_ref): New.
17952 (setup_live_bytes_from_ref, compute_trims): Likewise.
17953 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
17954 (maybe_trim_partially_dead_store): Likewise.
17955 (maybe_trim_complex_store): Likewise.
17956 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
17957 Track what bytes live from the original store. Return tri-state
17958 for dead, partially dead or live.
17959 (dse_dom_walker): Add constructor, destructor and new private members.
17960 (delete_dead_call, delete_dead_assignment): New extracted from
17962 (dse_optimize_stmt): Make a member of dse_dom_walker.
17963 Use initialize_ao_ref_for_dse.
17965 PR tree-optimization/33562
17966 PR tree-optimization/61912
17967 PR tree-optimization/77485
17968 * sbitmap.h (bitmap_count_bits): Prototype.
17969 (bitmap_clear_range, bitmap_set_range): Likewise.
17970 * sbitmap.c (bitmap_clear_range): New function.
17971 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
17973 2017-01-13 Martin Liska <mliska@suse.cz>
17976 * function.c (set_cfun): Add new argument force.
17977 * function.h (set_cfun): Likewise.
17978 * ipa-inline-transform.c (inline_call): Use the function when
17979 strict alising from is dropped for function we inline to.
17981 2017-01-13 Richard Biener <rguenther@suse.de>
17983 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
17984 for dumping GIMPLE INTEGER_CSTs.
17986 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17988 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
17989 to 201112L since C++17.
17991 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17994 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
17995 if -fsanitize=kernel-address is present.
17997 2017-01-13 Richard Biener <rguenther@suse.de>
17999 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18000 as _Literal ( type ) number in case usual suffixes do not
18001 preserve all information.
18003 2017-01-13 Richard Biener <rguenther@suse.de>
18005 PR tree-optimization/77283
18006 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18007 and ssa-iterators.h.
18008 (is_feasible_trace): Implement a cost model based on joiner
18011 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18014 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18015 char or short to __float128/_Float128 directly.
18017 2017-01-12 Martin Sebor <msebor@redhat.com>
18019 to -Wformat-overflow.
18020 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18021 (min_bytes_remaining): Same.
18022 (get_string_length): Same.
18023 (format_string): Same.
18024 (format_directive): Same.
18026 (pass_sprintf_length::handle_gimple_call): Same.
18028 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18030 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18031 info.nowrite calls with no lhs that can't throw. Return bool
18032 whether gsi_remove has been called or not.
18033 (pass_sprintf_length::handle_gimple_call): Return bool whether
18034 try_substitute_return_value called gsi_remove. Formatting fix.
18035 (pass_sprintf_length::execute): Don't use gsi_remove if
18036 handle_gimple_call returned true.
18039 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18040 be removed due to side-effects, don't remove following barrier nor
18041 turn the successor edge into fallthru edge.
18043 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18046 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18047 element-reversing loads and stores as not swappable.
18049 2017-01-12 Nathan Sidwell <nathan@acm.org>
18050 Nicolai Stange <nicstange@gmail.com>
18052 * combine.c (try_combine): Don't ignore result of overlap checking
18053 loop. Combine overlap & asm check into single loop.
18055 2017-01-12 Richard Biener <rguenther@suse.de>
18057 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18058 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18060 2017-01-12 Richard Biener <rguenther@suse.de>
18062 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18063 and TS_TARGET_OPTION directly derive from TS_BASE.
18064 * tree-core.h (tree_optimization_option): Derive from tree_base.
18065 (tree_target_option): Likewise.
18067 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18069 * config/i386/i386.c (memory_address_length): Increase len
18070 only when rip_relative_addr_p returns false.
18072 2017-01-11 Julia Koval <julia.koval@intel.com>
18074 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18075 (OPTION_MASK_ISA_SGX_SET): New.
18076 (ix86_handle_option): Handle OPT_msgx.
18077 * config.gcc: Added sgxintrin.h.
18078 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18079 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18080 * config/i386/i386.c (ix86_target_string): Add -msgx.
18082 (ix86_option_override_internal): Handle new options.
18083 (ix86_valid_target_attribute_inner_p): Add sgx.
18084 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18085 * config/i386/i386.opt: Add msgx.
18086 * config/i386/sgxintrin.h: New file.
18087 * config/i386/x86intrin.h: Add sgxintrin.h.
18089 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18092 * fold-const.c (maybe_nonzero_address): Return 1 for function
18094 (tree_single_nonzero_warnv_p): Don't handle function local objects
18098 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18101 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18104 * opts.c: Include "spellcheck.h"
18105 (struct string_fragment): New struct.
18106 (struct edit_distance_traits<const string_fragment &>): New
18108 (get_closest_sanitizer_option): New function.
