1 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
3 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4 earlier and only if MASK_FPU is set. Adjust formatting.
6 2017-07-27 Martin Liska <mliska@suse.cz>
8 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11 2017-07-27 Martin Liska <mliska@suse.cz>
13 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14 BBs and edges seen by autoFDO.
16 2017-07-27 Richard Biener <rguenther@suse.de>
18 PR tree-optimization/81502
19 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20 with incompatible but same sized type.
21 (execute_update_addresses_taken): Likewise.
23 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
25 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
26 flag_tree_loop_vectorize rather than flag_tree_vectorize.
28 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
32 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
33 Change s_operand to memory_operand.
35 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
37 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
38 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
39 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
40 Emit instructions rather than returning an expression. Handle TFmode
41 and KFmode by casting to TImode.
42 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
43 (rs6000_emit_le_vsx_store): Likewise.
44 * config/rs6000/vsx.md (VSX_TI): New iterator.
45 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
46 (*vsx_le_undo_permute_<mode>): Likewise.
47 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
48 emit the split sequence.
49 (*vsx_le_perm_store_<mode>): Likewise.
51 2017-07-27 Jakub Jelinek <jakub@redhat.com>
53 PR tree-optimization/81555
54 PR tree-optimization/81556
55 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
56 if true, force CHANGED for the recursive invocation.
57 (reassociate_bb): Remember original length of ops array, pass
58 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
60 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
61 attributes for noipa attribute. For naked attribute use
62 lookup_attribute first before lookup_attribute_spec.
63 * final.c (rest_of_handle_final): Disable IPA RA for functions with
65 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
66 for functions with noipa attribute.
67 (cgraph_externally_visible_p): Return true for functions with noipa
69 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
70 for functions with noipa attribute.
71 * doc/extend.texi: Document noipa function attribute.
72 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
73 also for functions with noipa attribute.
74 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
76 2017-07-26 Andrew Pinski <apinski@cavium.com>
78 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
79 vec_unalign_load_cost and vec_unalign_store_cost.
81 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
83 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
84 -mvsx-small-integer option.
85 (ISA_3_0_MASKS_IEEE): Likewise.
86 (OTHER_VSX_VECTOR_MASKS): Likewise.
87 (POWERPC_MASKS): Likewise.
88 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
89 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
90 code, only testing for DImode being allowed in non-VSX floating
92 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
93 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
95 (rs6000_option_override_internal): Delete -mvsx-small-integer.
96 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
97 TARGET_P8_VECTOR test.
98 (rs6000_secondary_reload_simple_move): Likewise.
99 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
100 since TARGET_P9_VECTOR was already tested.
101 (rs6000_opt_masks): Remove -mvsx-small-integer.
102 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
103 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
105 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
106 test for TARGET_VEXTRACTUB was used, and that uses
108 (p9 extract splitter): Likewise.
109 (vsx_extract_<mode>_di_p9): Likewise.
110 (vsx_extract_<mode>_store_p9): Likewise.
111 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
112 for TARGET_P9_VECTOR was used. Delete code that is now dead with
113 the elimination of TARGET_VSX_SMALL_INTEGER.
114 (vsx_extract_<mode>_p8): Likewise.
115 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
116 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
117 (vsx_set_<mode>_p9): Likewise.
118 (vsx_set_v4sf_p9): Likewise.
119 (vsx_set_v4sf_p9_zero): Likewise.
120 (vsx_insert_extract_v4sf_p9): Likewise.
121 (vsx_insert_extract_v4sf_p9_2): Likewise.
122 * config/rs6000/rs6000.md (sign extend splitter): Change
123 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
124 (floatsi<mode>2_lfiwax_mem): Likewise.
125 (floatunssi<mode>2_lfiwzx_mem): Likewise.
126 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
127 since a test for TARGET_P9_VECTOR was used.
128 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
129 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
130 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
131 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
132 TARGET_P8_VECTOR test.
133 (fix_trunc<mode>si2_stfiwx): Likewise.
134 (fix_trunc<mode>si2_internal): Likewise.
135 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
136 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
138 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
139 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
140 TARGET_P8_VECTOR test.
141 (fixuns_trunc<mode>si2_stfiwx): Likewise.
142 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
143 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
145 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
146 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
147 since a test for TARGET_P9_VECTOR was used.
148 (splitter for loading small constants): Likewise.
150 2017-07-26 Andrew Pinski <apinski@cavium.com>
152 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
155 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
158 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
159 (fp_valid_at): Likewise.
161 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
163 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
164 (qdf24xx_addrcost_table): Likewise.
165 (cortexa57_tunings): Update to use generic_branch_cost.
166 (cortexa72_tunings): Likewise.
167 (cortexa73_tunings): Likewise.
168 (qdf24xx_tunings): Likewise.
170 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
172 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
173 (thunderx2t99_branch_cost): Likewise.
174 (cortexa35_tunings): Update to use generic_branch_cost.
175 (cortexa53_tunings): Likewise.
176 (cortexa57_tunings): Likewise.
177 (cortexa72_tunings): Likewise.
178 (cortexa73_tunings): Likewise.
179 (thunderx2t99_tunings): Likewise.
181 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
183 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
184 (sparc_option_override): Honour MASK_FSMULD.
185 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
186 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
187 * config/sparc/sparc.opt (mfsmuld): New option.
188 * doc/invoke.texi (mfsmuld): Document option.
190 2017-07-26 Marek Polacek <polacek@redhat.com>
193 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
195 2017-07-26 Richard Biener <rguenther@suse.de>
197 * gimple-match-head.c (do_valueize): Return OP if valueize
199 (get_def): New helper to get at the def stmt of a SSA name
201 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
202 do_valueize to get at the def stmt.
203 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
205 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
208 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
210 2017-07-26 Martin Liska <mliska@suse.cz>
213 * function.c (expand_function_start): Make expansion of
214 nonlocal_goto_save_area after parm_birth_insn.
216 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
218 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
219 from all CPU target flags enable members.
221 2017-07-26 Richard Biener <rguenther@suse.de>
223 * genmatch.c (dt_simplify::gen): Make iterator vars const.
224 (decision_tree::gen): Make 'type' const.
225 (write_predicate): Likewise.
227 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
229 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
230 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
231 (rs6000_option_override_internal): Likewise.
232 (rs6000_expand_vector_set): Likewise.
233 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
234 (TARGET_UPPER_REGS_SF): Likewise.
235 (TARGET_UPPER_REGS_DI): Likewise.
236 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
237 (TARGET_DIRECT_MOVE_64BIT): Likewise.
238 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
239 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
240 (Splitters for DI constants in Altivec registers): Likewise.
241 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
242 (vsx_set_v4sf_p9): Likewise.
243 (vsx_set_v4sf_p9_zero): Likewise.
244 (vsx_insert_extract_v4sf_p9): Likewise.
245 (vsx_insert_extract_v4sf_p9_2): Likewise.
247 2017-07-25 Carl Love <cel@us.ibm.com>
249 * doc/extend.texi: Update the built-in documentation file for the
250 existing built-in functions
251 vector signed char vec_cnttz (vector signed char);
252 vector unsigned char vec_cnttz (vector unsigned char);
253 vector signed short vec_cnttz (vector signed short);
254 vector unsigned short vec_cnttz (vector unsigned short);
255 vector signed int vec_cnttz (vector signed int);
256 vector unsigned int vec_cnttz (vector unsigned int);
257 vector signed long long vec_cnttz (vector signed long long);
258 vector unsigned long long vec_cnttz (vector unsigned long long);
260 2017-07-25 Andrew Pinski <apinski@cavium.com>
262 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
263 accesses where the use is for the first operand of a BIT_INSERT.
265 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
268 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
269 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
271 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
273 * config/i386/gstabs.h: Delete.
274 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
276 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
278 * config/i386/i386.c (ix86_decompose_address): Do not check for
279 register RTX when looking at index_reg or base_reg.
280 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
282 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
284 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
285 to update EH info here.
287 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
289 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
291 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
293 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
295 2017-07-25 Torsten Duwe <duwe@suse.de>
297 * common.opt: Introduce -fpatchable-function-entry
298 command line option, and its variables function_entry_patch_area_size
299 and function_entry_patch_area_start.
300 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
301 including a two-value parser.
302 * target.def (print_patchable_function_entry): New target hook.
303 * targhooks.h (default_print_patchable_function_entry): New function.
304 * targhooks.c (default_print_patchable_function_entry): Likewise.
305 * toplev.c (process_options): Switch off IPA-RA if
306 patchable function entries are being generated.
307 * varasm.c (assemble_start_function): Look at the
308 patchable-function-entry command line switch and current
309 function attributes and maybe generate NOP instructions by
310 calling the print_patchable_function_entry hook.
311 * doc/extend.texi: Document patchable_function_entry attribute.
312 * doc/invoke.texi: Document -fpatchable_function_entry
314 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
316 * doc/tm.texi: Re-generate.
318 2017-07-25 Jakub Jelinek <jakub@redhat.com>
321 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
322 TARGET_AVX512DQ rather than TARGET_AVX512BW.
324 2017-07-25 Tamar Christina <tamar.christina@arm.com>
326 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
328 2017-07-25 Richard Biener <rguenther@suse.de>
330 PR tree-optimization/81455
331 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
332 not walk in cycles when looking for guards.
334 2017-07-25 Richard Biener <rguenther@suse.de>
336 PR tree-optimization/81529
337 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
338 when optimizing backedge uses.
340 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
342 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
344 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
345 to dl_section_ref. On AIX, append an expression to subtract
346 the size of the section length to dl_section_ref.
348 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
350 * configure.ac: If any of the config.* scripts fail, exit 1.
351 * configure: Regenerate.
353 2017-07-25 Richard Biener <rguenther@suse.de>
356 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
357 of immediate uses, be more verbose on errors.
359 2017-07-25 Richard Biener <rguenther@suse.de>
361 PR tree-optimization/81510
362 * tree-vect-loop.c (vect_is_simple_reduction): When the
363 reduction stmt is not inside the loop bail out.
365 2017-07-25 Richard Biener <rguenther@suse.de>
367 PR tree-optimization/81303
368 * tree-vect-loop-manip.c (vect_loop_versioning): Build
369 profitability check against LOOP_VINFO_NITERSM1.
371 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
373 * domwalk.c (cmp_bb_postorder): Simplify.
374 (sort_bbs_postorder): New function. Use it...
375 (dom_walker::walk): ...here to optimize common cases.
377 2017-07-25 Martin Liska <mliska@suse.cz>
380 * ipa-visibility.c (function_and_variable_visibility): Make the
381 redirection just on target that supports aliasing.
382 Fix GNU coding style.
384 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
387 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
389 * config/arm/rtems.h: Likewise.
390 * config/bfin/rtems.h: Likewise.
391 * config/i386/rtemself.h: Likewise.
392 * config/lm32/rtems.h: Likewise.
393 * config/m32c/rtems.h: Likewise.
394 * config/m68k/rtemself.h: Likewise.
395 * config/microblaze/rtems.h: Likewise.
396 * config/mips/rtems.h: Likewise.
397 * config/moxie/rtems.h: Likewise.
398 * config/nios2/rtems.h: Likewise.
399 * config/powerpcspe/rtems.h: Likewise.
400 * config/rs6000/rtems.h: Likewise.
401 * config/rtems.h: Likewise.
402 * config/sh/rtems.h: Likewise.
403 * config/sh/rtemself.h: Likewise.
404 * config/sparc/rtemself.h: Likewise.
406 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
409 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
410 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
411 * tree-ssa-structalias.c (alias_get_name): Same.
413 2017-07-25 Bin Cheng <bin.cheng@arm.com>
416 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
417 instructions if no du chain is found.
419 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
421 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
423 2017-07-25 Richard Biener <rguenther@suse.de>
426 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
429 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
431 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
433 (ISA_2_7_MASKS_SERVER): Likewise.
434 (ISA_3_0_MASKS_IEEE): Likewise.
435 (OTHER_P8_VECTOR_MASKS): Likewise.
436 (OTHER_VSX_VECTOR_MASKS): Likewise.
437 (POWERPC_MASKS): Likewise.
438 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
439 duplicate list of options.
440 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
441 explicit -mupper-regs options.
442 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
443 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
444 alias for -mupper-regs-df.
445 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
446 (rs6000_init_hard_regno_mode_ok): Likewise.
447 (rs6000_option_override_internal): Likewise.
448 (rs6000_opt_masks): Likewise.
449 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
450 options in terms of whether -mvsx or -mpower8-vector was used.
451 (TARGET_UPPER_REGS_DI): Likewise.
452 (TARGET_UPPER_REGS_SF): Likewise.
453 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
454 -mupper-regs-* options.
456 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
458 * passes.c (emergency_dump_function): Print some empty lines and a
459 header before the RTL dump.
461 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
463 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
465 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
468 * config/aarch64/aarch64.c (aarch64_classify_symbol):
469 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
471 2017-07-24 Carl Love <cel@us.ibm.com>
473 * config/rs6000/rs6000-c.c: Add support for built-in functions
474 vector float vec_extract_fp32_from_shorth (vector unsigned short);
475 vector float vec_extract_fp32_from_shortl (vector unsigned short);
476 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
477 vec_extract_fp_from_shortl): Add defines for the two builtins.
478 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
479 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
481 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
482 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
483 * doc/extend.texi: Update the built-in documentation file for the
484 new built-in function.
486 2017-07-24 Jakub Jelinek <jakub@redhat.com>
489 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
491 * doc/generic.texi: Likewise.
492 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
493 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
495 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
497 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
498 (aarch64_mls_elt_merge<mode>): Likewise.
500 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
502 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
505 2017-07-23 Michael Collison <michael.collison@arm.com>
507 * config/arm/arm.c (arm_option_override): Deprecate
508 use of -mstructure-size-boundary.
509 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
510 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
512 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
515 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
516 Reduce cost estimate for direct moves.
518 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
521 * config/i386/i386.c (ix86_option_override_internal): Disable
522 BMI, BMI2 and TBM instructions for -m16.
524 2017-07-21 Carl Love <cel@us.ibm.com>
526 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
527 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
528 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
529 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
530 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
531 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
532 VMULOSW): New enum "unspec" values.
533 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
534 altivec_vmulosw): New patterns.
535 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
536 VMULOSW): Add definitions.
538 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
540 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
542 * config/aarch64/aarch64-options-extensions.def (rdma); New.
543 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
544 (AARCH64_FL_V8_1): Renumber.
545 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
546 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
547 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
548 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
549 rdma to feature modifiers list.
551 2017-07-21 Yury Gribov <tetra2005@gmail.com>
554 * ipa-visibility (function_and_variable_visibility): Convert
555 recursive PLT call to direct call if appropriate.
557 2017-07-21 Andrew Pinski <apinski@cavium.com>
559 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
560 operand 1 to see if the types precision matches.
561 * fold-const.c (operand_equal_p): Likewise.
563 2017-07-21 Richard Biener <rguenther@suse.de>
565 PR tree-optimization/81303
566 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
567 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
568 (vect_peeling_hash_get_lowest_cost): Adjust.
569 (vect_enhance_data_refs_alignment): Likewise. Use
570 vect_get_peeling_costs_all_drs to compute the penalty for no
571 peeling to match up costs.
573 2017-07-21 Richard Biener <rguenther@suse.de>
575 PR tree-optimization/81500
576 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
577 we didn't identify a reduction path.
579 2017-07-21 Tom de Vries <tom@codesourcery.com>
580 Cesar Philippidis <cesar@codesourcery.com>
582 PR gcov-profile/81442
583 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
586 2017-07-21 Tom de Vries <tom@codesourcery.com>
589 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
591 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
592 nvptx_override_options_after_change.
594 2017-07-21 Ulrich Drepper <drepper@redhat.com>
596 * dwarf2out.c (output_file_names): Avoid double testing for
599 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
601 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
603 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
605 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
607 (try_crossjump_to_edge): Do not punt on partitioned functions.
609 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
611 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
612 Put all BBs reachable only via paths crossing cold region to cold
614 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
616 2016-07-21 Richard Biener <rguenther@suse.de>
618 PR tree-optimization/81303
619 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
620 into account prologue and epilogue iterations when raising
621 min_profitable_iters to sth at least covering one vector iteration.
623 2017-07-21 Tamar Christina <tamar.christina@arm.com>
625 * config/arm/arm.c (arm_test_cpu_arch_dat):
628 2017-07-20 Nathan Sidwell <nathan@acm.org>
631 * tree.h (TYPE_METHODS): Delete.
632 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
633 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
634 (dbxout_type_methods): Scan TYPE_FIELDS.
635 (dbxout_type): Don't check TYPE_METHODS here.
636 * function.c (use_register_for_decl): Always ignore register for
637 class types when not optimizing.
638 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
639 * tree.c (free_lang_data_in_type): Stitch out member functions and
640 templates from TYPE_FIELDS.
641 (build_distinct_type_copy, verify_type_variant,
642 verify_type): Member fns are on TYPE_FIELDS.
643 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
644 * tree-pretty-print.c (dump_generic_node): Likewise.
646 2017-07-20 Jakub Jelinek <jakub@redhat.com>
649 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
650 V2TImode and V4TImode.
651 (ix86_expand_vector_extract): Likewise.
652 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
653 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
654 (ssescalarmode): Handle V4TImode and V2TImode.
655 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
656 (*vec_extractv2ti, *vec_extractv4ti): New insns.
657 (VEXTRACTI128_MODE): New mode iterator.
658 (splitter for *vec_extractv?ti first element): New.
659 (VEC_INIT_MODE): New mode iterator.
660 (vec_init<mode>): Consolidate 3 expanders into one using
661 VEC_INIT_MODE mode iterator.
663 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
665 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
666 non_spilled_static_chain_regno_p.
668 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
670 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
672 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
674 * bb-reorder.c (connect_traces): Allow copying of blocks within
677 2017-07-20 Richard Biener <rguenther@suse.de>
679 * gimple.h (gimple_phi_result): Add gphi * overload.
680 (gimple_phi_result_ptr): Likewise.
681 (gimple_phi_arg): Likewise. Adjust index assert to only
682 allow actual argument accesses rather than all slots available
684 (gimple_phi_arg_def): Add gphi * overload.
685 * tree-phinodes.c (make_phi_node): Initialize only actual
687 (resize_phi_node): Clear memory not covered by old node,
688 do not initialize excess argument slots.
689 (reserve_phi_args_for_new_edge): Initialize new argument slot
692 2017-07-20 Bin Cheng <bin.cheng@arm.com>
694 PR tree-optimization/81388
696 2016-07-21 Bin Cheng <bin.cheng@arm.com>
698 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
699 by removing computation of may_be_zero.
701 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
702 Tom de Vries <tom@codesourcery.com>
705 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
706 when gimple level profile disagrees with what RTL expander did.
708 2017-07-20 Richard Biener <rguenther@suse.de>
710 PR tree-optimization/61171
711 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
712 (vect_analyze_stmt): Add slp instance parameter.
713 (vectorizable_reduction): Likewise.