18109 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18111 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18113 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18115 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18116 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18117 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18118 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18119 for initial die_offset if dwarf_split_debug_info.
18120 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18121 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18123 (output_skeleton_debug_sections): Formatting fix. Use
18124 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18125 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18127 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18129 * config/arm/cortex-a53.md: Add bypasses for
18130 cortex_a53_r2f_cvt.
18131 (cortex_a53_r2f): Only use for transfers.
18132 (cortex_a53_f2r): Likewise.
18133 (cortex_a53_r2f_cvt): Add reservation for conversions.
18134 (cortex_a53_f2r_cvt): Likewise.
18136 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18138 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18139 to all inlined functions, change static to extern.
18141 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18144 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18146 (arm_assemble_integer): Likewise.
18148 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18150 * config.gcc: Use new awk script to check CPU, FPU and architecture
18151 parameters for --with-... options.
18152 * config/arm/parsecpu.awk: New file
18153 * config/arm/arm-cpus.in: New file.
18154 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18156 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18158 * config/arm/t-arm: Update dependency rules.
18159 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18160 of processing .def files.
18161 * config/arm/genopt.sh: Deleted.
18162 * config/arm/gentune.sh: Deleted.
18163 * config/arm/arm-cores.def: Deleted.
18164 * config/arm/arm-arches.def: Deleted.
18165 * config/arm/arm-fpus.def: Deleted.
18166 * config/arm/arm-tune.md: Regenerated.
18167 * config/arm/arm-tables.opt: Regenerated.
18168 * config/arm/arm-cpu.h: New generated file.
18169 * config/arm/arm-cpu-data.h: New generated file.
18170 * config/arm/arm-cpu-cdata.h: New generated file.
18172 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18175 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18177 (input_varpool_node): Unpack dynamically_initialized bit.
18179 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18181 PR rtl-optimization/79032
18182 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18183 the alignment of the adjusted memory reference against that of MODE,
18184 instead of the alignment of the original memory reference.
18186 2017-01-11 Martin Jambor <mjambor@suse.cz>
18188 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18190 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18191 decorated functions.
18193 2017-01-11 Richard Biener <rguenther@suse.de>
18195 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18196 set range/nonnull info for PHI results. Do not set it on
18197 stmts marked for removal.
18199 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18201 * expr.c (store_field): In the bitfield case, fetch the return value
18202 from the registers before applying a single big-endian adjustment.
18203 Always do a final load for a BLKmode value not larger than a word.
18205 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18208 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18209 that we correctly handle column numbers greater than
18210 LINE_MAP_MAX_COLUMN_NUMBER.
18212 2017-01-10 Martin Sebor <msebor@redhat.com>
18214 PR middle-end/78245
18215 * gimple-ssa-sprintf.c (get_destination_size): Call
18216 {init,fini}object_sizes.
18217 * tree-object-size.c (addr_object_size): Adjust.
18218 (pass_through_call): Adjust.
18219 (pass_object_sizes::execute): Adjust.
18220 * tree-object-size.h (fini_object_sizes): Declare.
18222 2017-01-10 Martin Sebor <msebor@redhat.com>
18224 PR tree-optimization/78775
18225 * builtins.c (get_size_range): Move...
18226 * calls.c: ...to here.
18227 (alloc_max_size): Accept zero argument.
18228 (operand_signed_p): Remove.
18229 (maybe_warn_alloc_args_overflow): Call get_size_range.
18230 * calls.h (get_size_range): Declare.
18232 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18234 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18235 from TI's devices.csv file as of September 2016.
18236 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18238 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18240 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18241 * doc/invoke.texi: Likewise.
18242 * doc/md.texi: Likewise.
18243 * doc/objc.texi: Likewise.
18245 2017-01-10 Joshua Conner <joshconner@google.com>
18247 * config/arm/fuchsia-elf.h: New file.
18248 * config/fuchsia.h: New file.
18249 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18250 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18252 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18254 2016-01-10 Richard Biener <rguenther@suse.de>
18256 PR tree-optimization/79034
18257 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18258 Propagate out degenerate PHIs in the joiner.
18260 2017-01-10 Martin Liska <mliska@suse.cz>
18262 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18263 (sort_congruence_classes_by_decl_uid): Likewise.
18264 (sort_congruence_class_groups_by_decl_uid): Likewise.
18265 (sem_item_optimizer::merge_classes): Sort class, groups in these
18266 classes and members in the groups by DECL_UID of declarations.
18267 This would make merge operations stable.
18269 2017-01-10 Martin Liska <mliska@suse.cz>
18271 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18272 usage of m_classes_vec.
18273 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18274 (sem_item_optimizer::get_group_by_hash): Likewise.
18275 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18276 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18277 (sem_item_optimizer::verify_classes): Likewise.
18278 (sem_item_optimizer::process_cong_reduction): Likewise.