714 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
715 (vect_is_simple_reduction): Deal with chains not detected
716 as SLP reduction chain, specifically not properly associated
717 chains containing a mix of plus/minus.
718 (get_reduction_op): Remove.
719 (get_initial_defs_for_reduction): Simplify, pass in whether
720 this is a reduction chain, pass in the SLP node for the PHIs.
721 (vect_create_epilog_for_reduction): Get the SLP instance as
723 (vectorizable_reduction): Get the SLP instance as arg.
724 During analysis remember the SLP node with the PHIs in the
725 instance. Simplify getting at the vectorized reduction PHIs.
726 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
727 through SLP instance.
728 (vect_slp_analyze_operations): Likewise.
729 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
730 (vect_transform_stmt): Likewise.
732 2017-07-20 Tom de Vries <tom@codesourcery.com>
734 PR tree-optimization/81489
735 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
736 read of phi arg location to before loop that modifies phi.
738 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
740 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
743 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
746 * except.c (execute): Fix ordering issue.
748 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
750 PR rtl-optimization/81423
751 * combine.c (make_compound_operation_int): Don't try to optimize
752 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
754 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
756 PR rtl-optimization/81423
757 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
758 with a constant that is -1 in the truncated to mode.
760 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
762 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
763 (determine_unlikely_bbs): ... here.
764 * predict.h (propagate_unlikely_bbs_forward): Declare.
765 * cfgexpand.c (pass_expand::execute): Use it.
766 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
768 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
769 propagate_unlikely_bbs_forward.
771 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
774 * except.c (maybe_add_nop_after_section_switch): New function.
777 2017-07-19 Tom de Vries <tom@codesourcery.com>
779 * gimple.h (gimple_phi_set_arg): Make assert more strict.
781 2017-07-19 Tom de Vries <tom@codesourcery.com>
783 * gimple.h (gimple_phi_arg): Make assert more strict.
785 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
787 * config.gcc (powerpc*-*-*): Add mmintrin.h.
788 * config/rs6000/mmintrin.h: New file.
789 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
791 2017-07-19 Jakub Jelinek <jakub@redhat.com>
793 PR tree-optimization/81346
794 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
796 2017-07-19 Tom de Vries <tom@codesourcery.com>
798 * config/nvptx/nvptx.md (VECIM): Add V2DI.
800 2017-07-19 Tom de Vries <tom@codesourcery.com>
802 * config/nvptx/nvptx-modes.def: Add V2DImode.
803 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
804 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
805 (nvptx_output_mov_insn): Handle lack of mov.b128.
806 (nvptx_print_operand): Handle 'H' and 'L' codes.
807 (nvptx_vector_mode_supported): Allow V2DImode.
808 (nvptx_preferred_simd_mode): New function.
809 (nvptx_data_alignment): New function.
810 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
811 nvptx_preferred_simd_mode.
812 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
814 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
816 2017-07-19 Tom de Vries <tom@codesourcery.com>
818 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
819 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
820 (nvptx_vector_mode_supported): New function. Allow V2SImode.
821 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
822 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
823 (mov<VECIM>_insn): New define_insn.
824 (define_expand "mov<VECIM>): New define_expand.
826 2017-07-19 Tom de Vries <tom@codesourcery.com>
828 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
830 2017-07-19 Jakub Jelinek <jakub@redhat.com>
832 PR tree-optimization/81346
833 * fold-const.h (fold_div_compare, range_check_type): Declare.
834 * fold-const.c (range_check_type): New function.
835 (build_range_check): Use range_check_type.
836 (fold_div_compare): No longer static, rewritten into
837 a match.pd helper function.
838 (fold_comparison): Don't call fold_div_compare here.
839 * match.pd (X / C1 op C2): New optimization using fold_div_compare
842 2017-07-19 Nathan Sidwell <nathan@acm.org>
844 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
845 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
846 * tree.c (find_decls_types_r, verify_type): Use
847 TYPE_{MIN,MAX}_VALUE_RAW.
848 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
849 (hash_tree): Likewise.
850 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
852 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
855 2017-07-18 Tom de Vries <tom@codesourcery.com>
858 * omp-expand.c (expand_omp_for_static_chunk): Handle
859 equal-argument loop exit phi.
861 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
864 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
865 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
867 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
869 (ror,rol -> rorx splitters): Use const_int_operand as
872 2017-06-18 Richard Biener <rguenther@suse.de>
874 PR tree-optimization/81410
875 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
876 the gap in the ! slp_perm SLP case after each group.
878 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
881 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
884 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
887 * predict.c (set_even_probabilities): Cleanup; do not affect
888 probabilities that are already known.
889 (combine_predictions_for_bb): Call even when count is set.
891 2017-07-18 Nathan Sidwell <nathan@acm.org>
893 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
896 2017-07-18 Bin Cheng <bin.cheng@arm.com>
899 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
900 optimization for loop niter analysis.
902 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
905 * config/avr/avr.c (avr_optimize_casesi): Don't use
906 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
908 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
910 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
911 body_cost_vec from _vect_peel_extended_info.
912 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
913 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
916 2017-07-18 Bin Cheng <bin.cheng@arm.com>
918 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
920 2017-07-18 Richard Biener <rguenther@suse.de>
922 PR tree-optimization/80620
923 PR tree-optimization/81403
924 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
925 info when re-using a VN table entry.
927 2017-07-18 Richard Biener <rguenther@suse.de>
929 PR tree-optimization/81418
930 * tree-vect-loop.c (vectorizable_reduction): Properly compute
931 vectype_in. Verify that with lane-reducing reduction operations
932 we have a single def-use cycle.
934 2017-07-17 Carl Love <cel@us.ibm.com>
936 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
938 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
939 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
940 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
941 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
942 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
943 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
944 VMULOSW): New enum "unspec" values.
945 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
946 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
947 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
948 altivec_vmulosw): New patterns.
949 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
950 VMULOSW): Add definitions.
951 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
953 * config/alpha/alpha.c: Include predict.h.
955 2017-07-17 Yury Gribov <tetra2005@gmail.com>
957 * tree-vrp.c (compare_assert_loc): Fix comparison function
958 to return predictable results.
960 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
962 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
965 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
966 * doc/invoke.texi (mexpand-adddi): Update text.
968 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
970 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
971 that also clobbers the CC register. The old expand code is moved
973 (*arc_clzsi2): ... here.
974 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
975 the CC register. The old expand code is moved to ...
976 (arc_ctzsi2): ... here.
978 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
980 * config/arc/arc.opt (mindexed-loads): Use initial value
981 TARGET_INDEXED_LOADS_DEFAULT.
982 (mauto-modify-reg): Use initial value
983 TARGET_AUTO_MODIFY_REG_DEFAULT.
984 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
985 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
986 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
987 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
989 2017-07-17 Martin Liska <mliska@suse.cz>
992 * opts.c (finish_options): Do not allow -fgnu-tm
993 w/ -fsanitize={kernel-,}address. Say sorry.
995 2017-07-17 Bin Cheng <bin.cheng@arm.com>
998 * tree-loop-distribution.c (classify_partition): Only assert on
1000 (merge_dep_scc_partitions): Delete prameter. Update function call.
1001 (distribute_loop): Remove code handling loop with unknown niters.
1002 (pass_loop_distribution::execute): Skip loop with unknown niters.
1004 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1007 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1008 function sort_partitions_by_post_order.
1010 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1012 PR tree-optimization/81374
1013 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1014 the max index of basic blocks, rather than number of basic blocks.
1016 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1018 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1020 (arc_legitimate_pic_operand_p): Likewise.
1021 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1023 (arc_needs_pcl_p): Likewise.
1024 (arc_legitimate_pc_offset_p): Likewise.
1025 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1026 function is also used in constrains.md.
1027 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1028 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1029 PLUS. Only return true/false in known cases, otherwise assert.
1030 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1031 is already called in arc_legitimate_constant_p.
1032 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1034 (LEGITIMATE_PIC_OPERAND_P): Use
1035 arc_raw_symbolic_reference_mentioned_p function.
1036 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1041 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1042 Andrew Burgess <andrew.burgess@embecosm.com>
1044 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1045 (arc_return_address_register): New function.
1046 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1047 (arc_handle_fndecl_attribute): Add naked attribute.
1048 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1049 (TARGET_WARN_FUNC_RETURN): Likewise.
1050 (arc_allocate_stack_slots_for_args): New function.
1051 (arc_warn_func_return): Likewise.
1052 (machine_function): Change type fn_type.
1053 (arc_compute_function_type): Consider new naked function type,
1054 change function return type.
1055 (arc_must_save_register): Adapt to handle new
1056 arc_compute_function_type's return type.
1057 (arc_expand_prologue): Likewise.
1058 (arc_expand_epilogue): Likewise.
1059 (arc_return_address_regs): Delete.
1060 (arc_return_address_register): New function.
1061 (arc_epilogue_uses): Use above function.
1062 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1063 (arc_function_type): Change encoding, add naked type.
1064 (ARC_INTERRUPT_P): Change to handle the new encoding.
1065 (ARC_FAST_INTERRUPT_P): Likewise.
1066 (ARC_NORMAL_P): Define.
1067 (ARC_NAKED_P): Likewise.
1068 (arc_compute_function_type): Delete prototype.
1069 * config/arc/arc.md (in_ret_delay_slot): Use
1070 arc_return_address_register function.
1071 (simple_return): Likewise.
1072 (p_return_i): Likewise.
1074 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1076 PR tree-optimization/81428
1077 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1078 can't be built for those types.
1080 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1082 Remove stuff dead since r239246.
1084 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1085 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1086 (avr_inform_devices): Remove dead stuff.
1088 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1090 * config/arm/arm_neon.h: Fix softp typo.
1092 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1094 PR tree-optimization/81365
1095 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1096 aggregate moves onto bb predecessor edges, make sure there are no
1097 loads that could alias the lhs in between the start of bb and the
1100 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1103 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1104 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1105 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1107 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1109 PR tree-optimization/81396
1110 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1111 (init_symbolic_number): Initialize it to 1.
1112 (perform_symbolic_merge): Add n_ops from both operands into the new
1114 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1115 without base_addr as useless if they need more than one operation.
1116 (bswap_replace): Handle !bswap case for NULL base_addr.
1118 2017-07-17 Tom de Vries <tom@codesourcery.com>
1121 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1124 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1126 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1127 conditional builtin define __FIX_LEON3FT_B2BST.
1129 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1131 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1132 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1135 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1137 PR rtl-optimization/81424
1138 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1139 to remove potential trapping from operands if -fnon-call-exceptions.
1141 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1143 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1144 profile_proability for scalling.
1145 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1147 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1149 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1151 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1153 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1154 fixpoint arithmetics.
1156 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1158 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1159 fixpoint arithmetics.
1161 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1163 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1164 fixpoint arithmetics.
1166 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1168 * profile-count.h (profile_probability::from_reg_br_prob_note,
1169 profile_probability::to_reg_br_prob_note): New functions.
1170 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1171 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1172 * predict.c (probability_reliable_p): Update.
1173 (edge_probability_reliable_p): Update.
1174 (br_prob_note_reliable_p): Update.
1175 (invert_br_probabilities): Update.
1176 (add_reg_br_prob_note): New function.
1177 (combine_predictions_for_insn): Update.
1178 * asan.c (asan_clear_shadow): Update.
1179 * cfgbuild.c (compute_outgoing_frequencies): Update.
1180 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1181 (update_br_prob_note): Update.
1182 (rtl_verify_edges): Update.
1183 (purge_dead_edges): Update.
1184 (fixup_reorder_chain): Update.
1185 * emit-rtl.c (try_split): Update.
1186 * ifcvt.c (cond_exec_process_insns): Update.
1187 (cond_exec_process_if_block): Update.
1188 (dead_or_predicable): Update.
1189 * internal-fn.c (expand_addsub_overflow): Update.
1190 (expand_neg_overflow): Update.
1191 (expand_mul_overflow): Update.
1192 * loop-doloop.c (doloop_modify): Update.
1193 * loop-unroll.c (compare_and_jump_seq): Update.
1194 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1195 * predict.h: Update.
1196 * reorg.c (mostly_true_jump): Update.
1198 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1199 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1200 * config/arc/arc.c: (emit_unlikely_jump): Update.
1201 * config/arm/arm.c: (emit_unlikely_jump): Update.
1202 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1203 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1204 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1205 (ix86_print_operand): Update.
1206 (ix86_split_fp_branch): Update.
1207 (predict_jump): Update.
1208 * config/ia64/ia64.c (ia64_print_operand): Update.
1209 * config/mmix/mmix.c (mmix_print_operand): Update.
1210 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1211 (rs6000_expand_split_stack_prologue): Update.
1212 * config/rs6000/rs6000.c: Update.
1213 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1214 (s390_expand_vec_movstr): Update.
1215 (s390_expand_cs_tdsi): Update.
1216 (s390_expand_split_stack_prologue): Update.
1217 * config/sh/sh.c (sh_print_operand): Update.
1218 (expand_cbranchsi4): Update.
1219 (expand_cbranchdi4): Update.
1220 * config/sparc/sparc.c (output_v9branch): Update.
1221 * config/spu/spu.c (get_branch_target): Update.
1222 (ea_load_store_inline): Update.
1223 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1224 * config/tilepro/tilepro.c: Update.
1226 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1228 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1229 (gimplify_save_expr): Likewise.
1231 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1233 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1235 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1237 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1239 * timevar.def (TV_IPA_FNSUMMARY): Define.
1241 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1243 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1244 to back store errata sensitive sequence from being generated.
1245 (sqrtdf2_fix): Likewise.
1247 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1249 * tree-ssa-threadupdate.c (compute_path_counts,
1250 update_joiner_offpath_counts): Use profile_probability.
1252 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1255 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1257 * config/arm/arm-c.c (arm_cpu_builtins): Define
1258 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1260 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1262 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1263 array entries to represent __ieee128 versions of the
1264 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1265 scalar_extract_sig, and scalar_insert_exp built-in functions.
1266 (altivec_resolve_overloaded_builtin): Add special case handling
1267 for the __builtin_scalar_insert_exp function, as represented by
1268 the P9V_BUILTIN_VEC_VSIEDP constant.
1269 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1270 exponent support for __ieee128 argument.
1271 (VSESQP): Add scalar extract signature support for __ieee128
1273 (VSTDCNQP): Add scalar test negative support for __ieee128
1275 (VSIEQP): Add scalar insert exponent support for __int128 argument
1276 with __ieee128 result.
1277 (VSIEQPF): Add scalar insert exponent support for __ieee128
1278 argument with __ieee128 result.
1279 (VSTDCQP): Add scalar test data class support for __ieee128
1281 (VSTDCNQP): Add overload support for scalar test negative with
1283 (VSTDCQP): Add overload support for scalar test data class
1285 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1287 (UNSPEC_VSX_SIEXPQP): New constant.
1288 (xsxexpqp): New insn for VSX scalar extract exponent quad
1290 (xsxsigqp): New insn for VSX scalar extract significand quad
1292 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1293 precision with floating point argument.
1294 (xststdcqp): New expand for VSX scalar test data class quad
1296 (xststdcnegqp): New expand for VSX scalar test negative quad
1298 (xststdcqp): New insn to match expansions for VSX scalar test data
1299 class quad precision and VSX scalar test negative quad precision.
1300 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1301 special case operand checking to enforce that second operand of
1302 VSX scalar test data class with quad precision argument is a 7-bit
1304 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1305 prototypes and descriptions of __ieee128 versions of
1306 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1307 scalar_test_data_class, and scalar_test_neg built-in functions.
1309 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1311 PR tree-optimization/81162
1312 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1313 replace a negate with an add.
1315 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1317 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1319 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1321 * config/arm/arm-c.c (arm_cpu_builtins): Define
1322 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1324 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1326 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1327 (armv8-r): Set ARM Cortex-R52 as default CPU.
1328 * config/arm/arm-tables.opt: Regenerate.
1329 * config/arm/arm-tune.md: Regenerate.
1330 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1332 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1333 extension for -mcpu=cortex-r52.
1335 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1337 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1338 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1339 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1340 (fp-armv8): Define it as FP_ARMv8 only.
1341 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1342 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1344 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1345 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1346 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1347 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1348 than TARGET_FPU_ARMV8.
1349 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1350 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1351 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1353 * config/arm/neon.md (neon_vrint): Likewise.
1354 (neon_vcvt): Likewise.
1355 (neon_<fmaxmin_op><mode>): Likewise.
1356 (<fmaxmin><mode>3): Likewise.
1357 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1358 * config/arm/predicates.md (arm_cond_move_operator): Check against
1359 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1361 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1363 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1366 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1368 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1369 loop in comment with memset.
1371 2017-07-14 Martin Liska <mliska@suse.cz>
1373 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1374 * dwarf2out.c (is_java): Remove the function.
1375 (output_pubname): Remove usage of the function.
1376 (lower_bound_default): Remove usage of DW_LANG_Java.
1377 (gen_compile_unit_die): Likewise.
1378 * gcc.c: Remove compiler defaults for .java and .zip files.
1379 * gimple-expr.c (remove_suffix): Change as there's no longer
1380 extension than 4-letter one.
1381 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1382 (gimplify_save_expr): Likewise.
1383 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1384 as it's possible even for other languages than Java.
1385 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1386 * lto-opts.c (lto_write_options): Remove reference to Java.
1387 * opts.c (strip_off_ending): Update file extension handling.
1388 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1389 * tree-eh.c (lower_resx): Likewise.
1390 * tree.c (free_lang_data_in_type): Remove dead code.
1391 (find_decls_types_r): Likewise.
1392 (build_common_builtin_nodes): Remove Java from a comment.
1393 (verify_type): Remove dead code.
1394 * varasm.c (assemble_external): Remove Java from a comment.
1396 2017-07-14 Martin Liska <mliska@suse.cz>
1398 * opts.c (finish_options): Add quotes.
1399 (common_handle_option): Likewise.
1401 2017-07-14 Martin Liska <mliska@suse.cz>
1403 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1404 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1406 * dwarf2out.c (lower_bound_default): Do not handle
1408 (gen_compile_unit_die): Likewise.
1409 * gcc.c: Remove default extension binding for GNU Pascal.
1410 * stmt.c: Remove Pascal language from a comment.
1411 * xcoffout.c: Likewise.
1413 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1416 * diagnostic-show-locus.c (fixit_cmp): New function.
1417 (layout::layout): Sort m_fixit_hints.
1418 (column_range::column_range): Assert that the values are valid.
1419 (struct char_span): New struct.
1420 (correction::overwrite): New method.
1421 (struct source_line): New struct.
1422 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1423 calls in terms of classes source_line and char_span, and
1424 correction::overwrite.
1425 (selftest::test_overlapped_fixit_printing_2): New function.
1426 (selftest::diagnostic_show_locus_c_tests): Call it.
1428 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1430 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1431 early if there is no lhs.
1433 2017-07-13 Martin Liska <mliska@suse.cz>
1435 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1436 (gen_reference_type_die): Likewise.
1437 * stor-layout.c: Remove Pascal-related comment.