18279 (sem_item_optimizer::dump_cong_classes): Likewise.
18280 (sem_item_optimizer::merge_classes): Likewise.
18281 * ipa-icf.h (congruence_class_hash): Rename from
18282 congruence_class_group_hash. Remove declaration of m_classes_vec.
18284 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18286 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18287 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18288 * config.gcc: Add avx512vpopcntdqintrin.h.
18289 * config/i386/avx512vpopcntdqintrin.h: New.
18290 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18291 * config/i386/i386-builtin-types.def: Add new types.
18292 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18293 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18294 __builtin_ia32_vpopcountq_v8di_mask): New.
18295 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18296 __AVX512VPOPCNTDQ__.
18297 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18298 (PTA_AVX512VPOPCNTDQ): Define.
18299 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18300 TARGET_AVX512VPOPCNTDQ_P): Define.
18301 * config/i386/i386.opt: Add mavx512vpopcntdq.
18302 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18303 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18305 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18307 PR middle-end/77484
18308 * predict.def (PRED_CALL): Set to 67.
18310 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18312 * expr.c (store_field): In the bitfield case, if the value comes from
18313 a function call and is of an aggregate type returned in registers, do
18314 not modify the field mode; extract the value in all cases if the mode
18315 is BLKmode and the size is not larger than a word.
18317 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18320 * config/i386/cpuid.h: Fix undefined behavior.
18322 2017-01-04 Jeff Law <law@redhat.com>
18324 PR tree-optimization/79007
18325 PR tree-optimization/67955
18326 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18327 conservative for pt.null when flag_non_call_exceptions is on.
18329 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18331 PR translation/79019
18332 PR translation/79020
18333 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18334 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18335 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18337 * config/avr/avr.opt (maccumulate-args): Likewise.
18338 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18339 * common.opt (freport-bug): Likewise.
18340 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18341 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18342 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18343 translatable string.
18344 * config/i386/i386.c (function_value_32): Likewise.
18345 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18346 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18348 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18349 * common/config/msp430/msp430-common.c (msp430_handle_option):
18351 * symtab.c (symtab_node::verify_base): Likewise.
18352 * opts.c (set_debug_level): Likewise.
18353 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18354 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18355 missing whitespace to translatable strings.
18356 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18357 * config/sh/superh.h: Likewise.
18358 * config/i386/xopintrin.h: Likewise.
18359 * config/i386/znver1.md: Likewise.
18360 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18361 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18362 * double-int.h (struct double_int): Likewise.
18363 * double-int.c (div_and_round_double): Likewise.
18364 * wide-int.cc: Likewise.
18365 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18366 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18367 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18368 (crossjumps_occurred): ... this.
18369 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18372 PR tree-optimization/78899
18373 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18374 returning bool return struct loop *, NULL for failure and the new
18376 (versionable_outer_loop_p): Don't version outer loop if it has
18377 dont_vectorized bit set.
18378 (tree_if_conversion): When versioning outer loop, ensure
18379 tree_if_conversion is performed also on the inner loop of the
18380 non-vectorizable outer loop copy.
18381 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18382 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18383 prevent vectorization of it.
18384 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18385 the outer loop vectorization of the non-scalar version is attempted
18386 before vectorization of the inner loop in scalar version. If
18387 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18388 vectorization of its inner loop.
18389 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18390 has 2 inner loops, rename also on edges from bb whose single pred
18391 is outer_loop->header. Fix typo in function comment.
18393 2017-01-09 Martin Sebor <msebor@redhat.com>
18396 * asan.c (asan_emit_stack_protection): Increase local buffer size
18397 to avoid snprintf truncation warning.
18399 2017-01-09 Andrew Pinski <apinski@cavium.com>
18401 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18402 to reference thunderx2t99 for the tuning structure
18403 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18405 (thunderx2t99_extra_costs): This.
18406 * config/aarch64/aarch64-tune.md: Regenerate.
18407 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18408 (vulcan_addrcost_table): This.
18409 (vulcan_regmove_cost): Rename to ...
18410 (thunderx2t99_regmove_cost): This.
18411 (vulcan_vector_cost): Rename to ...
18412 (thunderx2t99_vector_cost): this.
18413 (vulcan_branch_cost): Rename to ...
18414 (thunderx2t99_branch_cost): This.
18415 (vulcan_tunings): Rename to ...
18416 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18417 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18419 2017-01-09 Martin Jambor <mjambor@suse.cz>
18423 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18424 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18425 (propagate_vr_accross_jump_function): Use the above function for all
18426 value range computations for pass-through jump functions and type
18427 converasion from explicit value range values.
18428 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18429 parameters from TYPE_ARG_TYPES.
18430 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18431 (ipa_write_node_info): Stream type of the actual argument.