1439 2017-07-13 Martin Liska <mliska@suse.cz>
1441 * opts.c (finish_options): Add quotes to error messages.
1442 (parse_sanitizer_options): Likewise.
1444 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1446 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1448 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1450 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1452 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1454 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1456 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1458 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1461 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1462 provides the hardware capability bits, define the macro
1463 __BUILTIN_CPU_SUPPORTS__.
1464 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1465 if GLIBC does not provide the hardware capability bits. Add a
1466 gcc_unreachable call if the built-in cpu function is neither
1467 __builtin_cpu_is nor __builtin_cpu_supports.
1468 (rs6000_get_function_versions_dispatcher): Change the warning
1469 that an old GLIBC is used which does not export the capability
1470 bits to be an error.
1471 * doc/extend.texi (target_clones attribute): Document the
1472 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1473 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1474 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1475 the macros defined by GCC if the newer GLIBC is available.
1477 2017-07-12 Jeff Law <law@redhat.com>
1479 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1480 remaining includes slightly.
1481 * config/riscv/riscv-builtins.c: Include profile-count.h.
1483 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1486 * config/avr/avr.c (avr_set_current_function): In diagnostic
1487 messages: Quote keywords and (parts of) identifiers.
1488 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1491 2017-07-12 Carl Love <cel@us.ibm.com>
1493 * config/rs6000/rs6000-c.c: Add support for built-in functions
1494 vector bool char vec_revb (vector bool char);
1495 vector bool short vec_revb (vector short char);
1496 vector bool int vec_revb (vector bool int);
1497 vector bool long long vec_revb (vector bool long long);
1498 * doc/extend.texi: Update the built-in documentation file for the
1499 new built-in functions.
1501 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1503 * config/s390/s390.md: Remove movcc splitter.
1505 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1507 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1508 load/store on condition.
1510 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1513 * config/avr/avr.c (avr_encode_section_info)
1514 [progmem && !TREE_READONLY]: Error if progmem object needs
1517 2017-07-11 Michael Collison <michael.collison@arm.com>
1519 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1522 2017-07-11 Carl Love <cel@us.ibm.com>
1524 * config/rs6000/rs6000-c.c: Add support for builtins
1525 vector unsigned int vec_parity_lsbb (vector signed int);
1526 vector unsigned int vec_parity_lsbb (vector unsigned int);
1527 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1528 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1529 vector unsigned long long vec_parity_lsbb (vector signed long long);
1530 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1531 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1532 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1533 * doc/extend.texi: Update the built-in documentation file for the
1534 new built-in functions.
1536 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1538 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1539 (layout::m_primary_loc): New field.
1540 (layout::layout): Initialize new field. Move location filtering
1541 logic from here to...
1542 (layout::maybe_add_location_range): ...this new method. Add
1543 support for filtering to just the lines already specified by other
1545 (layout::will_show_line_p): New method.
1546 (gcc_rich_location::add_location_if_nearby): New method.
1547 (selftest::test_add_location_if_nearby): New test function.
1548 (selftest::diagnostic_show_locus_c_tests): Call it.
1549 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1552 2017-07-11 Tom de Vries <tom@codesourcery.com>
1554 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1555 (bb_first_real_insn): New function.
1556 (nvptx_single): Add extra initialization of broadcasted condition
1559 2017-07-11 Nathan Sidwell <nathan@acm.org>
1561 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1563 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1565 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1566 Binutils doc as TEXI will mess them up.
1567 * doc/invoke.texi (AVR Options): Same here.
1569 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1571 * config/sparc/sparc.opt (mfix-ut700): New option.
1572 (mfix-gr712rc): Likewise.
1573 (sparc_fix_b2bst): New variable.
1574 * doc/invoke.texi (SPARC options): Document them.
1575 (ARM options): Fix warnings.
1576 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1577 instructions to prevent sequences that can trigger the store-store
1578 errata for certain LEON3FT processors.
1579 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1580 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1581 * config/sparc/sparc.md (fix_b2bst): New attribute.
1582 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1584 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1587 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1589 (*rsqrtsf2_sse): Ditto.
1591 (div<mode>3): Macroize insn from divdf3 and divsf3
1592 using MODEF mode iterator.
1594 2017-07-10 Martin Sebor <msebor@redhat.com>
1596 PR tree-optimization/80397
1597 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1598 instead of testing for equality to INTEGER_TYPE.
1600 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1602 * config.gcc: Remove uclibc from arc target spec.
1604 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1606 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1608 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1611 * lto-wrapper.c (remove_option): New function.
1612 (merge_and_complain): Merge PIC/PIE options more realistically.
1614 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1616 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1620 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1621 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1622 * config.in: Regenerate.
1623 * configure: Regenerate.
1624 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1625 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1626 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1627 (TARGET_GASISR_PROLOGUES): ...target mask.
1628 * common/config/avr/avr-common.c
1629 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1630 Set -mgas-isr-prologues.
1631 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1632 INSERT_PASS_BEFORE for it.
1633 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1634 * config/avr/avr.c (avr_option_override)
1635 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1636 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1637 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1638 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1639 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1641 (make_avr_pass_pre_proep): New function.
1642 (emit_push_sfr) <treg>: Add argument to function and use it
1644 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1645 and set machine->gasisr.yes.
1646 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1647 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1648 __gcc_isr.n_pushed to .L__stack_usage.
1649 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1650 (avr_asm_final_postscan_insn): ...this new static function.
1651 * config/avr/avr.h (machine_function)
1652 <is_no_gccisr, use_L__stack_usage>: New fields.
1653 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1654 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1655 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1656 (gasisr, *gasisr): New expander and insn.
1657 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1658 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1659 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1661 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1663 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1666 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1668 Move jump-tables out of .text again.
1671 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1672 (ASM_OUTPUT_ADDR_VEC): New function.
1673 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1674 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1675 INSN_ADDRESSes as asm comment.
1676 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1677 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1678 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1679 * config/avr/avr.md (*tablejump): Adjust comment.
1680 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1681 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1683 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1684 (avr_output_addr_vec): New proto.
1685 (avr_log_t) <insn_addresses>: New field.
1687 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1690 * config/i386/i386.c (ix86_function_arg_advance): Set
1691 outgoing_args_on_stack to true if there are outgoing arguments
1693 (ix86_function_arg): Likewise.
1694 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1695 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1696 * config/i386/i386.h (machine_function): Add
1697 outgoing_args_on_stack.
1699 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1701 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1703 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1705 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1707 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1708 (REAL_H): Remove $(MACHMODE_H).
1709 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1711 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1712 $(MACHMODE_H) and double-int.h.
1713 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1715 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1718 2017-07-07 Andrew Pinski <apinski@cavium.com>
1720 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1721 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1723 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1725 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1726 Add warning if GCC was not configured to link against a GLIBC that
1727 exports the hardware capability bits.
1728 (make_resolver_func): Make resolver function private and not a
1729 COMDAT function. Create the name with clone_function_name instead
1730 of make_unique_name.
1733 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1734 correct operand in doing the split.
1736 2017-07-07 Carl Love <cel@us.ibm.com>
1738 * config/rs6000/rs6000-c: Add support for built-in function
1739 vector unsigned short vec_pack_to_short_fp32 (vector float,
1741 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1742 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1743 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1744 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1745 (convert_4f32_8i16): Add define_expand.
1746 * doc/extend.texi: Update the built-in documentation file for the
1747 new built-in function.
1749 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1751 * config/sparc/m8.md: New file.
1752 * config/sparc/sparc.md: Include m8.md.
1754 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1756 * config/sparc/sparc.opt: New option -mvis4b.
1757 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
1758 (sparc_option_override): Handle VIS4B.
1759 (enum sparc_builtins): Define
1760 SPARC_BUILTIN_DICTUNPACK{8,16,32},
1761 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
1762 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
1763 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
1764 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
1765 (check_constant_argument): New function.
1766 (sparc_vis_init_builtins): Define builtins
1767 __builtin_vis_dictunpack{8,16,32},
1768 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
1769 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
1770 __builtin_vis_fpcmpde{8,16,32}shl and
1771 __builtin_vis_fpcmpur{8,16,32}shl.
1772 (sparc_expand_builtin): Check that the constant operands to
1773 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
1774 constant and in range.
1775 * config/sparc/sparc-c.c (sparc_target_macros): Handle
1777 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
1778 (SPARC_IMM5_P): Likewise.
1779 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
1780 (enabled): Handle vis4b.
1781 (UNSPEC_DICTUNPACK): New unspec.
1782 (UNSPEC_FPCMPSHL): Likewise.
1783 (UNSPEC_FPUCMPSHL): Likewise.
1784 (UNSPEC_FPCMPDESHL): Likewise.
1785 (UNSPEC_FPCMPURSHL): Likewise.
1786 (cpu_feature): New CPU feature `vis4b'.
1787 (dictunpack{8,16,32}): New insns.
1788 (FPCSMODE): New mode iterator.
1789 (fpcscond): New code iterator.
1790 (fpcsucond): Likewise.
1791 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
1792 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
1793 (fpcmpde{8,16,32}{si,di}shl): Likewise.
1794 (fpcmpur{8,16,32}{si,di}shl): Likewise.
1795 * config/sparc/constraints.md: Define constraints `q' for unsigned
1796 2-bit integer constants and `t' for unsigned 5-bit integer
1798 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
1800 (imm5_operand_dictunpack16): Likewise.
1801 (imm5_operand_dictunpack32): Likewise.
1802 (imm2_operand): Likewise.
1803 * doc/invoke.texi (SPARC Options): Document -mvis4b.
1804 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
1805 ditunpack* and fpcmp*shl builtins.
1807 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1809 * config.gcc: Handle m8 in --with-{cpu,tune} options.
1810 * config.in: Add HAVE_AS_SPARC6 define.
1811 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
1813 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
1815 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
1816 (CPP_CPU_SPEC): Handle m8.
1817 (ASM_CPU_SPEC): Likewise.
1818 * config/sparc/sparc-opts.h (enum processor_type): Add
1820 * config/sparc/sparc.c (m8_costs): New struct.
1821 (sparc_option_override): Handle TARGET_CPU_m8.
1822 (sparc32_initialize_trampoline): Likewise.
1823 (sparc64_initialize_trampoline): Likewise.
1824 (sparc_issue_rate): Likewise.
1825 (sparc_register_move_cost): Likewise.
1826 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
1827 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
1828 (ASM_CPU64_DEFAULT_SPEC): Likewise.
1829 (CPP_CPU_SPEC): Handle M8.
1830 (ASM_CPU_SPEC): Likewise.
1831 (AS_M8_FLAG): Define.
1832 * config/sparc/sparc.md: Add m8 to the cpu attribute.
1833 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
1834 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
1836 * configure: Regenerate.
1837 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
1840 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1842 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
1844 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
1845 ("*movdi_insn_sp32"): Do not set v3pipe.
1846 ("*movsi_insn"): Likewise.
1847 ("*movdi_insn_sp64"): Likewise.
1848 ("*movsf_insn"): Likewise.
1849 ("*movdf_insn_sp32"): Likewise.
1850 ("*movdf_insn_sp64"): Likewise.
1851 ("*zero_extendsidi2_insn_sp64"): Likewise.
1852 ("*sign_extendsidi2_insn"): Likewise.
1853 ("*mov<VM32:mode>_insn"): Likewise.
1854 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1855 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1856 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1857 ("<vlop:code><VL:mode>3"): Likewise.
1858 ("*not_<vlop:code><VL:mode>3"): Likewise.
1859 ("*nand<VL:mode>_vis"): Likewise.
1860 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
1861 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
1862 ("one_cmpl<VL:mode>2"): Likewise.
1863 ("faligndata<VM64:mode>_vis"): Likewise.
1864 ("alignaddrsi_vis"): Likewise.
1865 ("alignaddrdi_vis"): Likweise.
1866 ("alignaddrlsi_vis"): Likewise.
1867 ("alignaddrldi_vis"): Likewise.
1868 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1869 ("bmaskdi_vis"): Likewise.
1870 ("bmasksi_vis"): Likewise.
1871 ("bshuffle<VM64:mode>_vis"): Likewise.
1872 ("cmask8<P:mode>_vis"): Likewise.
1873 ("cmask16<P:mode>_vis"): Likewise.
1874 ("cmask32<P:mode>_vis"): Likewise.
1875 ("pdistn<P:mode>_vis"): Likewise.
1876 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1878 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1880 * config/sparc/sparc.md ("subtype"): New insn attribute.
1881 ("*wrgsr_sp64"): Set insn subtype.
1882 ("*rdgsr_sp64"): Likewise.
1883 ("alignaddrsi_vis"): Likewise.
1884 ("alignaddrdi_vis"): Likewise.
1885 ("alignaddrlsi_vis"): Likewise.
1886 ("alignaddrldi_vis"): Likewise.
1887 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1888 ("fexpand_vis"): Likewise.
1889 ("fpmerge_vis"): Likewise.
1890 ("faligndata<VM64:mode>_vis"): Likewise.
1891 ("bshuffle<VM64:mode>_vis"): Likewise.
1892 ("cmask8<P:mode>_vis"): Likewise.
1893 ("cmask16<P:mode>_vis"): Likewise.
1894 ("cmask32<P:mode>_vis"): Likewise.
1895 ("fchksm16_vis"): Likewise.
1896 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
1897 ("fmean16_vis"): Likewise.
1898 ("fp<plusminus_insn>64_vis"): Likewise.
1899 ("<plusminus_insn>v8qi3"): Likewise.
1900 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1901 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
1902 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
1903 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
1904 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
1905 ("*movqi_insn"): Likewise.
1906 ("*movhi_insn"): Likewise.
1907 ("*movsi_insn"): Likewise.
1908 ("movsi_pic_gotdata_op"): Likewise.
1909 ("*movdi_insn_sp32"): Likewise.
1910 ("*movdi_insn_sp64"): Likewise.
1911 ("movdi_pic_gotdata_op"): Likewise.
1912 ("*movsf_insn"): Likewise.
1913 ("*movdf_insn_sp32"): Likewise.
1914 ("*movdf_insn_sp64"): Likewise.
1915 ("*zero_extendhisi2_insn"): Likewise.
1916 ("*zero_extendqihi2_insn"): Likewise.
1917 ("*zero_extendqisi2_insn"): Likewise.
1918 ("*zero_extendqidi2_insn"): Likewise.
1919 ("*zero_extendhidi2_insn"): Likewise.
1920 ("*zero_extendsidi2_insn_sp64"): Likewise.
1921 ("ldfsr"): Likewise.
1922 ("prefetch_64"): Likewise.
1923 ("prefetch_32"): Likewise.
1924 ("tie_ld32"): Likewise.
1925 ("tie_ld64"): Likewise.
1926 ("*tldo_ldub_sp32"): Likewise.
1927 ("*tldo_ldub1_sp32"): Likewise.
1928 ("*tldo_ldub2_sp32"): Likewise.
1929 ("*tldo_ldub_sp64"): Likewise.
1930 ("*tldo_ldub1_sp64"): Likewise.
1931 ("*tldo_ldub2_sp64"): Likewise.
1932 ("*tldo_ldub3_sp64"): Likewise.
1933 ("*tldo_lduh_sp32"): Likewise.
1934 ("*tldo_lduh1_sp32"): Likewise.
1935 ("*tldo_lduh_sp64"): Likewise.
1936 ("*tldo_lduh1_sp64"): Likewise.
1937 ("*tldo_lduh2_sp64"): Likewise.
1938 ("*tldo_lduw_sp32"): Likewise.
1939 ("*tldo_lduw_sp64"): Likewise.
1940 ("*tldo_lduw1_sp64"): Likewise.
1941 ("*tldo_ldx_sp64"): Likewise.
1942 ("*mov<VM32:mode>_insn"): Likewise.
1943 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1944 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1946 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1948 * config/sparc/sparc.md ("type"): New insn type viscmp.
1949 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
1951 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
1952 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
1953 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1954 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
1956 ("n7_vis_logical_11cycle"): Likewise.
1957 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
1958 * config/sparc/niagara2.md ("niag3_vis": Likewise.
1959 * config/sparc/niagara.md ("niag_vis"): Likewise.
1960 * config/sparc/ultra3.md ("us3_fga"): Likewise.
1961 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
1963 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1965 * config/sparc/sparc.md: New instruction type `bmask'.
1966 (bmaskdi_vis): Use the `bmask' type.
1967 (bmasksi_vis): Likewise.
1968 * config/sparc/ultra3.md (us3_array): Likewise.
1969 * config/sparc/niagara7.md (n7_array): Likewise.
1970 * config/sparc/niagara4.md (n4_array): Likewise.
1971 * config/sparc/niagara2.md (niag2_vis): Likewise.
1972 (niag3_vis): Likewise.
1973 * config/sparc/niagara.md (niag_vis): Likewise.
1975 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1977 * ipa-comdats.c: Remove optimize check from gate.
1978 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
1979 for functions not optimized.
1980 (ipa_fn_summary_read): Skip optimize check.
1981 (ipa_fn_summary_write): Likewise.
1982 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
1984 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
1986 (can_inline_edge_p): Check flag_pcc_struct_return for match.
1987 (check_callers): Give up on caller which is not optimized.
1988 (inline_small_functions): Likewise.
1989 (ipa_inline): Do not give up when not optimizing.
1990 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
1991 away unoptimizes cdtors.
1992 (whole_program_function_and_variable_visibility): Do
1993 ipa_discover_readonly_nonaddressable_vars in LTO mode.
1994 * ipa.c (process_references): Do not check optimize.
1995 (symbol_table::remove_unreachable_nodes): Update optimize check.
1996 (set_writeonly_bit): Update optimize check.
1997 (pass_ipa_cdtor_merge::gate): Do not check optimize.
1998 (pass_ipa_single_use::gate): Remove.
2000 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2002 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2003 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2004 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2005 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2006 permute_load, permute_store, adjust_extract, adjust_splat,
2007 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2008 replace_swap_with_copy, dump_swap_insn_table,
2009 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2010 recombine_lvx_pattern, recombine_stvx_pattern,
2011 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2012 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2013 to file rs6000-p8swap.c.
2014 * config/rs6000/rs6000-p8swap.c: New file.
2015 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2016 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2017 and rs6000*-*-* targets.
2019 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2021 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2023 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2025 * lto-wrapper.c (merge_and_complain): Do not merge
2026 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2027 fsigned_zeros, ftrapping_math, fwrapv.
2028 (append_compiler_options): Do not track these options.
2029 (append_linker_options): Likewie
2031 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2033 * cgraphunit.c (cgraph_node::finalize_function): When
2034 !flag_toplevel_reorde set no_reorder flag.
2035 (varpool_node::finalize_decl): Likewise.
2036 (symbol_table::compile): Drop no toplevel reorder path.
2038 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2040 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2041 edges; zero probability is not better than uninitialized.
2043 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2045 * asan.h (asan_sanitize_allocas_p): Declare.
2046 * asan.c (asan_sanitize_allocas_p): New function.
2047 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2048 (handle_builtin_alloca): Likewise.
2049 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2050 if !asan_sanitize_allocas_p.
2051 * params.def (asan-instrument-allocas): Add new option.
2052 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2053 * opts.c (common_handle_option): Disable allocas sanitization for
2056 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2058 * asan.c: Include gimple-fold.h.
2059 (get_last_alloca_addr): New function.
2060 (handle_builtin_stackrestore): Likewise.
2061 (handle_builtin_alloca): Likewise.
2062 (asan_emit_allocas_unpoison): Likewise.
2063 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2064 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2065 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2066 (instrument_builtin_call): Pass gimple iterator to
2067 get_mem_refs_of_builtin_call.
2068 (last_alloca_addr): New global.
2069 * asan.h (asan_emit_allocas_unpoison): Declare.
2070 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2071 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2072 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2073 if function calls alloca.
2074 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2075 * gimple-fold.h (replace_call_with_value): Declare.
2076 * internal-fn.c: Include asan.h.
2077 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2078 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2080 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2082 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2083 (C_SELFTEST_FLAGS): New.
2084 (CPP_SELFTEST_FLAGS): New.
2085 (SELFTEST_DEPS): New, from deps of s-selftest.
2086 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2087 (CPP_SELFTEST_DEPS): New.
2088 (selftest): Add dependency on s-selftest-c++.
2089 (s-selftest): Rename to...
2090 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2091 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2092 than SELFTEST_FLAGS.
2093 (selftest-gdb): Rename to...
2094 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2096 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2097 (selftest-valgrind): Rename to...
2098 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2100 (selftest-valgrind): Reintroduce as an alias for
2101 selftest-c-valgrind.
2102 (s-selftest-c++): New.
2103 (selftest-c++-gdb): New.
2104 (selftest-c++-valgrind): New.
2106 2017-07-06 Olivier Hainque <hainque@adacore.com>
2108 * gcc.c (process_command): When deciding if undefined variables
2109 should be ignored when processing specs, accept "gcc -v" as well.
2111 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2113 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2114 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2116 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2118 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2119 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2120 * config/arm/arm-tables.opt: Regenerate.
2121 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2123 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2125 2017-07-06 Carl Love <cel@us.ibm.com>
2127 * ChangeLog: Clean up from mid air collision
2129 2017-07-06 Carl Love <cel@us.ibm.com>
2131 * config/rs6000/rs6000-c.c: Add support for built-in functions
2132 vector signed int vec_subc (vector signed int, vector signed int);
2133 vector signed __int128 vec_subc (vector signed __int128,
2134 vector signed __int128);
2135 vector unsigned __int128 vec_subc (vector unsigned __int128,
2136 vector unsigned __int128);
2137 vector signed int vec_sube (vector signed int, vector signed int,
2139 vector unsigned int vec_sube (vector unsigned int,
2140 vector unsigned int,
2141 vector unsigned int);
2142 vector signed __int128 vec_sube (vector signed __int128,
2143 vector signed __int128,
2144 vector signed__int128);
2145 vector unsigned __int128 vec_sube (vector unsigned __int128,
2146 vector unsigned __int128,
2147 vector unsigned __int128);
2148 vector signed int vec_subec (vector signed int, vector signed int,
2150 vector unsigned int vec_subec (vector unsigned int,
2151 vector unsigned int,
2152 vector unsigned int);
2153 vector signed __int128 vec_subec (vector signed __int128,
2154 vector signed __int128,
2155 vector signed__int128);
2156 vector unsigned __int128 vec_subec (vector unsigned __int128,
2157 vector unsigned __int128,
2158 vector unsigned __int128);
2159 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2160 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2161 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2162 BU_ALTIVEC_OVERLOAD_X definitions.
2163 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2164 * doc/extend.texi: Update the built-in documentation file for the new
2167 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2170 * diagnostic-show-locus.c (layout::layout): Use start and finish
2171 spelling location for the start and finish of each range.
2172 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2173 Add unused aspect param.
2174 * input.c (expand_location_1): Add "aspect" param, and use it
2175 to access the correct part of the location.
2176 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2178 (expand_location_to_spelling_point): Likewise.
2179 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2180 param, and pass it to expand_location_1.
2182 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2184 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2185 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2186 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2187 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2188 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2189 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2190 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2191 _mm_maskz_getmant_ss): New intrinsics.
2192 (__builtin_ia32_getexpss128_mask): Changed to ...
2193 __builtin_ia32_getexpss128_round ... this.
2194 (__builtin_ia32_getexpsd128_mask): Changed to ...
2195 __builtin_ia32_getexpsd128_round ... this.
2196 * config/i386/i386-builtin-types.def
2197 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2198 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2199 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2200 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2201 __builtin_ia32_getmantss_mask_round): New builtins.
2202 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2203 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2204 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2205 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2206 * config/i386/sse.md
2207 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2208 avx512f_sgetexp<mode><mask_scalar_name>
2209 <round_saeonly_scalar_name> ... this.
2210 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2211 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2212 vgetexp<ssescalarmodesuffix>
2213 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2214 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2215 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2216 avx512f_vgetmant<mode><mask_scalar_name>
2217 <round_saeonly_scalar_name> ... this.
2218 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2219 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2220 vgetmant<ssescalarmodesuffix>
2221 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2222 %0<mask_scalar_operand4>, %1, %2
2223 <round_saeonly_scalar_mask_op4>, %3} ... this.
2224 * config/i386/subst.md (mask_scalar_operand4,
2225 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2226 round_saeonly_scalar_nimm_predicate): New subst attributes.
2228 2017-07-06 Julia Koval <julia.koval@intel.com>
2230 * config/i386/i386.c (ix86_erase_embedded_rounding):
2231 Remove code for old rounding pattern.
2233 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2235 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2237 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2239 * doc/sourcebuild.texi (Test Directives, Variants of
2240 dg-require-support): Add documentation for dg-require-stack-check.
2242 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2244 * config/i386/subst.md (mask_scalar, round_scalar,
2245 round_saeonly_scalar): New meta-templates.
2246 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2247 round_scalar_mask_operand3, round_scalar_mask_op3,
2248 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2249 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2250 round_saeonly_scalar_constraint,
2251 round_saeonly_scalar_prefix): New subst attribute.
2252 * config/i386/sse.md
2253 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2254 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2255 <round_scalar_name> ... this.
2256 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2257 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2258 <round_scalar_name> ... this.
2259 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2260 <sse>_vm<code><mode>3<mask_scalar_name>
2261 <round_saeonly_scalar_name> ... this.
2262 (v<plusminus_mnemonic><ssescalarmodesuffix>
2263 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2264 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2265 v<plusminus_mnemonic><ssescalarmodesuffix>
2266 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2267 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2268 (v<multdiv_mnemonic><ssescalarmodesuffix>
2269 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2270 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2271 v<multdiv_mnemonic><ssescalarmodesuffix>
2272 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2273 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2274 (v<maxmin_float><ssescalarmodesuffix>
2275 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2276 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2277 v<maxmin_float><ssescalarmodesuffix>
2278 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2279 %0<mask_scalar_operand3>, %1, %<iptr>2
2280 <round_saeonly_scalar_mask_op3>} ... this.
2282 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2284 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2285 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2287 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2288 Alan Hayward <alan.hayward@arm.com>
2289 David Sherwood <david.sherwood@arm.com>
2291 * combine.c (simplify_if_then_else): Remove "enum" before
2293 * compare-elim.c (can_eliminate_compare): Likewise.
2294 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2296 (aarch64_lookup_simd_builtin_type): Likewise.
2297 (aarch64_simd_builtin_type): Likewise.
2298 (aarch64_init_simd_builtin_types): Likewise.
2299 (aarch64_simd_expand_args): Likewise.
2300 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2302 (aarch64_reverse_mask): Likewise.
2303 (aarch64_simd_emit_reg_reg_move): Likewise.
2304 (aarch64_gen_adjusted_ldpstp): Likewise.
2305 (aarch64_ccmp_mode_to_code): Likewise.
2306 (aarch64_operands_ok_for_ldpstp): Likewise.
2307 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2308 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2310 (aarch64_min_divisions_for_recip_mul): Likewise.
2311 (aarch64_reassociation_width): Likewise.
2312 (aarch64_get_condition_code_1): Likewise.
2313 (aarch64_simd_emit_reg_reg_move): Likewise.
2314 (aarch64_simd_attr_length_rglist): Likewise.
2315 (aarch64_reverse_mask): Likewise.
2316 (aarch64_operands_ok_for_ldpstp): Likewise.
2317 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2318 (aarch64_gen_adjusted_ldpstp): Likewise.
2319 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2321 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2322 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2323 (arm_lookup_simd_builtin_type): Likewise.
2324 (arm_simd_builtin_type): Likewise.
2325 (arm_init_simd_builtin_types): Likewise.
2326 (arm_expand_builtin_args): Likewise.
2327 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2328 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2329 (ft32_setup_incoming_varargs): Likewise.
2330 (ft32_function_arg): Likewise.
2331 (ft32_function_arg_advance): Likewise.
2332 (ft32_pass_by_reference): Likewise.
2333 (ft32_arg_partial_bytes): Likewise.
2334 (ft32_valid_pointer_mode): Likewise.
2335 (ft32_addr_space_pointer_mode): Likewise.
2336 (ft32_addr_space_legitimate_address_p): Likewise.
2337 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2339 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2340 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2341 (iamcu_alignment): Likewise.
2342 (canonicalize_vector_int_perm): Likewise.
2343 (ix86_noce_conversion_profitable_p): Likewise.
2344 (ix86_mpx_bound_mode): Likewise.
2345 (ix86_operands_ok_for_move_multiple): Likewise.
2346 * config/microblaze/microblaze-protos.h
2347 (microblaze_expand_conditional_branch_reg): Likewise.
2348 * config/microblaze/microblaze.c
2349 (microblaze_expand_conditional_branch_reg): Likewise.
2350 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2352 (rs6000_reassociation_width): Likewise.
2353 (rs6000_invalid_binary_op): Likewise.
2354 (fusion_p9_p): Likewise.
2355 (emit_fusion_p9_load): Likewise.
2356 (emit_fusion_p9_store): Likewise.
2357 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2359 (riscv_hard_regno_mode_ok_p): Likewise.
2360 (riscv_address_insns): Likewise.
2361 (riscv_split_symbol): Likewise.
2362 (riscv_legitimize_move): Likewise.
2363 (riscv_function_value): Likewise.
2364 (riscv_hard_regno_nregs): Likewise.
2365 (riscv_expand_builtin): Likewise.
2366 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2367 (riscv_build_integer): Likewise.
2368 (riscv_split_integer): Likewise.
2369 (riscv_legitimate_constant_p): Likewise.
2370 (riscv_cannot_force_const_mem): Likewise.
2371 (riscv_regno_mode_ok_for_base_p): Likewise.
2372 (riscv_valid_base_register_p): Likewise.
2373 (riscv_valid_offset_p): Likewise.
2374 (riscv_valid_lo_sum_p): Likewise.
2375 (riscv_classify_address): Likewise.
2376 (riscv_legitimate_address_p): Likewise.
2377 (riscv_address_insns): Likewise.
2378 (riscv_load_store_insns): Likewise.
2379 (riscv_force_binary): Likewise.
2380 (riscv_split_symbol): Likewise.
2381 (riscv_force_address): Likewise.
2382 (riscv_legitimize_address): Likewise.
2383 (riscv_move_integer): Likewise.
2384 (riscv_legitimize_const_move): Likewise.
2385 (riscv_legitimize_move): Likewise.
2386 (riscv_address_cost): Likewise.
2387 (riscv_subword): Likewise.
2388 (riscv_output_move): Likewise.
2389 (riscv_canonicalize_int_order_test): Likewise.
2390 (riscv_emit_int_order_test): Likewise.
2391 (riscv_function_arg_boundary): Likewise.
2392 (riscv_pass_mode_in_fpr_p): Likewise.
2393 (riscv_pass_fpr_single): Likewise.
2394 (riscv_pass_fpr_pair): Likewise.
2395 (riscv_get_arg_info): Likewise.
2396 (riscv_function_arg): Likewise.
2397 (riscv_function_arg_advance): Likewise.
2398 (riscv_arg_partial_bytes): Likewise.
2399 (riscv_function_value): Likewise.
2400 (riscv_pass_by_reference): Likewise.
2401 (riscv_setup_incoming_varargs): Likewise.
2402 (riscv_print_operand): Likewise.
2403 (riscv_elf_select_rtx_section): Likewise.
2404 (riscv_save_restore_reg): Likewise.
2405 (riscv_for_each_saved_reg): Likewise.
2406 (riscv_register_move_cost): Likewise.
2407 (riscv_hard_regno_mode_ok_p): Likewise.
2408 (riscv_hard_regno_nregs): Likewise.
2409 (riscv_class_max_nregs): Likewise.
2410 (riscv_memory_move_cost): Likewise.
2411 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2412 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2413 (rl78_addr_space_address_mode): Likewise.
2414 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2416 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2417 (rs6000_reassociation_width): Likewise.
2418 (rs6000_invalid_binary_op): Likewise.
2419 (fusion_p9_p): Likewise.
2420 (emit_fusion_p9_load): Likewise.
2421 (emit_fusion_p9_store): Likewise.
2422 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2423 (ok_for_simple_move_operands): Likewise.
2424 (ok_for_simple_move_strict_operands): Likewise.
2425 (ok_for_simple_arith_logic_operands): Likewise.
2426 (visium_legitimize_reload_address): Likewise.
2427 (visium_select_cc_mode): Likewise.
2428 (output_cbranch): Likewise.
2429 (visium_split_double_move): Likewise.
2430 (visium_expand_copysign): Likewise.
2431 (visium_expand_int_cstore): Likewise.
2432 (visium_expand_fp_cstore): Likewise.
2433 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2434 (visium_function_arg): Likewise.
2435 (visium_function_arg_advance): Likewise.
2436 (visium_libcall_value): Likewise.
2437 (visium_setup_incoming_varargs): Likewise.
2438 (visium_legitimate_constant_p): Likewise.
2439 (visium_legitimate_address_p): Likewise.
2440 (visium_legitimize_address): Likewise.
2441 (visium_secondary_reload): Likewise.
2442 (visium_register_move_cost): Likewise.
2443 (visium_memory_move_cost): Likewise.
2444 (prepare_move_operands): Likewise.
2445 (ok_for_simple_move_operands): Likewise.
2446 (ok_for_simple_move_strict_operands): Likewise.
2447 (ok_for_simple_arith_logic_operands): Likewise.
2448 (visium_function_value_1): Likewise.
2449 (rtx_ok_for_offset_p): Likewise.
2450 (visium_legitimize_reload_address): Likewise.
2451 (visium_split_double_move): Likewise.
2452 (visium_expand_copysign): Likewise.
2453 (visium_expand_int_cstore): Likewise.
2454 (visium_expand_fp_cstore): Likewise.
2455 (visium_split_cstore): Likewise.
2456 (visium_select_cc_mode): Likewise.
2457 (visium_split_cbranch): Likewise.
2458 (output_cbranch): Likewise.
2459 (visium_print_operand_address): Likewise.
2460 * expmed.c (flip_storage_order): Likewise.
2461 * expmed.h (emit_cstore): Likewise.
2462 (flip_storage_order): Likewise.
2463 * genrecog.c (validate_pattern): Likewise.
2464 * hsa-gen.c (gen_hsa_addr): Likewise.
2465 * internal-fn.c (expand_arith_overflow): Likewise.
2466 * ira-color.c (allocno_copy_cost_saving): Likewise.
2467 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2468 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2469 (process_invariant_for_inheritance): Likewise.
2470 * lra-eliminations.c (move_plus_up): Likewise.
2471 * omp-low.c (lower_oacc_reductions): Likewise.
2472 * simplify-rtx.c (simplify_subreg): Likewise.
2473 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2474 (TARGET_CHKP_BOUND_MODE): Likewise..
2475 * targhooks.c (default_chkp_bound_mode): Likewise.
2476 (default_setup_incoming_vararg_bounds): Likewise.
2477 * targhooks.h (default_chkp_bound_mode): Likewise.
2478 (default_setup_incoming_vararg_bounds): Likewise.
2479 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2480 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2481 (have_whole_vector_shift): Likewise.
2482 * tree-vect-stmts.c (vectorizable_load): Likewise.
2483 * doc/tm.texi: Regenerate.
2485 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2487 Graceful degrade if Binutils PR21472 is not available.
2490 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2491 .rodata in flash test fails.
2492 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2493 * confgure: Regenerate.
2494 * config.in: Regenerate.
2495 * config/avr/avr.c (avr_asm_named_section)
2496 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2497 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2498 (avr_asm_init_sections): Same.
2500 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2502 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2503 (fma<VH:mode>4_intrinsic): Likewise.
2504 (*fmsub<VCVTF:mode>4): Likewise.
2505 (*fmsub<VH:mode>4_intrinsic): Likewise.
2507 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2510 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2511 Don't depend on "optimize > 0".
2512 (out_movhi_r_mr, out_movqi_mr_r): Same.
2513 (out_movhi_mr_r, out_movqi_r_mr): Same.
2514 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2515 io_address_operand on "optimize > 0".
2517 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2519 * tree-loop-distribution.c: Add general explanantion on the pass.
2520 (generate_loops_for_partition): Mark distributed loop.
2521 (pg_add_dependence_edges): New parameter. Handle alias data
2522 dependence specially and record it in the parameter if asked.
2523 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2524 (init_partition_graph_vertices, add_partition_graph_edge): New.
2525 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2526 (free_partition_graph_vdata, build_partition_graph): New.
2527 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2528 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2529 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2530 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2531 (version_for_distribution_p, finalize_partitions): New.
2532 (distribute_loop): Handle alias data dependence specially. Factor
2533 out loop fusion code as functions and call these functions.
2535 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2537 * tree-loop-distribution.c (classify_partition): New parameter and
2538 better handle reduction statement.
2539 (rdg_build_partitions): Revise comment.
2540 (distribute_loop): Compute statements in all partitions and pass it
2541 to classify_partition.
2543 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2545 * tree-loop-distribution.c (enum partition_type): New.
2546 (struct partition): New field type.
2547 (partition_merge_into): Add parameter. Update partition type.
2548 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2549 (build_rdg_partition_for_vertex): Compute partition type.
2550 (rdg_build_partitions): Dump partition type.
2551 (distribute_loop): Update calls to partition_merge_into.
2553 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2555 * tree-loop-distribution.c (struct ddr_hasher): New.
2556 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2558 (classify_partition): Call get_data_dependence.
2559 (pg_add_dependence_edges): Ditto.
2560 (distribute_loop): Release data dependence hash table.
2562 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2564 * tree-loop-distribution.c (ref_base_address): Delete.
2565 (similar_memory_accesses): Rename ...
2566 (share_memory_accesses): ... to this. Check if partitions access
2567 the same memory reference.
2568 (distribute_loop): Call share_memory_accesses.
2570 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2572 * tree-loop-distribution.c (struct partition): New field recording
2574 (partition_alloc, partition_free): Init and release data refs.