18432 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18434 2017-01-09 Martin Liska <mliska@suse.cz>
18437 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18438 (lookup_compiler): Do not show error message with have_E.
18440 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18442 PR tree-optimization/78938
18443 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18444 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18445 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18446 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18449 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18451 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18454 2017-01-09 Richard Biener <rguenther@suse.de>
18456 PR tree-optimization/78997
18457 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18458 name condition properly.
18460 2017-01-09 Richard Biener <rguenther@suse.de>
18463 * dwarf2out.c (is_cxx): New overload with context.
18464 (is_naming_typedef_decl): Use it.
18466 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18468 * invoke.texi (Option Summary): Correct spacing in option lists
18469 and add line breaks to fix over-long lines.
18471 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18473 PR middle-end/17660
18475 * extend.texi (Common Variable Attributes): Add xref to GCC
18476 Internals manual to explain mode attribute keywords.
18478 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18481 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18482 and Preprocessor Options.
18483 (Options for Linking): Document -pthread here....
18484 (RS/6000 and PowerPC Options): ...not here.
18485 (Solaris 2 Options): ...or here.
18486 * doc/cppopts.texi: Document -pthread.
18488 2017-01-08 Martin Sebor <msebor@redhat.com>
18490 PR middle-end/77708
18491 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18492 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18493 New member functions.
18494 (format_directive): Used them.
18496 (pass_sprintf_length::handle_gimple_call): Same.
18497 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18498 to avoid truncation for any argument.
18499 (extract_affine_mul): Same.
18500 * tree.c (get_file_function_name): Same.
18502 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18504 PR middle-end/77484
18505 * predict.def (PRED_INDIR_CALL): Set to 86.
18507 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18509 PR preprocessor/54124
18510 * doc/cppopts.texi: Reformat -d subtable to list the full name
18511 of the options. Add cross-reference to the docs for the general
18512 compiler -d options.
18513 * doc/invoke.texi (Developer Options): Add cross-reference to the
18514 preprocessor-specific -d option documentation.
18516 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18518 PR preprocessor/13498
18519 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18520 redudant material, and reflect new command-line options.
18521 (System Headers): Likewise.
18523 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18525 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18526 -isystem, and -idirafter. Copy-edit.
18527 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18528 default for -ftrack-macro-expansion. Delete obsolete and
18529 badly-formatted implementation details about -fdebug-cpp output.
18530 * doc/cppwarnopts.texi: Copy-edit.
18532 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18535 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18536 that the transition from a max line width >= 1<<10 to narrower
18537 lines works correctly.
18539 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18541 * doc/options.texi (PerFunction): New.
18542 * opt-functions.awk (switch_flags): Map both Optimization and
18543 PerFunction to CL_OPTIMIZATION.
18544 * opth-gen.awk: Test for PerFunction flag along with
18546 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18547 it only when the latter is present. Skip those that don't in
18548 the hash function generator.
18549 * common.opt (fvar-tracking): Mark as PerFunction instead of
18551 (fvar-tracking-assignments): Likewise.
18552 (fvar-tracking-assignments-toggle): Likewise.
18553 (fvar-tracking-uninit): Likewise.
18555 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18557 PR translation/79018
18558 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18561 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18564 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18565 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18566 TARGET_LONG_JUMP_TABLE_OFFSETS.
18567 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18568 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18569 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18570 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18571 * config/m68k/m68k.md (tablejump expander): Likewise.
18572 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18573 TARGET_LONG_JUMP_TABLE_OFFSETS.
18574 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18575 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18577 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18578 David Holsgrove <david.holsgrove@xilinx.com>
18580 * common/config/microblaze/microblaze-common.c
18581 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18582 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18584 * config/microblaze/microblaze.c (microblaze_must_save_register)
18585 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18587 (microblaze_eh_return): New function.
18588 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18589 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18590 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18591 * config/microblaze/microblaze.md (eh_return): New pattern.
18593 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18595 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18596 GCC_DIAGNOSTIC_STRINGIFY): Define.
18598 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18600 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18602 * config/arm/arm.md (<mcrr>): New.
18604 * config/arm/arm.c (arm_arch5te): New.
18605 (arm_option_override): Set arm_arch5te.
18606 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18608 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18609 (arm_mcrr_qualifiers): ... this. New.
18610 (MRRC_QUALIFIERS): Define to...
18611 (arm_mrrc_qualifiers): ... this. New.
18612 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18614 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18615 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18616 (MRRCI, mrrc, MRRC): New.
18617 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18618 VUNSPEC_MRRC2): New.
18620 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18622 * config/arm/arm.md (<mcr>): New.
18624 * config/arm/arm.c (arm_coproc_builtin_available): Add
18625 support for mcr, mrc, mcr2 and mrc2.