2575 (partition_merge_into): Merge data refs.
2576 (build_rdg_partition_for_vertex): Collect data refs for partition.
2577 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2579 (distribute_loop): Remve data refs from vertice data of partition
2582 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2584 * tree-loop-distribution.c (params.h): Include header file.
2585 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2586 (datarefs_vec): New global var.
2587 (create_rdg_vertices): Use datarefs_vec directly.
2588 (free_rdg): Don't free data references.
2589 (build_rdg): Update use. Don't free data references.
2590 (distribute_loop): Compute global variable for data references.
2591 Bail out if there are too many data references.
2593 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2595 * tree-loop-distribution.c (loop_nest): New global var.
2596 (build_rdg): Use loop directly, rather than loop nest.
2597 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2599 (distribute_loop): Compute global variable loop nest. Update use.
2601 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2603 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2604 (partition_merge_into): New parameter. Dump reason for fusion.
2605 (distribute_loop): Update use of partition_merge_into.
2607 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2609 * tree-loop-distribution.c (bb_top_order_index): New.
2610 (bb_top_order_index_size, bb_top_order_cmp): New.
2611 (stmts_from_loop): Use topological order.
2612 (pass_loop_distribution::execute): Compute and release topological
2613 order for basic blocks.
2615 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2617 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2620 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2622 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2623 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2624 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2625 * internal-fn.def (LOOP_DIST_ALIAS): New.
2626 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2627 (fold_loop_internal_call): ... this.
2628 (vect_loop_dist_alias_call): New function.
2629 (set_uid_loop_bbs): Call fold_loop_internal_call.
2630 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2633 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2636 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2637 Require dead FLAGS_REG at the beginning of a peephole.
2639 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2642 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2643 arguments in the call to __builtin_ia32_sbb_u32.
2644 (_subborrow_u64): Swap _X and _Y arguments in the call to
2645 __builtin_ia32_sbb_u64.
2647 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2650 * tree-vrp.c (compare_assert_loc): Turn into a function template
2651 with stable template parameter. Only test if a->e is NULL,
2652 !a->e == !b->e has been verified already. Use e == NULL or
2653 e != NULL instead of e or ! e tests. If stable is true, don't use
2654 iterative_hash_expr, on the other side allow a or b or both NULL
2655 and sort the NULLs last.
2656 (process_assert_insertions): Sort using compare_assert_loc<false>
2657 instead of compare_assert_loc, later sort using
2658 compare_assert_loc<true> before calling process_assert_insertions_for
2659 in a loop. Use break instead of continue once seen NULL pointer.
2661 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2663 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2664 Cortex-R7 and Cortex-R8 processors.
2666 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2668 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2669 uninitialized while src is not.
2671 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2673 * common/config/arm/arm-common.c: Adjust include path for
2675 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2676 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2677 (arm-cpu-data.h): Likewise.
2678 (arm-cpu-cdata.h): Likewise.
2679 * config/arm/arm-cpu.h: Delete.
2680 * config/arm/arm-cpu-cdata.h: Delete.
2681 * config/arm/arm-cpu-data.h: Delete.
2683 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2685 * config/arm/arm-cpus.in (cortex-a55): New.
2686 (cortex-a75): Likewise.
2687 (cortex-a75.cortex-a55): Likewise.
2688 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2690 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2691 * config/arm/arm-cpu-cdata.h: Regenerate.
2692 * config/arm/arm-cpu-data.h: Regenerate.
2693 * config/arm/arm-cpu.h: Regenerate.
2694 * config/arm/arm-tables.opt: Regenerate.
2695 * config/arm/arm-tune.md: Regenerate.
2697 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2699 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2701 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2703 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2706 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2708 PR tree-optimization/81292
2709 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2710 full_string_p, also call adjust_related_strinfos if the adjustment
2711 is simple, otherwise invalidate related strinfos.
2713 2017-07-04 Martin Liska <mliska@suse.cz>
2716 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2717 newly created variable as DECL_IGNORED_P.
2719 2017-07-04 Martin Liska <mliska@suse.cz>
2722 * ipa-inline.c (inline_small_functions):
2723 Use xstrdup_for_dump.
2725 2017-07-04 Tom de Vries <tom@codesourcery.com>
2727 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2729 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2732 * config/darwin.c (darwin_function_switched_text_sections):
2735 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2737 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2739 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2741 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2743 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2745 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2746 min_profitable_iters, and th as inclusive lower bounds.
2747 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
2748 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
2749 for min_profitable_iters and min_profitable_estimate.
2750 (vect_transform_loop): Treat th as an inclusive lower bound.
2751 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2753 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2756 * config/darwin.c (darwin_function_switched_text_sections):
2757 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
2758 in two pieces, and suppress the use of buf.
2760 2017-07-03 Nathan Sidwell <nathan@acm.org>
2762 * hash-table.h (hash_table_mod1): Fix indentation.
2764 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2767 * predict.c (force_edge_cold): Be more careful about propagation
2769 * profile-count.h (profile_probability::guessed,
2770 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
2772 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
2774 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
2776 * doc/invoke.texi (rcpc architecture extension): Document it.
2778 2017-07-03 Richard Biener <rguenther@suse.de>
2780 PR tree-optimization/60510
2781 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
2782 the scalar reduction PHI and use it.
2783 (vectorizable_reduction): Properly guard the single_defuse_cycle
2784 path for non-SLP reduction chains where we cannot use it.
2785 Rework reduc_def/index and vector type deduction. Rework
2786 vector operand gathering during reduction op code-gen.
2787 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
2788 chains dissolve the chain and leave it to non-SLP reduction
2791 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2793 * tree-data-ref.h (dr_alignment): Declare.
2794 * tree-data-ref.c (dr_alignment): New function.
2795 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
2796 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
2798 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
2800 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2802 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
2803 and base_misalignment fields.
2804 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
2805 * tree-data-ref.c: Include builtins.h.
2806 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
2807 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
2808 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
2809 * tree-vect-data-refs.c: Include tree-cfg.h.
2810 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
2811 fields instead of calculating an alignment here.
2812 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
2813 innermost_loop_behavior fields.
2815 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2817 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
2819 (DR_STEP_ALIGNMENT): New macro.
2820 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
2821 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
2822 (create_data_ref): Print it.
2823 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
2824 to tell whether the step preserves vector (mis)alignment.
2825 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2826 Move the check for an integer step and generalise to all INTEGER_CST.
2827 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
2828 Print the outer step alignment.
2830 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2832 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
2833 with offset_alignment.
2834 (DR_ALIGNED_TO): Delete.
2835 (DR_OFFSET_ALIGNMENT): New macro.
2836 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
2837 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
2838 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
2839 (create_data_ref): Likewise.
2840 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2841 (vect_analyze_data_refs): Likewise.
2842 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
2843 creating dummy innermost behavior.
2845 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2847 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
2848 with a "innermost_loop_behavior *" and refeence tree.
2849 * tree-data-ref.c (dr_analyze_innermost): Likewise.
2850 (create_data_ref): Update call accordingly.
2851 * tree-predcom.c (find_looparound_phi): Likewise.
2853 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2855 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
2856 fields with dr_wrt_vec_loop.
2857 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
2858 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
2859 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
2860 (vect_dr_behavior): New function.
2861 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2862 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
2863 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
2864 track whether the step preserves the misalignment.
2865 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2866 Use vect_dr_behavior.
2867 (vect_setup_realignment): Update call accordingly.
2868 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
2869 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
2870 call to vect_create_addr_base_for_vector_ref.
2871 (vect_create_cond_for_align_checks): Likewise.
2872 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
2873 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
2874 (vect_recog_mask_conversion_pattern): Likewise.
2875 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
2876 (new_stmt_vec_info): Remove redundant zeroing.
2878 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2880 * common/config/arm/arm-common.c (arm_be8_option): New function.
2881 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
2882 (ISA_ARMv6): Add isa_bit_be8.
2883 * config/arm/arm.h (arm_be8_option): Add prototype.
2884 (BE8_SPEC_FUNCTION): New define.
2885 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
2886 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
2887 (mlittle-endian): Similarly.
2888 (mbe8, mbe32): New options.
2889 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
2890 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
2892 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2894 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
2896 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2898 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
2899 (cleanup_tree_cfg_bb): Use it.
2900 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
2902 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
2904 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2907 * loop-doloop.c (add_test): Update profile.
2909 2017-07-03 Martin Liska <mliska@suse.cz>
2912 * sanopt.c (rewrite_usage_of_param): New function.
2913 (sanitize_rewrite_addressable_params): Likewise.
2914 (pass_sanopt::execute): Call rewrite_usage_of_param.
2916 2017-07-03 Richard Biener <rguenther@suse.de>
2918 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
2919 back to using VIEW_CONVERT_EXPR.
2921 2017-07-03 Martin Liska <mliska@suse.cz>
2924 * doc/extend.texi: Document when a resolver function is
2925 generated for target_clones.
2927 2017-07-03 Martin Liska <mliska@suse.cz>
2929 * asan.c (asan_emit_stack_protection): Unpoison just red zones
2930 and shadow memory of auto variables which are subject of
2931 use-after-scope sanitization.
2932 (asan_expand_mark_ifn): Add do set only when is_poison.
2934 2016-07-03 Richard Biener <rguenther@suse.de>
2936 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
2938 (vect_force_simple_reduction): Record reduction def -> phi mapping.
2939 (vectorizable_reduction): Perform reduction PHI creation when
2940 visiting a reduction PHI and adjust and simplify code generation
2941 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
2942 (vect_transform_loop): Visit reduction PHIs.
2943 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
2944 defs into the SLP tree.
2945 (vect_build_slp_tree): Reduction defs terminate the recursion.
2946 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
2948 (vect_get_vec_defs_for_stmt_copy): Export.
2949 (vect_get_vec_defs): Likewise.
2950 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
2952 (vect_get_vec_defs_for_stmt_copy): Declare.
2953 (vect_get_vec_defs): Likewise.
2955 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2957 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
2958 parameter with a "loop" parameter and use it instead of the
2959 loop containing DR_STMT. Don't check simple_iv when doing
2960 BB analysis. Describe the two analysis modes in the comment.
2962 2017-07-03 Tom de Vries <tom@codesourcery.com>
2964 PR tree-optimization/69468
2965 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
2966 (find_same_succ_bb): Handle ignore_edge_flags.
2968 2017-07-03 Tom de Vries <tom@codesourcery.com>
2970 PR tree-optimization/81192
2971 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
2973 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
2975 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
2977 2017-07-03 Tom de Vries <tom@codesourcery.com>
2979 PR tree-optimization/81192
2980 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
2981 BB_SAME_SUCC (bb) == NULL.
2983 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2985 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
2988 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2990 * dumpfile.c: Include profile-count.h
2991 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
2993 (insert_cond_bb): Update profile.
2994 * tree-cfg.h (insert_cond_bb): Update prototype.
2995 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
2996 * tree-dump.c: Do not include tree-cfg.
2998 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3000 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3002 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3004 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3007 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3009 * tree-complex.c (expand_complex_div_wide): update profile.
3011 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3012 Alan Hayward <alan.hayward@arm.com>
3013 David Sherwood <david.sherwood@arm.com>
3015 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3016 (CORETYPES_H): New define.
3017 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3018 (insn-modes-inline.h, s-modes-inline-h): New rules.
3019 (generated_files): Add insn-modes-inline.h.
3020 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3021 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3022 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3023 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3024 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3025 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3026 (build/gencodes.o, build/genconditions.o): Likewise.
3027 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3028 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3029 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3030 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3031 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3032 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3033 * coretypes.h: Include everything up to real.h for generators.
3034 Include insn-modes.h first. Include wide-int-print.h after
3035 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3036 * machmode.h: Don't include insn-modes.h here.
3037 * function-tests.c: Remove includes of signop.h, machmode.h,
3038 double-int.h and wide-int.h.
3040 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3042 * optc-save-gen.awk: Likewise.
3043 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3044 * godump.c: Remove include of wide-int-print.h.
3045 * pretty-print.h: Likewise.
3046 * wide-int-print.cc: Likewise.
3047 * wide-int.cc: Likewise.
3048 * hash-map-tests.c: Remove include of signop.h.
3049 * hash-set-tests.c: Likewise.
3050 * rtl-tests.c: Likewise.
3051 * mkconfig.sh: Remove include of machmode.h.
3052 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3054 (emit_insn_modes_inline_h): ...this new function. Emit the code
3055 into an insn-modes-inline.h header file, adding appropriate
3056 include guards and end comments.
3057 (emit_insn_modes_c_header): Remove include of machmode.h.
3058 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3060 (main): Handle -i flag and call emit_insn_modes_inline_h when
3063 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3065 * tree-ssa-strlen.c (strinfo): Rename the length field to
3066 nonzero_chars. Add a full_string_p field.
3067 (compare_nonzero_chars, zero_length_string_p): New functions.
3068 (get_addr_stridx): Add an offset_out parameter.
3069 Use compare_nonzero_chars.
3070 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3071 (new_strinfo): Update after above changes to strinfo.
3072 (set_endptr_and_length): Set full_string_p.
3073 (get_string_length): Update after above changes to strinfo.
3074 (unshare_strinfo): Update call to new_strinfo.
3075 (maybe_invalidate): Likewise.
3076 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3077 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3078 as a uhwi instead of an shwi. Update after above changes to
3079 strinfo and new_strinfo.
3080 (zero_length_string): Assert that chainsi contains full strings.
3081 Use zero_length_string_p. Update call to new_strinfo.
3082 (adjust_related_strinfos): Update after above changes to strinfo.
3083 Copy full_string_p from origsi.
3084 (adjust_last_stmt): Use zero_length_string_p.
3085 (handle_builtin_strlen): Update after above changes to strinfo and
3086 new_strinfo. Install the lhs as the string length if the previous
3087 entry didn't describe a full string.
3088 (handle_builtin_strchr): Update after above changes to strinfo
3090 (handle_builtin_strcpy): Likewise.
3091 (handle_builtin_strcat): Likewise.
3092 (handle_builtin_malloc): Likewise.
3093 (handle_pointer_plus): Likewise.
3094 (handle_builtin_memcpy): Likewise. Track nonzero characters
3095 that aren't necessarily followed by a nul terminator.
3096 (handle_char_store): Likewise.
3098 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3100 PR tree-optimization/80769
3101 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3102 for malloc and calloc. Document the new invariant that all related
3103 strinfos have delayed lengths or none do.
3104 (verify_related_strinfos): Move earlier in file.
3105 (set_endptr_and_length): New function, split out from...
3106 (get_string_length): ...here. Also set the lengths of related
3108 (zero_length_string): Assert that chainsi has known (rather than
3110 (adjust_related_strinfos): Likewise.
3112 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3114 PR tree-optimization/81136
3115 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3116 assert that two references with the same misalignment have the same
3117 compile-time misalignment if those compile-time misalignments
3120 2017-07-01 Andi Kleen <ak@linux.intel.com>
3122 * print-tree.c (print_node): Print all attributes.
3124 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3126 * cfg.c (scale_bbs_frequencies): New function.
3127 * cfg.h (scale_bbs_frequencies): Declare it.
3128 * cfgloopanal.c (single_likely_exit): Cleanup.
3129 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3131 (scale_loop_profile): Likewise.
3132 (loop_version): Likewise.
3133 (create_empty_loop_on_edge): Update.
3134 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3135 scale_loop_frequencies, scale_loop_profile, loopify,
3136 loop_version): Update prototypes.
3137 * modulo-sched.c (sms_schedule): Update.
3138 * predict.c (unlikely_executed_edge_p): Also check probability.
3139 (probably_never_executed_edge_p): Fix typo.
3140 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3141 * tree-parloops.c (gen_parallel_loop): Update.
3142 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3143 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3144 * tree-ssa-loop-split.c (split_loop): Update.
3145 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3146 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3147 (vect_loop_versioning): Update.
3148 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3150 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3152 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3154 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3156 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3157 to keep profile consistent.
3159 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3161 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3162 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3163 * profile-count.h (max_safe_multiplier): Make unsigned.
3164 (profile_count::guessed_zero): New.
3166 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3168 * bb-reorder.c (fix_up_crossing_landing_pad,
3169 fix_crossing_conditional_branches): Use make_single_succ_edge
3170 to keep profile consistent.
3172 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3174 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3177 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3180 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3181 the right scopes, make sure cond_jump isn't preserved between multiple
3182 iterations. Search for fallthru edge whenever there are 3+ edges and
3183 use find_fallthru_edge for it.
3185 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3187 Patch by Alexander Monakov <amonakov@ispras.ru>
3188 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3189 probabilities consistently.
3191 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3193 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3194 emit_cmp_and_jump_insns.
3196 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3199 * tree-inline.c (expand_call_inline): Combine profile statuses.
3201 2017-06-30 Andrew Pinski <apinski@cavium.com>
3203 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3204 fold_stmt returned true.
3206 2017-06-30 Nathan Sidwell <nathan@acm.org>
3208 * ggc.h (empty_string): Delete.
3209 * cfgexpand.c (expand_asm_stmt): Use plain "".
3210 * optabs.c (expand_asm_memory_barrier): Likewise.
3211 * stringpool.c (empty_string): Delete.
3212 (digit_vector, digit_string): Delete.
3213 (ggc_alloc_string): Use plain "", don't optimize single digit
3214 strings. Use ggc_alloc_atomic.
3216 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3218 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3219 comparison set and one other set, use the cost of the non-comparison
3222 2017-06-30 Nathan Sidwell <nathan@acm.org>
3224 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3227 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3229 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3230 loops. Remove now unneeded calls to gimple_switch_set_label() that
3231 just set removed labels to NULL_TREE.
3233 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3235 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3236 (set_range_info): ...here. Only call set_range_info_raw if domain
3238 (set_nonzero_bits): Call set_range_info_raw.
3239 * tree-ssanames.h (set_range_info_raw): New.
3241 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3244 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3245 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3246 of nonimmediate_operand and <store_mask_constraint> instead of m
3247 for the input operand. For V8FI iterator, always split if input
3248 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3249 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3250 <store_mask_predicate> instead of register_operand and
3251 <store_mask_constraint> instead of v for the input operand. Make
3252 sure both operands aren't MEMs for if not <mask_applied>.
3254 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3256 * lto-wrapper.c (copy_file) Close both file descriptors before
3259 2017-06-30 Martin Liska <mliska@suse.cz>
3262 * multiple_target.c (create_dispatcher_calls): Make ifunc
3263 also for function that don't have calls or are not referenced.
3265 2017-06-30 Richard Biener <rguenther@suse.de>
3267 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3268 analyze the first scalar stmt. Move vector type computation
3269 for the BB case here from ...
3270 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3271 live operation processing in the SLP case properly.
3273 2017-06-30 Richard Biener <rguenther@suse.de>
3275 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3277 2017-06-30 Martin Liska <mliska@suse.cz>
3280 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3281 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3283 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3285 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3288 2017-06-30 Richard Biener <rguenther@suse.de>
3290 PR tree-optimization/81249
3291 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3292 condition reduction result to original scalar type.