18626 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18627 (arm_mcr_qualifiers): ... this. New.
18628 (MRC_QUALIFIERS): Define to ...
18629 (arm_mrc_qualifiers): ... this. New.
18630 (MCR_QUALIFIERS): Define to ...
18631 (arm_mcr_qualifiers): ... this. New.
18632 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18634 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18635 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18636 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18637 VUNSPEC_MRC2): New.
18639 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18641 * config/arm/arm.md (*ldc): New.
18645 * config/arm/arm.c (arm_coproc_builtin_available): Add
18646 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18647 (arm_coproc_ldc_stc_legitimate_address): New.
18648 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18649 'qualifier_const_pointer'.
18650 (LDC_QUALIFIERS): Define to...
18651 (arm_ldc_qualifiers): ... this. New.
18652 (STC_QUALIFIERS): Define to...
18653 (arm_stc_qualifiers): ... this. New.
18654 * config/arm/arm-protos.h
18655 (arm_coproc_ldc_stc_legitimate_address): New.
18656 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18657 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18658 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18659 stc2, stcl, stc2l): New.
18660 * config/arm/constraints.md (Uz): New.
18661 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18662 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18663 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18664 VUNSPEC_STC2L): New.
18666 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18668 * config/arm/arm.md (<cdp>): New.
18669 * config/arm/arm.c (neon_const_bounds): Rename this ...
18670 (arm_const_bounds): ... this.
18671 (arm_coproc_builtin_available): New.
18672 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18673 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18674 (CDP_QUALIFIERS): Define to...
18675 (arm_cdp_qualifiers): ... this. New.
18677 (arm_expand_builtin_args): Add case for 6 arguments.
18678 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18679 (arm_const_bounds): ... this.
18680 (arm_coproc_builtin_available): New.
18681 * config/arm/arm_acle.h (__arm_cdp): New.
18683 * config/arm/arm_acle_builtins.def (cdp): New.
18685 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18686 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18687 'arm_const_bounds'.
18688 * config/arm/types.md (coproc): New.
18689 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18690 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18691 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18692 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18694 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18696 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18697 (UBINOP_QUALIFIERS): New.
18699 (acle_builtin_data): New. Change comment.
18700 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18701 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18702 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18703 arm_acle_builtins.def.
18704 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18705 (arm_init_acle_builtins): New.
18706 (CRC32_BUILTIN): Remove.
18707 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18708 crc32cb, crc32ch and crc32cw.
18709 (arm_init_crc32_builtins): Remove.
18710 (arm_init_builtins): Use arm_init_acle_builtins rather
18711 than arm_init_crc32_builtins.
18712 (arm_expand_acle_builtin): New.
18713 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18714 * config/arm/arm_acle_builtins.def: New.
18716 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18718 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18719 (arm_builtin_datum): ... this.
18720 (arm_init_neon_builtin): Rename to ...
18721 (arm_init_builtin): ... this. Add a new parameters PREFIX
18722 and USE_SIG_IN_NAME.
18723 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
18724 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
18725 'arm_builtin_datum'.
18726 (arm_init_vfp_builtins): Likewise.
18727 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
18728 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
18729 (arm_expand_neon_args): Rename to ...
18730 (arm_expand_builtin_args): ... this. Rename builtin_arg
18731 enum values and differentiate between ARG_BUILTIN_MEMORY
18732 and ARG_BUILTIN_NEON_MEMORY.
18733 (arm_expand_neon_builtin_1): Rename to ...
18734 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
18735 values, arm_expand_builtin_args and add bool parameter NEON.
18736 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
18737 (arm_expand_vfp_builtin): Likewise.
18738 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
18740 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18742 PR middle-end/77484
18743 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
18744 * predict.c (tree_estimate_probability_bb): Reverse direction of
18745 polymorphic call predictor.
18747 2017-01-06 David Malcolm <dmalcolm@redhat.com>
18749 * passes.c (execute_one_pass): Split out pass-skipping logic into...
18750 (determine_pass_name_match): ...this new function and...
18751 (should_skip_pass_p): ...this new function.
18753 2017-01-06 Nathan Sidwell <nathan@acm.org>
18755 * ipa-visibility.c (function_and_variable_visibility): Reformat
18756 comments and long lines. Remove extrneous if.
18757 * symtab.c (symtab_node::make_decl_local): Fix code format.
18758 (symtab_node::set_section_for_node): Fix comment typo.
18760 2017-01-06 Martin Liska <mliska@suse.cz>
18763 * lra-constraints.c: Rename invariant to lra_invariant.
18764 * predict.c (set_even_probabilities): Initialize e to NULL.
18766 2017-01-05 Martin Sebor <msebor@redhat.com>
18768 PR tree-optimization/78910
18769 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
18770 (format_integer): Correct off-by-one error in the handling
18771 of precision with negative numbers in signed conversions..