3294 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3296 * profile-count.h (enum profile_quality): Fix typos and whitespace
3299 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3301 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3302 type for branch probabilities.
3304 2017-06-29 Julian Brown <julian@codesourcery.com>
3305 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3307 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3308 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3309 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3310 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3312 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3314 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3315 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3316 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3317 CC usage from generic code to here.
3318 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3319 CC usage into the target macros.
3321 2017-06-29 Maya Rashish <coypu@sdf.org>
3323 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3326 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3328 * arm/arm-builtins.c: Include profile-count.h
3329 * except.c (sjlj_emit_function_enter): Use
3330 profile_probability::unlikely.
3332 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3334 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3335 and tocrel_offset be pointer args rather than implicitly using
3337 (legitimate_constant_pool_address_p, rs6000_emit_move,
3338 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3339 tocrel_offset and use in toc_relative_expr_p call.
3340 (print_operand, print_operand_address): Use static tocrel_base_oac
3341 and tocrel_offset_oac.
3342 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3345 2017-06-29 Maya Rashish <coypu@sdf.org>
3347 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3349 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3351 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3352 objects, take into account only the alignment of 'op0' and 'mode1' if
3355 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3357 * ccmp.c (ccmp_tree_comparison_p): New function.
3358 (ccmp_candidate_p): Update to use above function.
3359 (get_compare_parts): New function.
3360 (expand_ccmp_next): Update to use new functions.
3361 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3363 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3364 take mode as argument.
3365 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3366 * expr.c (expand_expr_real_1): Pass mode as argument.
3368 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3370 * combine.c (combine_instructions): Print insns to dump_file, together
3373 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3375 * asan.c (asan_emit_stack_protection): Update.
3376 (create_cond_insert_point): Update.
3377 * auto-profile.c (afdo_propagate_circuit): Update.
3378 * basic-block.h (struct edge_def): Turn probability to
3379 profile_probability.
3380 (EDGE_FREQUENCY): Update.
3381 * bb-reorder.c (find_traces_1_round): Update.
3382 (better_edge_p): Update.
3383 (sanitize_hot_paths): Update.
3384 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3385 (make_single_succ_edge): Update.
3386 (check_bb_profile): Update.
3387 (dump_edge_info): Update.
3388 (update_bb_profile_for_threading): Update.
3389 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3390 probabilitycount to 0.
3391 * cfgbuild.c (compute_outgoing_frequencies): Update.
3392 * cfgcleanup.c (try_forward_edges): Update.
3393 (outgoing_edges_match): Update.
3394 (try_crossjump_to_edge): Update.
3395 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3396 (expand_gimple_tailcall): Update.
3397 (construct_init_block): Use make_single_succ_edge.
3398 (construct_exit_block): Use make_single_succ_edge.
3399 * cfghooks.c (verify_flow_info): Update.
3400 (redirect_edge_succ_nodup): Update.
3401 (split_edge): Update.
3402 (account_profile_record): Update.
3403 * cfgloopanal.c (single_likely_exit): Update.
3404 * cfgloopmanip.c (scale_loop_profile): Update.
3405 (set_zero_probability): Remove.
3406 (duplicate_loop_to_header_edge): Update.
3407 * cfgloopmanip.h (loop_version): Update prototype.
3408 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3409 (force_nonfallthru_and_redirect): Update.
3410 (update_br_prob_note): Update.
3411 (rtl_verify_edges): Update.
3412 (purge_dead_edges): Update.
3413 (rtl_lv_add_condition_to_bb): Update.
3414 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3415 * cgraphunit.c (init_lowered_empty_function): Update.
3416 (cgraph_node::expand_thunk): Update.
3417 * cilk-common.c: Include profile-count.h
3418 * dojump.c (inv): Remove.
3419 (jumpifnot): Update.
3420 (jumpifnot_1): Update.
3421 (do_jump_1): Update.
3423 (do_jump_by_parts_greater_rtx): Update.
3424 (do_compare_rtx_and_jump): Update.
3425 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3426 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3427 * dwarf2cfi.c: Include profile-count.h
3428 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3429 (sjlj_emit_dispatch_table): Likewise.
3430 * explow.c: Include profile-count.h
3431 * expmed.c (emit_store_flag_force): Update.
3432 (do_cmp_and_jump): Update.
3433 * expr.c (compare_by_pieces_d::generate): Update.
3434 (compare_by_pieces_d::finish_mode): Update.
3435 (emit_block_move_via_loop): Update.
3436 (store_expr_with_bounds): Update.
3437 (store_constructor): Update.
3438 (expand_expr_real_2): Update.
3439 (expand_expr_real_1): Update.
3440 * expr.h (try_casesi, try_tablejump): Update prototypes.
3441 * gimple-pretty-print.c (dump_probability): Update.
3442 (dump_profile): New.
3443 (dump_gimple_label): Update.
3444 (dump_gimple_bb_header): Update.
3445 * graph.c (draw_cfg_node_succ_edges): Update.
3446 * hsa-gen.c (convert_switch_statements): Update.
3447 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3448 (find_if_case_1): Update.
3449 (find_if_case_2): Update.
3450 * internal-fn.c (expand_arith_overflow_result_store): Update.
3451 (expand_addsub_overflow): Update.
3452 (expand_neg_overflow): Update.
3453 (expand_mul_overflow): Update.
3454 (expand_vector_ubsan_overflow): Update.
3455 * ipa-cp.c (good_cloning_opportunity_p): Update.
3456 * ipa-split.c (split_function): Use make_single_succ_edge.
3457 * ipa-utils.c (ipa_merge_profiles): Update.
3458 * loop-doloop.c (add_test): Update.
3459 (doloop_modify): Update.
3460 * loop-unroll.c (compare_and_jump_seq): Update.
3461 (unroll_loop_runtime_iterations): Update.
3462 * lra-constraints.c (lra_inheritance): Update.
3463 * lto-streamer-in.c (input_cfg): Update.
3464 * lto-streamer-out.c (output_cfg): Update.
3465 * mcf.c (adjust_cfg_counts): Update.
3466 * modulo-sched.c (sms_schedule): Update.
3467 * omp-expand.c (expand_omp_for_init_counts): Update.
3468 (extract_omp_for_update_vars): Update.
3469 (expand_omp_ordered_sink): Update.
3470 (expand_omp_for_ordered_loops): Update.
3471 (expand_omp_for_generic): Update.
3472 (expand_omp_for_static_nochunk): Update.
3473 (expand_omp_for_static_chunk): Update.
3474 (expand_cilk_for): Update.
3475 (expand_omp_simd): Update.
3476 (expand_omp_taskloop_for_outer): Update.
3477 (expand_omp_taskloop_for_inner): Update.
3478 * omp-simd-clone.c (simd_clone_adjust): Update.
3479 * optabs.c (expand_doubleword_shift): Update.
3480 (expand_abs): Update.
3481 (emit_cmp_and_jump_insn_1): Update.
3482 (expand_compare_and_swap_loop): Update.
3483 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3484 * predict.c (predictable_edge_p): Update.
3485 (edge_probability_reliable_p): Update.
3486 (set_even_probabilities): Update.
3487 (combine_predictions_for_insn): Update.
3488 (combine_predictions_for_bb): Update.
3489 (propagate_freq): Update.
3490 (estimate_bb_frequencies): Update.
3491 (force_edge_cold): Update.
3492 * profile-count.c (profile_count::dump): Add missing space into dump.
3493 (profile_count::debug): Add newline.
3494 (profile_count::differs_from_p): Explicitly convert to unsigned.
3495 (profile_count::stream_in): Update.
3496 (profile_probability::dump): New member function.
3497 (profile_probability::debug): New member function.
3498 (profile_probability::differs_from_p): New member function.
3499 (profile_probability::differs_lot_from_p): New member function.
3500 (profile_probability::stream_in): New member function.
3501 (profile_probability::stream_out): New member function.
3502 * profile-count.h (profile_count_quality): Rename to ...
3503 (profile_quality): ... this one.
3504 (profile_probability): New.
3505 (profile_count): Update.
3506 * profile.c (compute_branch_probabilities): Update.
3507 * recog.c (peep2_attempt): Update.
3508 * sched-ebb.c (schedule_ebbs): Update.
3509 * sched-rgn.c (find_single_block_region): Update.
3510 (compute_dom_prob_ps): Update.
3511 (schedule_region): Update.
3512 * sel-sched-ir.c (compute_succs_info): Update.
3513 * stmt.c (struct case_node): Update.
3514 (do_jump_if_equal): Update.
3515 (get_outgoing_edge_probs): Update.
3516 (conditional_probability): Update.
3517 (emit_case_dispatch_table): Update.
3518 (expand_case): Update.
3519 (expand_sjlj_dispatch_table): Update.
3520 (emit_case_nodes): Update.
3521 * targhooks.c: Update.
3522 * tracer.c (better_p): Update.
3523 (find_best_successor): Update.
3524 * trans-mem.c (expand_transaction): Update.
3525 * tree-call-cdce.c: Update.
3526 * tree-cfg.c (gimple_split_edge): Upate.
3527 (move_sese_region_to_fn): Upate.
3528 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3529 * tree-eh.c (lower_resx): Upate.
3530 (cleanup_empty_eh_move_lp): Upate.
3531 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3532 * tree-inline.c (copy_edges_for_bb): Update.
3533 (copy_cfg_body): Update.
3534 * tree-parloops.c (gen_parallel_loop): Update.
3535 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3536 (gimple_gen_time_profiler): Update.
3537 * tree-ssa-dce.c (remove_dead_stmt): Update.
3538 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3539 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3540 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3541 (unloop_loops): Update.
3542 (try_peel_loop): Update.
3543 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3544 * tree-ssa-loop-split.c (connect_loops): Update.
3545 (split_loop): Update.
3546 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3547 (hoist_guard): Update.
3548 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3549 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3550 (value_replacement): Update.
3551 * tree-ssa-reassoc.c (branch_fixup): Update.
3552 * tree-ssa-tail-merge.c (replace_block_by): Update.
3553 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3554 (create_edge_and_update_destination_phis): Update.
3555 (compute_path_counts): Update.
3556 (recompute_probabilities): Update.
3557 (update_joiner_offpath_counts): Update.
3558 (freqs_to_counts_path): Update.
3559 (duplicate_thread_path): Update.
3560 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3561 (struct switch_conv_info): Update.
3562 (gen_inbound_check): Update.
3563 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3564 (vect_do_peeling): Update.
3565 (vect_loop_versioning): Update.
3566 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3567 (optimize_mask_stores): Update.
3568 * ubsan.c (ubsan_expand_null_ifn): Update.
3569 * value-prof.c (gimple_divmod_fixed_value): Update.
3570 (gimple_divmod_fixed_value_transform): Update.
3571 (gimple_mod_pow2): Update.
3572 (gimple_mod_pow2_value_transform): Update.
3573 (gimple_mod_subtract): Update.
3574 (gimple_mod_subtract_transform): Update.
3575 (gimple_ic): Update.
3576 (gimple_stringop_fixed_value): Update.
3577 (gimple_stringops_transform): Update.
3578 * value-prof.h: Update.
3580 2017-06-29 Carl Love <cel@us.ibm.com>
3582 * config/rs6000/rs6000-c.c: Add support for built-in functions
3583 vector signed int vec_signed (vector float);
3584 vector signed long long vec_signed (vector double);
3585 vector signed int vec_signed2 (vector double, vector double);
3586 vector signed int vec_signede (vector double);
3587 vector signed int vec_signedo (vector double);
3588 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3589 instruction generator.
3590 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3591 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3592 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3594 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3595 vunsignede_v2df): Add define_expands.
3596 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3597 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3598 VEC_UNSIGNEDO): Add definitions.
3599 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3600 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3601 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3602 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3603 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3604 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3605 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3606 vec_unsignede, vec_unsignedo): Add builtin defines.
3607 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3609 * doc/extend.texi: Update the built-in documentation file for the
3610 new built-in functions.
3612 2017-06-29 Richard Biener <rguenther@suse.de>
3614 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3615 reduction chains to LOOP_VINFO_REDUCTIONS.
3616 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3617 SLP reductions after processing reduction chains.
3619 2017-06-29 Nathan Sidwell <nathan@acm.org>
3621 * builtins.c (fold_builtin_FUNCTION): Use
3622 lang_hooks.decl_printable_name.
3624 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3627 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3628 with only one label.
3629 * stmt.c (expand_case): Assert NCASES is greater than one.
3631 2017-06-29 Richard Biener <rguenther@suse.de>
3633 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3635 (group_case_labels): Likewise.
3636 (find_taken_edge): Push sanity checking on val to workers...
3637 (find_taken_edge_cond_expr): ... here
3638 (find_taken_edge_switch_expr): ... and here, handle cases
3639 with just a default label.
3640 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3641 (group_case_labels): Likewise.
3642 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3643 group_case_labels does anything cleanup the CFG again.
3645 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3647 PR tree-optimization/81196
3648 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3649 exit condition comparing two IVs.
3651 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3653 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3654 profile to the dummy entry at the end of the list of architectures.
3655 * config/arm/arm-cpu-cdata.h: Regenerated.
3657 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3658 Michael Collison <michael.collison@arm.com>
3661 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3663 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3664 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3665 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3666 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3667 of shift when the shift amount is masked with constant equal to
3668 the size of the mode.
3669 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3672 2017-06-29 Martin Liska <mliska@suse.cz>
3674 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3676 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3679 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3682 (edge_growth_cache_entry::edge_growth_cache_entry): New
3684 (reset_edge_growth_cache): Update to use constructor.
3686 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3688 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3689 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3690 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3692 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3694 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3695 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3697 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3699 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3700 (*-linux-uclibc*): Add t-uclibc tmake_file.
3701 * config/t-musl: New.
3702 * config/t-uclibc: New.
3704 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3706 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3708 (gen_comm_data): Emit architectural setting of arch_prof.
3709 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3711 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3712 (armv8-m.base, armv8-m.main): Likewise.
3713 * arm-protos.h (arm_build_target): Add profile field.
3714 (arch_option): Likewise.
3715 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3717 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3718 arm_active_target.profile.
3720 2017-06-28 Richard Biener <rguenther@suse.de>
3723 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3724 TYPE_OVERFLOW_WRAPS.
3725 * match.pd (negate_expr_p): Likewise.
3726 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3727 fold_build2, not fold_binary.
3729 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3731 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3732 Convert memory address to Pmode.
3733 (aarch64_print_operand): Assert MEM operands are always Pmode.
3735 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3738 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3739 Remove redundant if.
3740 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3741 * config/arm/aarch-common-protos.h
3742 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3743 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3745 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3748 * multiple_target.c (create_dispatcher_calls): Set the default
3749 clone to be static, not public.
3751 2017-06-28 Richard Biener <rguenther@suse.de>
3753 * tree-vect-loop.c (vectorizable_reduction): Move special
3754 cond reduction IV var creation ...
3755 (vect_create_epilog_for_reduction): ... here. Remove induction_index
3756 parameter. Use STMT_VINFO_VECTYPE.
3757 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
3760 2017-06-28 Martin Liska <mliska@suse.cz>
3763 * ipa-visibility.c (non_local_p): Handle visibility.
3765 2017-06-28 Martin Liska <mliska@suse.cz>
3768 * common.opt: Add IntegerRange to various options.
3769 * opt-functions.awk (integer_range_info): New function.
3770 * optc-gen.awk: Add integer_range_info to cl_options struct.
3771 * opts-common.c (decode_cmdline_option): Handle
3772 CL_ERR_INT_RANGE_ARG.
3773 (cmdline_handle_error): Likewise.
3774 * opts.c (print_filtered_help): Show valid interval in
3775 when --help is provided.
3776 * opts.h (struct cl_option): Add range_min and range_max fields.
3777 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
3779 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
3781 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
3782 (x * C EQ/NE y * C): New transformation.
3784 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
3786 * genmultilib (combination_space): Accept '+' in option names.
3788 2017-06-28 Martin Liska <mliska@suse.cz>
3791 * asan.c (instrument_derefs): Bail out inner references
3792 that are hard register variables.
3794 2017-06-28 Jakub Jelinek <jakub@redhat.com>
3797 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
3798 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
3800 2017-06-28 Richard Biener <rguenther@suse.de>
3802 * tree-vectorizer.h (vect_get_vec_defs): Remove.
3803 (vect_get_slp_defs): Adjust.
3804 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
3806 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
3808 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3809 get_initial_defs_for_reduction instead of vect_get_vec_defs.
3810 (vectorizable_reduction): Adjust.
3811 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
3813 (vect_get_slp_defs): Likewise.
3814 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
3815 (vectorizable_bswap): Adjust.
3816 (vectorizable_call): Likewise.
3817 (vectorizable_conversion): Likewise.
3818 (vectorizable_assignment): Likewise.
3819 (vectorizable_shift): Likewise.
3820 (vectorizable_operation): Likewise.
3821 (vectorizable_store): Likewise.
3822 (vectorizable_condition): Likewise.
3823 (vectorizable_comparison): Likewise.
3825 2017-06-28 Michael Collison <michael.collison@arm.com>
3828 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
3830 (arm_gen_movmemqi): Removed unused variable 'i'.
3831 Convert 'for' loop into 'while' loop.
3832 (arm_expand_prologue): Remove last unnecessary set of insn.
3833 (thumb_pop): Remove unused variable 'pushed_words'.
3834 (thumb_exit): Remove last unnecessary set of regs_to_pop.
3836 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3838 * config/s390/predicates.md: Use s390_rel_address_ok_p.
3839 * config/s390/s390-protos.h: Add prototype of
3840 s390_rel_address_ok_p.
3841 * config/s390/s390.c (s390_got_symbol): New function.
3842 (s390_rel_address_ok_p): New function.
3843 (legitimize_pic_address): Use s390_rel_address_ok_p.
3844 (s390_load_got): Use s390_got_symbol.
3845 (s390_option_override): Issue error if
3846 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
3847 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
3849 * config/s390/s390.opt: New option mpic-data-is-text-relative.
3851 2017-06-27 Andrew Pinski <apinski@cavium.com>
3853 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
3854 (X * copysign (1.0, X)): New pattern.
3855 (X * copysign (1.0, -X)): New pattern.
3856 (copysign (-1.0, CST)): New pattern.
3858 2017-06-27 Joseph Myers <joseph@codesourcery.com>
3860 * genmultilib (combination_space): Remove variable.
3861 Validate reuse rules against regular expression for any sequence
3862 of multilib options in any order.
3864 2017-06-27 Michael Collison <michael.collison@arm.com>
3866 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
3867 call aarch64_split_simd_combine.
3868 * (aarch64_combine_internal<mode>): Delete pattern.
3869 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
3870 Allow register and subreg operands.
3872 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3874 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
3875 specific need, just fallback on defaults.
3876 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
3878 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3879 Olivier Hainque <hainque@adacore.com>
3881 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
3883 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
3884 targets. Pick a default if no particular attempt applied.