18773 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
18775 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
18777 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18779 PR tree-optimization/71016
18780 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
18781 factor_out_conditional_conversion. Formatting fix.
18782 (factor_out_conditional_conversion): Add cond_stmt argument.
18783 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
18784 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
18787 2017-01-05 David Malcolm <dmalcolm@redhat.com>
18789 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
18790 read-rtl-function.o, and selftest-rtl.o.
18791 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
18792 (selftest::aarch64_test_loading_full_dump): New function.
18793 (selftest::aarch64_run_selftests): New function.
18794 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
18795 selftest::aarch64_run_selftests.
18796 * config/i386/i386.c
18797 (selftest::ix86_test_loading_dump_fragment_1): New function.
18798 (selftest::ix86_test_loading_call_insn): New function.
18799 (selftest::ix86_test_loading_full_dump): New function.
18800 (selftest::ix86_test_loading_unspec): New function.
18801 (selftest::ix86_run_selftests): Call the new functions.
18802 * emit-rtl.c (maybe_set_max_label_num): New function.
18803 * emit-rtl.h (maybe_set_max_label_num): New decl.
18804 * function.c (instantiate_decls): Guard call to
18805 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
18806 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
18808 * gensupport.c (gen_reader::gen_reader): Pass "false"
18809 for new "compact" param of rtx_reader.
18810 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
18811 rather than an empty string for NULL strings.
18812 * read-md.c: Potentially include config.h rather than bconfig.h.
18813 Wrap include of errors.h with #ifdef GENERATOR_FILE.
18814 (have_error): New global, copied from errors.c.
18815 (md_reader::read_name): Rename to...
18816 (md_reader::read_name_1): ...this, adding "out_loc" param,
18817 and converting "missing name or number" to returning false, rather
18819 (md_reader::read_name): Reimplement in terms of read_name_1.
18820 (md_reader::read_name_or_nil): New function.
18821 (md_reader::read_string): Handle "(nil)" by returning NULL.
18822 (md_reader::md_reader): Add new param "compact".
18823 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
18824 (md_reader::read_file): New method.
18825 * read-md.h (md_reader::md_reader): Add new param "compact".
18826 (md_reader::read_file): New method.
18827 (md_reader::is_compact): New accessor.
18828 (md_reader::read_name): Convert return type from void to file_location.
18829 (md_reader::read_name_or_nil): New decl.
18830 (md_reader::read_name_1): New decl.
18831 (md_reader::m_compact): New field.
18832 (noop_reader::noop_reader): Pass "false" for new "compact" param
18834 (rtx_reader::rtx_reader): Add new "compact" param.
18835 (rtx_reader::read_rtx_operand): Make virtual and convert return
18836 type from void to rtx.
18837 (rtx_reader::read_until): New decl.
18838 (rtx_reader::handle_any_trailing_information): New virtual function.
18839 (rtx_reader::postprocess): New virtual function.
18840 (rtx_reader::finalize_string): New virtual function.
18841 (rtx_reader::m_in_call_function_usage): New field.
18842 (rtx_reader::m_reuse_rtx_by_id): New field.
18843 * read-rtl-function.c: New file.
18844 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
18845 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
18846 (selftest::verify_three_block_rtl_cfg): New decl.
18847 * read-rtl-function.h: New file.
18848 * read-rtl.c: Potentially include config.h rather than bconfig.h.
18849 For host, include function.h, memmodel.h, and emit-rtl.h.
18850 (one_time_initialization): New function.
18851 (struct compact_insn_name): New struct.
18852 (compact_insn_names): New array.
18853 (find_code): Handle insn codes in compact dumps.
18854 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
18855 (bind_subst_iter_and_attr): Likewise.
18856 (add_condition_to_string): Likewise.
18857 (add_condition_to_rtx): Likewise.
18858 (apply_attribute_uses): Likewise.
18859 (add_current_iterators): Likewise.
18860 (apply_iterators): Likewise.
18861 (initialize_iterators): Guard usage of apply_subst_iterator with
18862 #ifdef GENERATOR_FILE.
18863 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
18864 (md_reader::read_mapping): Likewise.
18865 (add_define_attr_for_define_subst): Likewise.
18866 (add_define_subst_attr): Likewise.
18867 (read_subst_mapping): Likewise.
18868 (check_code_iterator): Likewise.
18869 (rtx_reader::read_rtx): Likewise. Move one-time initialization
18871 (one_time_initialization): New function.
18872 (rtx_reader::read_until): New method.
18873 (read_flags): New function.
18874 (parse_reg_note_name): New function.
18875 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
18876 Handle reuse_rtx ids.
18877 Wrap iterator lookup within #ifdef GENERATOR_FILE.
18878 Add parsing support for RTL dumps, mirroring the special-cases in
18879 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
18880 values, and calling handle_any_trailing_information.
18881 (rtx_reader::read_rtx_operand): Convert return type from void
18882 to rtx, returning return_rtx. Handle case 'e'. Call
18883 finalize_string on XSTR and XTMPL fields.
18884 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
18885 "(nil)" values were omitted. Call the postprocess vfunc on the
18887 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
18888 class ctor. Initialize m_in_call_function_usage. Call
18889 one_time_initialization.
18890 * rtl-tests.c (selftest::test_uncond_jump): Call
18891 set_new_first_and_last_insn.
18892 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
18893 * selftest-rtl.c: New file.
18894 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
18895 (selftest::get_insn_by_uid): New decl.
18896 * selftest-run-tests.c (selftest::run_tests): Call
18897 read_rtl_function_c_tests.
18898 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
18899 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
18902 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
18904 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
18905 operands in a special way. Assert that pos+len <= mode precision.
18907 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18909 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
18910 3 argument Alias with unlimited for the negative form.
18911 (fno-vect-cost-model): Removed.
18913 2017-01-05 Martin Liska <mliska@suse.cz>
18915 * hsa-gen.c (gen_hsa_divmod): New function.
18916 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
18918 2017-01-05 Martin Liska <mliska@suse.cz>
18921 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
18924 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18926 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
18927 small constant length operands.
18929 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18931 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
18932 between loop iterations.
18934 2017-01-05 Martin Liska <mliska@suse.cz>
18937 * gimplify.c (gimplify_decl_expr): Compare to
18938 asan_poisoned_variables instread of checking flags.
18939 (gimplify_target_expr): Likewise.
18940 (gimplify_expr): Likewise.
18941 (gimplify_function_tree): Conditionally initialize
18942 asan_poisoned_variables.
18944 2017-01-04 Jeff Law <law@redhat.com>
18946 PR tree-optimizatin/78812
18947 * rtl.h (contains_mem_rtx_p): Prototype.
18948 * ifcvt.c (containts_mem_rtx_p): Move from here to...
18949 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
18950 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
18951 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
18952 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
18954 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18956 * input.c (assert_char_at_range): Default-initialize actual_range.
18958 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18960 * df-scan.c (df_ref_create_structure): Make regno unsigned,
18961 to match the caller.
18963 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18965 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
18966 insns after final jump in test to emit dummy move.
18968 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18970 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
18971 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
18973 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18975 * multiple_target.c (create_dispatcher_calls): Init e_next.
18976 * tree-ssa-loop-split.c (split_loop): Init border.
18977 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
18980 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
18985 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
18986 (altivec_register_operand): Do not return true if the operand
18987 contains a SUBREG mixing SImode and SFmode.
18988 (vsx_register_operand): Likewise.
18989 (vsx_reg_sfsubreg_ok): New predicate.
18990 (vfloat_operand): Do not return true if the operand contains a
18991 SUBREG mixing SImode and SFmode.
18992 (vint_operand): Likewise.
18993 (vlogical_operand): Likewise.
18994 (gpc_reg_operand): Likewise.
18995 (int_reg_operand): Likewise.
18996 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
18997 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
18998 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19000 (rs6000_emit_move_si_sf_subreg): New helper function.
19001 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19002 fixup SUBREGs involving SImode and SFmode.
19003 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19004 numbers for the new peephole2 optimization.
19005 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19006 the GLIBC math library that do AND/IOR/XOR operations on single
19007 precision floating point.
19008 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19009 target macros to say whether we need to avoid SUBREGs mixing
19011 (TARGET_ALLOW_SF_SUBREG): Likewise.
19012 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19013 (UNSPEC_SI_FROM_SF): Likewise.
19014 (iorxor): Change spacing.
19015 (and_ior_xor): New iterator for AND, IOR, and XOR.
19016 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19017 (movdi_from_sf_zero_ext): Likewise.
19018 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19019 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19020 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19021 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19022 (fms<mode>4): Likewise.
19023 (fnma<mode>4): Likewise.
19024 (fnms<mode>4): Likewise.
19025 (nfma<mode>4): Likewise.
19026 (nfms<mode>4): Likewise.
19028 2017-01-04 Marek Polacek <polacek@redhat.com>
19031 * doc/invoke.texi: Document -Wpointer-compare.
19033 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19035 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19038 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19039 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19042 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19045 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19046 documentation of the powerpc_popcntb_ok attribute.
19047 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19048 code to issue warning messages if a requested CPU configuration is
19049 not supported by the binary (assembler and loader) toolchain.
19050 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19051 made to define a built-in function that has been disabled.
19052 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19053 made to define a built-in function that has been disabled.