3885 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
3888 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3890 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
3891 (x86_64-wrs-vxworks7): Likewise.
3893 2017-06-27 Marek Polacek <polacek@redhat.com>
3896 * ubsan.c (instrument_null): Check get_base_address's result for null.
3898 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3900 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
3902 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3904 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
3905 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
3906 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
3907 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
3909 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
3910 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
3911 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
3912 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
3913 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
3914 BUILT_IN_FEUPDATEENV): New builtins.
3915 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
3916 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
3917 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
3918 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
3920 (builtin_structptr_types): Adjust size.
3921 * tree.c (builtin_structptr_types): Add four entries.
3923 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3924 Olivier Hainque <hainque@adacore.com>
3926 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
3927 (TLS_SYM): New local macro, forcing reference to __tls__ on
3928 link command lines for VxWorks 7 RTPs, triggering initialization
3930 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
3931 OS features TLS support, true for RTPs on VxWorks 7.
3932 * config/vxworks.c (vxworks_override_options): Setup emutls
3935 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3937 * predict.c (test_prediction_value_range): Use -1U instead of -1
3938 to avoid narrowing conversion warning.
3939 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
3940 to avoid narrowing conversion warning.
3941 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
3943 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
3945 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3947 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
3948 64bit configurations.
3949 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
3950 (SIZE_TYPE): Likewise.
3951 * config/vxworks.c (vxworks_emutls_var_fields): Use
3952 long_unsigned_type_node instead of unsigned_type_node as the offset
3953 field type, which is "pointer" mode in emutls.c.
3955 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3958 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
3961 * gimple-fold.c (replace_call_with_call_and_fold): Handle
3962 gimple_vuse copying separately from gimple_vdef copying.
3964 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3966 * value-prof.c (free_hist): Remove call to memset and the enclosing if
3969 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3970 Olivier Hainque <hainque@adacore.com>
3972 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
3973 for all vxworks7 targets.
3974 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
3975 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
3976 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
3977 variations for VX6/VX7 and 32/64bits later on in ...
3978 (VXWORKS_LIB_SPEC): Leverage new macros.
3979 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
3980 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
3982 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3984 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
3985 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
3987 2017-06-26 Carl Love <cel@us.ibm.com>
3989 * config/rs6000/rs6000-c.c: Add support for built-in functions
3990 vector bool char vec_reve (vector bool char);
3991 vector signed char vec_reve (vector signed char);
3992 vector unsigned char vec_reve (vector unsigned char);
3993 vector bool int vec_reve (vector bool int);
3994 vector signed int vec_reve (vector signed int);
3995 vector unsigned int vec_reve (vector unsigned int);
3996 vector bool long long vec_reve (vector bool long long);
3997 vector signed long long vec_reve (vector signed long long);
3998 vector unsigned long long vec_reve (vector unsigned long long);
3999 vector bool short vec_reve (vector bool short);
4000 vector signed short vec_reve (vector signed short);
4001 vector double vec_reve (vector double);
4002 vector float vec_reve (vector float);
4003 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4004 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4005 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4006 (altivec_vreve): New pattern.
4007 * config/rs6000/altivec.h (vec_reve): New define.
4008 * doc/extend.texi (vec_rev): Update the built-in documentation file
4009 for the new built-in functions.
4011 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4013 PR tree-optimization/71815
4014 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4016 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4018 (slsr_process_phi): Likewise.
4019 (replace_uncond_cands_and_profitable_phis): Don't replace a
4020 multiply candidate with a stride of 1 (copy or cast).
4021 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4023 (lowest_cost_path): Likewise.
4024 (total_savings): Likewise.
4026 2017-06-26 Richard Biener <rguenther@suse.de>
4029 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4030 Use def_builtin_pure for all gather builtins.
4032 2017-06-26 Richard Biener <rguenther@suse.de>
4034 PR tree-optimization/81203
4035 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4038 2017-06-26 Marek Polacek <polacek@redhat.com>
4041 * doc/invoke.texi: Document -Wmultistatement-macros.
4043 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4045 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4046 arm_neon_ok_no_float_abi effective target.
4048 2017-06-26 Richard Biener <rguenther@suse.de>
4050 PR tree-optimization/80928
4051 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4052 (copy_bbs): Set BB_DUPLICATED flag early.
4053 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4055 (execute_on_shrinking_pred): Likewise.
4056 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4057 BB_DUPLICATED blocks.
4058 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4059 iterate over all PHIs considering removal of *gsi.
4061 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4063 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4066 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4068 * config/rs6000/rs6000-string.c: (expand_block_clear,
4069 do_load_for_compare, select_block_compare_mode,
4070 compute_current_alignment, expand_block_compare,
4071 expand_strncmp_align_check, expand_strn_compare,
4072 expand_block_move, rs6000_output_load_multiple)
4073 Move functions related to string/block move/compare
4075 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4076 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4077 for this function which is now used in two files.
4078 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4079 * config.gcc: Add rs6000-string.o to extra_objs for
4080 targets powerpc*-*-* and rs6000*-*-*.
4082 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4085 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4086 32-bit, since indexed is not valid for DImode.
4087 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4088 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4089 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4090 (define_peephole2 for Altivec d-form store): Likewise.
4093 * multiple_target.c (create_dispatcher_calls): Only create the
4094 dispatcher call if the function is the default clone of a
4097 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4100 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4101 a call, force the call to not be a tail call.
4103 2017-06-23 Jeff Law <law@redhat.com>
4105 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4108 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4110 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4111 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4112 handling for early expansion of vector shifts (sl,sr,sra,rl).
4113 (builtin_function_type): Add vector shift right instructions
4114 to the unsigned argument list.
4116 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4118 rtl-optimizatoin/79286
4119 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4120 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4121 trap. PIC register plus a const unspec without offset can never trap.
4123 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4125 * tree.h (builtin_structptr_type): New type.
4126 (builtin_structptr_types): Declare new array.
4127 * tree.c (builtin_structptr_types): New array.
4128 (free_lang_data, build_common_tree_nodes): Use it.
4130 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4133 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4136 2017-06-22 Matt Turner <mattst88@gmail.com>
4138 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4139 Lake models to skylake case. Assume skylake for unknown
4140 models with clflushopt.
4142 2017-06-22 Jeff Law <law@redhat.com>
4144 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4145 frame sizes that do not satisfy aarch64_uimm12_shift.
4147 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4149 * profile-count.h (apply_probability,
4150 apply_scale, probability_in): Fix checks for zero.
4152 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4154 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4155 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4157 2016-06-22 Richard Biener <rguenther@suse.de>
4159 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4160 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4161 REDUC_MAX_EXPR support.
4162 (vectorizable_reduction): Likewise.
4163 (vect_create_epilog_for_reduction): Likewise.
4165 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4167 * match.pd (A / (1 << B) -> A >> B): New.
4168 * generic-match-head.c: Include optabs-tree.h.
4169 * gimple-match-head.c: Likewise.
4170 * optabs-tree.h (target_supports_op_p): New.
4171 * optabs-tree.c (target_supports_op_p): New.
4173 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4175 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4176 $gcc_cv_ld --help output.
4177 (gcc_cv_ld_demangle): Likewise.
4178 (gcc_cv_ld_eh_frame_hdr): Likewise.
4179 (gcc_cv_ld_pie): Likewise.
4180 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4181 (gcc_cv_ld_buildid): Likewise.
4182 (gcc_cv_ld_sysroot): Likewise.
4183 (ld_bndplt_support): Likewise.
4184 (ld_pushpopstate_support): Likewise.
4185 * configure: Regenerate.
4186 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4188 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4191 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4192 operands indexes to avoid gap between operands and match_dups.
4194 2017-06-21 Andrew Pinski <apinski@cavium.com>
4196 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4197 Increment Arith_shift and Arith_shift_reg by 1.
4198 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4200 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4201 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4202 (aarch64_strip_extend): Add new argument and test for it.
4203 (aarch64_cheap_mult_shift_p): New function.
4204 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4205 add a cost if it is true.
4206 Update calls to aarch64_strip_extend.
4207 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4209 2017-06-21 Andrew Pinski <apinski@cavium.com>
4211 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4213 (thunderxt88): Likewise.
4214 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4215 (thunderx_prefetch_tune): New variable.
4216 (thunderx2t99_prefetch_tune): Update for the correct values.
4217 (thunderxt88_tunings): New variable.
4218 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4219 generic_prefetch_tune.
4220 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4222 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4224 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4225 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4226 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4227 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4228 (aarch64_atomic_cas<mode>, GPI): Likewise.
4230 2017-06-21 Martin Liska <mliska@suse.cz>
4232 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4233 statements on cold and hot labels.
4234 * predict.c (tree_estimate_probability_bb): Remove the
4235 prediction from this place.
4237 2017-06-21 Martin Liska <mliska@suse.cz>
4239 PR tree-optimization/79489
4240 * gimplify.c (maybe_add_early_return_predict_stmt): New
4242 (gimplify_return_expr): Call the function.
4243 * predict.c (tree_estimate_probability_bb): Remove handling
4245 * predict.def: Update comment about early return predictor.
4246 * gimple-predict.h (is_gimple_predict): New function.
4247 * predict.def: Change default value of early return to 66.
4248 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4250 * passes.def: Put pass_strip_predict_hints to the beginning of
4253 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4255 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4256 FUNCTION_DECL declarations.
4257 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4259 (dwaf2out_decl): Likewise.
4260 * godump.c (go_early_global_decl): Skip call to the real debug hook
4261 for FUNCTION_DECL declarations.
4262 * passes.c (rest_of_decl_compilation): Skip call to the
4263 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4264 -fdump-go-spec is passed.
4266 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4268 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4269 Reorder for compactness.
4270 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4271 (def_builtin_pure, def_builtin_pure2): New functions.
4272 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4274 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4276 * match.pd (nop_convert): New predicate.
4277 ((A +- CST1) +- CST2): Allow some NOP conversions.
4279 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4282 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4283 with ctors/dtors if GOVD_SHARED is set.
4285 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4287 * config/aarch64/aarch64.md (movti_aarch64):
4288 Emit mov rather than orr.
4289 (movtf_aarch64): Likewise.
4290 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4291 Emit mov rather than orr.
4293 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4295 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4296 Swap alternatives, make integer dup more expensive.
4298 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4300 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4301 Return true for non-tls symbols.
4303 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4305 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4306 (cortex-a75): Likewise.
4307 (cortex-a75.cortex-a55): Likewise.
4308 * config/aarch64/aarch64-tune.md: Regenerate.
4309 * doc/invoke.texi (-mtune): Document new values for -mtune.
4311 2017-06-21 Tom de Vries <tom@codesourcery.com>
4313 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4315 (Effective-Target Keywords, Other attributes): Suggest using
4316 dg-add-options stack_size feature to get stack limit in stack_size
4317 effective target documentation.
4319 2017-06-21 Julian Brown <julian@codesourcery.com>
4320 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4322 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4323 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4324 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4326 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4327 attribute type list for neon_multiply.
4328 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4329 attribute type list for neon_multiply.
4330 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4331 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4332 attribute type list for neon_multiply.
4333 * config/arm/types.md (crypto_pmull): Add.
4334 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4335 attribute type list.
4337 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4339 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4342 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4344 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4345 to make sure not to dereference a NULL cost_classes_ptr pointer.
4347 2017-06-20 Carl Love <cel@us.ibm.com>
4349 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4350 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4351 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4352 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4353 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4354 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4355 VMULOSW): New enum "unspec" values.
4356 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4357 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4358 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4359 altivec_vmulosw): New patterns.
4360 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4361 VMULOSW): Add definitions.
4363 2017-06-20 Julia Koval <julia.koval@intel.com>
4365 * config/i386/i386.c: Fix rounding expand for new pattern.
4366 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4368 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4370 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4371 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4373 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4375 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4378 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4380 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4381 architecture, then by implementer ID.
4382 * config/aarch64/aarch64-tune.md: Regenerate.
4384 2017-06-20 Richard Biener <rguenther@suse.de>
4387 * fold-const.c (split_tree): Fold to type before negating.
4389 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4391 * diagnostic-show-locus.c
4392 (selftest::test_fixit_deletion_affecting_newline): New function.
4393 (selftest::diagnostic_show_locus_c_tests): Call it.
4395 2017-06-20 Andreas Schwab <schwab@suse.de>
4398 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4401 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4403 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4404 __ARM_FEATURE_COPROC according to support.
4406 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4408 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4409 Rewritten to avoid overflow for > 32-bit pointers.
4412 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4413 by removing enum keyword.
4414 (ubsan_type_descriptor): Likewise. Formatting fix.
4417 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4418 splitter): Require TARGET_SSE2 in the condition.
4420 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4423 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4424 for doing vector set of SFmode on ISA 3.0.
4425 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4426 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4428 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4429 SFmode value into a V4SF variable that was extracted from another
4430 V4SF variable without converting the element to double precision
4431 and back to single precision vector format.
4432 (vsx_insert_extract_v4sf_p9_2): Likewise.
4434 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4436 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4437 in UWHI to avoid undefined overflow.
4440 * ubsan.h (enum ubsan_encode_value_phase): New.
4441 (ubsan_encode_value): Change second argument to
4442 enum ubsan_encode_value_phase with default value of
4443 UBSAN_ENCODE_VALUE_GENERIC.
4444 * ubsan.c (ubsan_encode_value): Change second argument to
4445 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4446 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4447 create_tmp_var_raw instead of create_tmp_var and use a
4449 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4450 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4451 ubsan_encode_value callers.
4454 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4455 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4456 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4458 2017-06-19 Richard Biener <rguenther@suse.de>
4461 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4462 estimates if we changed anything.
4464 2017-06-19 Richard Biener <rguenther@suse.de>
4466 PR tree-optimization/80887
4467 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4468 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4469 simplified lookups, then reset mprts_hook.
4470 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4472 (try_to_simplify): Likewise.
4474 2017-06-19 Martin Liska <mliska@suse.cz>
4477 * gimplify.c (gimplify_switch_expr):
4478 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4480 2017-06-19 Martin Liska <mliska@suse.cz>
4482 * doc/install.texi: Document that PGO runs in 4 stages.
4484 2017-06-19 Martin Liska <mliska@suse.cz>
4487 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4488 to dispatcher function name.
4489 * multiple_target.c (replace_function_decl): New function.
4490 (create_dispatcher_calls): Redirect both edges and references.
4492 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4494 * profile-count.c (profile_count::dump): Dump quality.
4495 (profile_count::differs_from_p): Update for unsigned val.
4496 * profile-count.h (profile_count_quality): New enum.
4497 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4499 2017-06-19 Richard Biener <rguenther@suse.de>
4501 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4502 struct function as arg.
4503 (estimate_numbers_of_iterations): Export overload with loop arg.
4504 (free_numbers_of_iterations_estimates_loop): Use an overload of
4505 free_numbers_of_iterations_estimates instead.
4506 * tree-cfg.c (remove_bb): Adjust.
4507 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4508 * tree-parloops.c (gen_parallel_loop): Likewise.
4509 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4511 (tree_unroll_loops_completely): Likewise.
4512 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4513 Use an overload instead and export.
4514 (estimated_loop_iterations): Adjust.
4515 (max_loop_iterations): Likewise.
4516 (likely_max_loop_iterations): Likewise.
4517 (estimate_numbers_of_iterations): Take struct function as arg
4519 (loop_exits_before_overflow): Adjust.
4520 (free_numbers_of_iterations_estimates_loop): Use an overload.
4521 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4522 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4524 2017-06-19 Richard Biener <rguenther@suse.de>
4527 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4528 RANGE_EXPR conservatively.
4530 2017-06-16 Carl Love <cel@us.ibm.com>
4532 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4533 definitions for vec_float, vec_float2, vec_floato,
4534 vec_floate built-ins.
4535 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4536 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4538 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4539 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4540 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4541 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4542 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4543 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4544 vec_floato): Add builtin defines.
4545 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4546 Update the built-in documentation file for the new built-in
4549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4551 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4552 (mthumb): Mark as the negative of -marm.
4554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4556 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4558 (ARM Options, -mtune): Document that this accepts the same
4559 extension options as -mcpu.
4560 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4562 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4564 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4565 permitted extensions.
4567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4569 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4570 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4571 (armv8-m.main): Add option +nodsp.
4572 * config/arm/arm-cpu-cdata.h: Regenerated.
4574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4576 * config/arm/t-fuchsia: New file.
4577 * config.gcc (arm*-*-fuchsia*): Use it.
4579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4581 * config/arm/t-symbian: Rewrite for new option infrastructure.
4583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4585 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4586 (MULTILIB_REQUIRED): Likewise.
4588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4590 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4591 (MULTILIB_RESUE): Likewise.
4592 (MULTILIB_MATCHES): Likewise.
4593 (MULTLIB_REQUIRED): Likewise.
4595 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4597 * config/arm/t-rtems: Rewrite for new option framework.
4599 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4601 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4602 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4603 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4604 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4605 * config/arm/t-multilib: ... here.
4606 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4607 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4608 armv7-a and armv8*-a when A-profile libraries have not been built.
4609 * config/arm/t-rmprofile: Rewrite.
4611 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4613 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4614 with a backslash. Remove the backslash after substituting unescaped
4616 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4620 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4621 the driver to pass through the default setting of -mfloat-abi.
4622 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4624 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4625 (all_feat_combs): New rule.
4626 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4628 * config/arm/t-aprofile: Rewrite.
4630 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4632 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4633 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4634 fpu is not specified by the user/command-line.
4635 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4636 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4637 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4638 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4639 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4640 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4644 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4645 * config/arm/t-arm-elf: Rewritten.
4647 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4649 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4650 have some floating-point instructions.
4651 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4652 (TARGET_MAYBE_HARD_FLOAT): New macro.
4653 * config/arm/arm-builtins.c (arm_init_builtins): Use
4654 TARGET_MAYBE_HARD_FLOAT.
4655 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4659 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4660 (configargs.h): Include it.
4661 (arm_print_hint_for_fpu_option): New function.
4662 (arm_parse_fpu_option): New function.
4663 (candidate_extension): New class.
4664 (arm_canon_for_multilib): New function.
4665 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4666 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4667 (ARCH_CANONICAL_SPECS): New macro.
4668 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4672 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4673 are set after handling multilib fragments. Set target_cpu_default2
4676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4678 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4680 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4681 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4683 (check_arch): Likewise.
4684 * config/arm/arm.c (arm_configure_build_target): Handle
4685 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4686 options in the default.
4688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4690 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4691 when an option is an alias of another.
4692 * config/arm/parsecpu.awk (optalias): New parser token.
4693 (gen_comm_data): Mark non-alias options as such. Emit entries
4694 for extension aliases.
4695 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4696 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4697 (armv6kz, armv6zk, armv6t2): Likewise.
4698 (armv7): Make vfpv3-d16 an alias.
4699 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4701 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4702 Sort in canonical order.
4703 (armv8-a): Sort in canonical order.
4704 (armv8.1-a, armv8.2-a): Likewise.
4705 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4707 (cortex-a9): Sort in canonical order.