19054 (altivec_init_builtins): Add comment explaining why definition
19055 of the DST built-in functions is not preceded by an assertion
19056 check. Add assertions to prevent ICE if attempts are made to
19057 define an altivec predicate or an abs* built-in function that has
19059 (htm_init_builtins): Add comment explaining why definition of the
19060 htm built-in functions is not preceded by an assertion check.
19062 2017-01-04 Jeff Law <law@redhat.com>
19064 PR tree-optimizatin/67955
19065 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19066 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19067 the points-to solution does not include pt_null. Use DECL_PT_UID
19070 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19072 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19073 Use gen_int_mode instead of gen_lopwart for const_int operands.
19075 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19077 PR tree-optimization/71563
19078 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19079 out of range value - has low bits known to be zero.
19081 2017-01-04 Alan Modra <amodra@gmail.com>
19083 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19084 * configure: Regenerate.
19085 * config.in: Regenerate.
19087 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19090 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19091 a substring of the message, but strcmp with the whole message. Ifdef
19092 ENABLE_NLS, translate the message first using dgettext.
19094 2017-01-03 Jeff Law <law@redhat.com>
19096 PR tree-optimizatin/78856
19097 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19098 (mark_threaded_blocks): Remove code to truncate thread paths that
19099 cross multiple loop headers. Instead invalidate the cached loop
19100 iteration information and handle case of a thread path walking
19101 into an irreducible region.
19103 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19106 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19107 assertions. Add support for doing the signbit if the IEEE 128-bit
19108 floating point value is in a GPR.
19109 * config/rs6000/rs6000.md (Fsignbit): Delete.
19110 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19111 Update the length attribute if the value is in a GPR.
19112 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19113 the sign or zero extension instruction, since the value is always 0/1.
19114 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19117 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19118 extracting SImode to a GPR register so that we can generate a
19119 store, limit the vector to be in a traditional Altivec register
19120 for the vextuwrx instruction.
19122 2017-01-03 Ian Lance Taylor <iant@google.com>
19124 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19126 2017-01-03 Martin Sebor <msebor@redhat.com>
19128 PR tree-optimization/78696
19129 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19130 precision. Use MPFR for %f for greater fidelity. Correct handling
19132 (pass_sprintf_length::compute_format_length): Set width and precision
19133 specified by asrerisk to void_node for vararg functions.
19134 (try_substitute_return_value): Adjust dump output.
19136 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19138 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19140 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19142 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19143 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19144 -mlra/-mno-lra was passed to the compiler.
19146 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19148 PR rtl-optimization/65618
19149 * emit-rtl.c (try_split): Move initialization of "before" and
19150 "after" to just before the call to emit_insn_after_setloc.
19152 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19154 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19156 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19158 * dwarf2out.c (gen_enumeration_type_die): When
19159 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19161 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19163 PR tree-optimization/78965
19164 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19165 Change first argument from const call_info & to call_info &. For %n
19166 set info.nowrite to false.
19168 PR middle-end/78901
19169 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19170 possibly throwing calls.
19172 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19173 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19174 and fns handling, rather than in a separate case SSA_NAME.
19176 2017-01-02 Jeff Law <law@redhat.com>
19178 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19179 fixes for first_period and second_period variables.
19181 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19184 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19185 (*insvqi_1): New insn pattern.
19186 (*insvqi_1_mem_rex64): Ditto.
19187 (*insvqi_2): Ditto.
19188 (*insvqi_3): Rename from *insvqi.
19190 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19192 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19194 * doc/cfg.texi (Edges): Remove reference to Java.
19195 (Maintaining the CFG): Ditto.
19197 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19199 PR middle-end/77674
19200 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19201 transparent aliases.
19203 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19205 PR middle-end/77484
19206 * predict.def (PRED_CALL): Update hitrate.
19207 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19208 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19209 into direct/indirect/polymorphic variants.
19211 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19213 Update copyright years.
19215 * gcc.c (process_command): Update copyright notice dates.
19216 * gcov-dump.c (print_version): Ditto.
19217 * gcov.c (print_version): Ditto.
19218 * gcov-tool.c (print_version): Ditto.
19219 * gengtype.c (create_file): Ditto.
19220 * doc/cpp.texi: Bump @copying's copyright year.
19221 * doc/cppinternals.texi: Ditto.
19222 * doc/gcc.texi: Ditto.
19223 * doc/gccint.texi: Ditto.
19224 * doc/gcov.texi: Ditto.
19225 * doc/install.texi: Ditto.
19226 * doc/invoke.texi: Ditto.
19228 Copyright (C) 2017 Free Software Foundation, Inc.
19230 Copying and distribution of this file, with or without modification,
19231 are permitted in any medium without royalty provided the copyright
19232 notice and this notice are preserved.