4708 * config/arm/arm.c (selftests.h): Include it.
4709 (arm_test_cpu_arch_data): New function.
4710 (arm_run_self_tests): New function.
4711 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4712 (targetm): Move declaration to the end of the file.
4713 * arm-cpu-cdata.h: Regenerated.
4715 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4717 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4718 call to target_mode_check describing the type of option passed.
4719 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4720 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4721 arm_parse_cpu_option_name to match parameters against list of
4723 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4724 arm_arch_core_flags data structure.
4725 * config/arm/arm-cpu_cdata.h: Regenerated.
4727 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4729 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4731 (arm_print_hint_for_cpu_option): Likewise.
4732 (arm_print_hint_for_arch_option): Likewise.
4733 (arm_parse_cpu_option_name): Likewise.
4734 (arm_parse_arch_option_name): Likewise.
4735 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4736 of entries in the all_fpus list.
4737 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4738 (arm_parse_cpu_option_name): Declare.
4739 (arm_parse_arch_option_name): Declare.
4740 (arm_parse_option_features): Declare.
4741 (arm_intialize_isa): Declare.
4742 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4744 (gen_comm_data): ... here. Make definitions non-static.
4745 * config/arm/arm-cpu-data.h: Regenerated.
4746 * config/arm/arm-cpu-cdata.h: Regenerated.
4748 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4750 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
4751 (cpu_arch_extension): New structure.
4752 (cpu_arch_option, arch_option, cpu_option): New structures.
4753 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
4755 (gen_data): Generate new format data tables.
4756 * config/arm/arm.c (cpu_tune): New structure.
4757 (cpu_option, processors): Delete.
4758 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
4759 (arm_print_hint_for_cpu_option): ... this and ...
4760 (arm_print_hint_for_arch_option): ... this.
4761 (arm_parse_arch_cpu_name): Delete. Replace with ...
4762 (arm_parse_cpu_option_name): ... this and ...
4763 (arm_parse_arch_option_name): ... this.
4764 (arm_unrecognized_feature): Change type of target parameter to
4766 (arm_parse_arch_cpu_features): Delete. Replace with ...
4767 (arm_parse_option_features): ... this.
4768 (arm_configure_build_target): Rework to use new configuration data
4770 (arm_print_tune_info): Rework for new configuration data tables.
4771 * config/arm/arm-cpu-data.h: Regenerated.
4772 * config/arm/arm-cpu.h: Regenerated.
4774 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4776 * Makefile.in (OBJS): Move sbitmap.o from here ...
4777 (OBJS-libcommon): ... to here.
4779 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4781 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
4782 (ISA_ALL_CRYPTO): New macro.
4783 (ISA_ALL_SIMD): New macro
4784 (ISA_ALL_FP): New macro.
4785 * config/arm/arm.c (fpu_bitlist): Update initializer.
4786 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
4788 (arm9e): Add fpu. Add option for nofp
4789 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
4790 (arm926ej-s, arm1026ej-s): Likewise.
4791 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
4792 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
4793 neon-fp16, neon-vfpv4, nofp and nosimd.
4794 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
4795 (cortex-a8): Add fpu. Add option for nofp.
4796 (cortex-a9): Add fpu. Add options for nosimd and nofp.
4797 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
4798 (cortex-r4f): Add fpu.
4799 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
4800 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
4802 (cortex-r8): Likewise.
4803 (cortex-m4): Add fpu. Add option for nofp.
4804 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
4805 (cortex-a17.cortex-a7): Likewise.
4806 (cortex-a32): Add fpu. Add options for crypto and nofp.
4807 (cortex-a35, cortex-a53): Likewise.
4808 (cortex-a57): Add fpu. Add option for crypto.
4809 (cortex-a72, cortex-a73): Likewise.
4810 (exynos-m1): Likewise.
4811 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
4812 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
4813 (cortex-m33): Add fpu. Add option for nofp.
4814 * config/arm/arm-cpu-cdata.h: Regenerated
4815 * config/arm/arm-cpu-data.h: Regenerated.
4817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4819 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
4820 (armv5te, armv5tej): Likewise.
4821 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
4822 (armv7): Add options fp and vfpv3-d16.
4823 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
4824 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
4826 (armv7ve): Likewise.
4827 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
4828 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
4829 (armv8-a): Add nocrypto option.
4830 (armv8.1-a, armv8.2-a): Likewise.
4831 (armv8-m.main): add options fp, fp.dp and nofp.
4833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4835 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
4837 (armv8-a+crc): Delete.
4838 (armv8.1-a): Add options simd, crypto and nofp.
4839 (armv8.2-a): Add options fp16, simd, crypto and nofp.
4840 (armv8.2-a+fp16): Delete.
4841 (armv8-m.main): Add option dsp.
4842 (armv8-m.main+dsp): Delete.
4843 (cortex-a8): Add fpu. Add nofp option.
4844 (cortex-a9): Add fpu. Add nofp and nosimd options.
4845 * config/arm/parsecpu.awk (gen_data): Generate option tables and
4846 link to main cpu and architecture data structures.
4847 (gen_comm_data): Only put isa attributes from the main architecture
4849 (option): New statement for architecture and CPU entries.
4850 * arm.c (struct cpu_option): New structure.
4851 (struct processors): Add entry for options.
4852 (arm_unrecognized_feature): New function.
4853 (arm_parse_arch_cpu_name): Ignore any characters after the first
4855 (arm_parse_arch_cpu_feature): New function.
4856 (arm_configure_build_target): Separate out any CPU and architecture
4857 features and parse separately. Don't error out if -mfpu=auto is
4858 used with only an architecture string.
4859 (arm_print_asm_arch_directives): New function.
4860 (arm_file_start): Call it.
4861 * config/arm/arm-cpu-cdata.h: Regenerated.
4862 * config/arm/arm-cpu-data.h: Likewise.
4863 * config/arm/arm-tables.opt: Likewise.
4865 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4867 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
4868 assembler when it is not -mfpu=auto.
4870 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4872 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
4873 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
4874 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
4875 (ASM_CPU_SPEC): Rewrite.
4876 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
4877 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
4878 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
4879 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
4880 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
4881 copied string is NUL-terminated. Also strip any characters prefixed
4883 (arm_rewrite_selected_arch): New function.
4884 (arm_rewrite_march): New function.
4886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4888 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
4889 (x_arm_cpu_string, x_arm_tune_string): Likewise.
4890 (march, mcpu, mtune): Convert to string-based options.
4891 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
4892 (arm_parse_arch_cpu_name): New function.
4893 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
4894 identify selected architecture or CPU.
4895 (arm_option_save): New function.
4896 (TARGET_OPTION_SAVE): Redefine.
4897 (arm_option_restore): Restore string options.
4898 (arm_option_print): Print string options.
4900 2017-06-16 Martin Sebor <msebor@redhat.com>
4902 PR tree-optimization/80933
4903 PR tree-optimization/80934
4904 * builtins.c (fold_builtin_3): Do not handle bcmp here.
4905 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
4906 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
4907 (gimple_fold_builtin): Call them.
4909 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4911 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
4912 as unlikely; update profile.
4914 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4916 * predict.c (force_edge_cold): Handle declaring edges impossible
4919 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4921 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
4923 (try_unroll_loop_completely): Fix reporting.
4925 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4927 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
4929 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
4932 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
4933 if given a non-constant argument for an intrinsic which requires a
4936 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4938 * profile.c (compare_freqs): New function.
4939 (branch_prob): Sort edge list.
4940 (find_spanning_tree): Assume that the list is priority sorted.
4942 2017-06-16 Richard Biener <rguenther@suse.de>
4944 PR tree-optimization/81090
4945 * passes.def (pass_record_bounds): Remove.
4946 * tree-pass.h (make_pass_record_bounds): Likewise.
4947 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
4948 make_pass_record_bounds): Likewise.
4949 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
4950 not free niter estimates at the beginning but at the end.
4951 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
4953 2017-06-16 Richard Biener <rguenther@suse.de>
4955 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
4956 initializer to workaround ICE in host GCC 4.8.
4958 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4960 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
4962 (clone_inlined_nodes): Update.
4964 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4966 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
4967 prefetch settings, and enable prefetching by default at -O3.
4969 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4971 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4972 Set flag_prefetch_loop_arrays according to tuning data.
4974 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4976 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
4978 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
4979 [Unrelated to main purpose of the patch] Place the pointer field last
4980 to enable type checking errors when tune structure are wrongly merged.
4981 * config/aarch64/aarch64.c (generic_prefetch_tune,)
4982 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
4983 (thunderx2t99_prefetch_tune): New tune constants.
4984 (tune_params *_tunings): Update all tunings (no functional change).
4985 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
4986 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
4987 from tunings structures.
4989 2017-06-16 Jakub Jelinek <jakub@redhat.com>
4992 * ubsan.c (instrument_null): Add T argument, use it instead
4993 of computing it based on IS_LHS.
4994 (instrument_object_size): Likewise.
4995 (pass_ubsan::execute): Adjust instrument_null and
4996 instrument_object_size callers to pass gimple_get_lhs or
4997 gimple_assign_rhs1 result to it. Use instrument_null instead of
4998 calling get_base_address and instrument_mem_ref. Handle
4999 aggregate call arguments for object-size sanitization.
5001 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5003 PR tree-optimization/81089
5004 * tree-vrp.c (is_masked_range_test): Validate operands of
5007 2017-06-15 Martin Sebor <msebor@redhat.com>
5010 * dumpfile.c (dump_register): Avoid calling memset to initialize
5011 a class with a default ctor.
5012 * gcc.c (struct compiler): Remove const qualification.
5013 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5014 * hash-table.h: Ditto.
5015 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5017 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5018 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5020 * params.h (struct param_info): Make struct members non-const.
5021 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5022 with default initialization.
5023 * vec.h (vec_copy_construct, vec_default_construct): New helper
5025 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5026 with vec_copy_construct.
5027 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5028 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5029 * doc/invoke.texi (-Wclass-memaccess): Document.
5031 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5033 * emit-rtl.h (is_leaf): Update comment about local
5036 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5039 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5040 for a variable to have a section before checking if the section has a
5042 Set section to.persistent if persistent attribute is set.
5043 Warn if .persistent attribute is used on an automatic variable.
5045 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5047 PR rtl-optimization/80474
5048 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5050 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5052 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5055 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5057 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5058 (casesi): Emit bounds checking as RTL.
5059 (casesi_internal_mips16_<mode>): Remove bounds checking.
5061 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5063 * config/xtensa/xtensa.c (xtensa_option_override): Append
5064 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5065 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5066 xtensa_doloop_hooks): Define unconditionally.
5067 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5069 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5070 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5071 for it in xtensa_option_override.
5072 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5073 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5075 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5077 * doc/cppopts.texi: Document '-' special value to -MF.
5079 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5081 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5082 (cortex_a53_fconst): Likewise.
5083 (cortex_a53_fpmul): Likewise.
5084 (cortex_a53_f_load_64): Likewise.
5085 (cortex_a53_f_load_many): Likewise.
5086 (cortex_a53_advsimd_alu): Likewise.
5087 (cortex_a53_advsimd_alu_q): Likewise.
5088 (cortex_a53_advsimd_mul): Likewise.
5089 (cortex_a53_advsimd_mul_q): Likewise.
5090 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5091 Add missing fmul, r2f_cvt and fconst cases.
5093 2017-06-14 Richard Biener <rguenther@suse.de>
5096 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5098 (fold_binary_loc): When associating do not treat pre-existing
5099 TREE_OVERFLOW on literal constants as a reason to allow
5100 TREE_OVERFLOW on associated literal constants.
5102 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5104 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5105 (MASK_FEATURES): New macro.
5106 * config/sparc/sparc.c (sparc_option_override): Remove the special
5107 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5109 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5111 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5112 a division of 0 if non-call exceptions are enabled.
5114 2017-06-14 Andrew Pinski <apinski@cavium.com>
5115 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5118 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5119 Improve vector initialization code gen for only variable case.
5121 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5123 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5125 2017-06-14 Richard Biener <rguenther@suse.de>
5127 PR tree-optimization/81083
5128 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5131 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5133 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5134 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5135 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5136 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5137 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5138 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5140 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5142 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5143 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5145 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5147 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5149 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5151 * config/rs6000/t-rtems: Don't handle SPE.
5153 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5155 * config/rs6000/t-linux: Don't handle SPE.
5157 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5159 * config/rs6000/eabispe.h: Delete file.
5161 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5163 * config/rs6000/t-spe: Delete file.
5165 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5167 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5168 (rs6000_legitimate_offset_address_p): Return false for anything in
5169 V2SImode or V2SFmode.
5171 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5173 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5174 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5175 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5177 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5178 (rs6000_legitimate_offset_address_p): Ditto.
5179 (rs6000_emit_move): Ditto.
5180 (rs6000_init_builtins): Remove V4HI_type_node.
5182 2017-06-13 Martin Liska <mliska@suse.cz>
5185 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5186 (gate_asan): Likewise.
5187 * asan.h (asan_no_sanitize_address_p): Remove the function.
5188 (sanitize_flags_p): New function.
5189 * builtins.def: Fix coding style.
5190 * common.opt: Use renamed enum value.
5191 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5192 * doc/extend.texi: Document no_sanitize attribute.
5193 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5194 to SANITIZE_UNDEFINED_NONDEFAULT.
5195 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5196 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5197 Use sanitize_flags_p.
5198 * gimplify.c (gimplify_function_tree): Likewise.
5199 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5200 * opts.c (parse_no_sanitize_attribute): New function.
5201 (common_handle_option): Use renamed enum value.
5202 * opts.h (parse_no_sanitize_attribute): Declare.
5203 * tree.c (sanitize_flags_p): New function.
5204 * tree.h: Declared here.
5205 * tsan.c: Use sanitize_flags_p.
5206 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5207 (instrument_mem_ref): Likewise.
5208 (instrument_bool_enum_load): Likewise.
5209 (do_ubsan_in_current_function): Remove the function.
5210 (pass_ubsan::execute): Use sanitize_flags_p.
5211 * ubsan.h: Remove do_ubsan_in_current_function
5212 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5213 (dump_function_to_file): Use it here.
5215 2017-06-13 Martin Jambor <mjambor@suse.cz>
5217 PR tree-optimization/80803
5218 PR tree-optimization/81063
5219 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5220 (propagate_subaccesses_across_link): Enqueue subtree whenever
5221 necessary instead of relying on the caller.
5223 2017-06-13 Martin Jambor <mjambor@suse.cz>
5225 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5226 that have a first_link.
5227 (sort_and_splice_var_accesses): Do not check first_link before
5229 (subtree_mark_written_and_enqueue): Likewise.
5230 (propagate_all_subaccesses): Likewise and do not stop at first
5231 parent with a first_link.
5233 2017-06-13 Martin Jambor <mjambor@suse.cz>
5235 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5238 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5240 * match.pd: New pattern.
5242 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5244 * tree-vrp.c (is_masked_range_test): New function.
5245 (register_edge_assert_for): Determine ranges for
5248 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5250 PR tree-optimization/67328
5251 * fold-const.c (maskable_range_p): New function.
5252 (build_range_check): Generate bittests if possible.
5254 2017-06-13 Martin Liska <mliska@suse.cz>
5256 * gimple-pretty-print.c (dump_probability): Add new argument.
5257 (dump_edge_probability): Dump both probability and count.
5258 (dump_gimple_label): Likewise.
5259 (dump_gimple_bb_header): Likewise.
5261 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5264 * config/avr/avr-devices.c: Fix indentation.
5265 * config/avr/gen-avr-mmcu-specs.c: Dito.
5267 2017-06-13 Richard Biener <rguenther@suse.de>
5269 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5270 instead get vector type from stmt_info.
5271 (vectorizable_reduction): Adjust. Remove dead code.
5273 2017-06-13 Richard Biener <rguenther@suse.de>
5276 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5277 case of C * (x * C2 + C3).
5278 (fold_addr_of_array_ref_difference): Properly fold index difference.
5280 2017-06-12 David S. Miller <davem@davemloft.net>
5283 * config/sparc/sparc.md (return expander): Emit frame blockage if
5284 function uses alloca.
5286 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5288 * combine.c (make_field_assignment): Check len rather than the mode
5289 precision when calling force_to_mode.
5291 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5293 Support multilibs and devices that see flash in RAM address range.
5296 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5297 (avr_mcu_t) <flash_pm_offset>: New field.
5298 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5299 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5300 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5301 (AVR_TINY_PM_OFFSET): Remove macro.
5302 * config/avr/avr.opt (-mshort-calls): New option.
5303 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5304 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5305 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5306 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5307 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5308 instead of avr_arch->have_jmp_call.
5309 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5310 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5311 avr_arch->flash_pm_offset to define.
5312 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5313 new field flash_pm_offset. Add entry for avrxmega3.
5314 (avr_texinfo): Add entry for avrxmega3.
5315 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5316 attiny212, attiny214,
5317 attiny412, attiny414, attiny416, attiny417,
5318 attiny814, attiny816, attiny817,
5319 attiny1614, attiny1616, attiny1617,
5320 attiny3214, attiny3216, attiny3217.
5321 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5322 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5323 (avr_print_operand_address) [AVR_TINY]: Same.
5324 (avr_asm_init_sections) <readonly_data_section>: Only patch
5325 callback if avr_arch->flash_pm_offset = 0.
5326 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5327 for rodata if avr_arch->flash_pm_offset != 0.
5328 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5329 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5330 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5331 (m_options): Append opt_rcall.
5332 (m_dirnames): Append dir_rcall.
5333 * config/avr/t-multilib: Regenerate.
5335 * configure.ac [target=avr]: Check whether avrxmega3 default
5336 linker description file works as needed.
5337 * configure: Regenerate.
5338 * doc/avr-mmcu.texi: Regenerate.
5339 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5340 <__AVR_ARCH__>: Document avrxmega3 and 103.
5341 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5342 <__AVR_SHORT_CALLS__>: Document it.
5343 <__AVR_PM_BASE_ADDRESS__>: Document it.
5344 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5345 (AVR Variable Attributes) <progmem>: Document this is
5346 not needed for avrxmega3.
5347 (AVR Named Address Spaces) <__flash>: Dito.
5349 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5351 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5353 2017-06-12 Doug Rupp <rupp@adacore.com>
5355 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5356 Append vxworks-stdint.h to the tm_file list.
5357 * config/vxworks-stdint.h: New file.
5359 2017-06-12 Martin Liska <mliska@suse.cz>
5361 PR tree-optimization/81041
5362 * tree-profile.c (gimple_gen_ic_func_profiler):
5363 Create an extra BB in profile-generate
5364 (gimple_gen_time_profiler): Likewise.
5366 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5368 PR tree-optimization/81003
5369 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5370 (update_range_test): Use it instead of force_gimple_operand_gsi.
5372 2017-06-12 Richard Biener <rguenther@suse.de>
5374 PR tree-optimization/81053
5375 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5376 with backedge value not defined in loop. Simplify def stmt
5379 2017-06-11 Tom de Vries <tom@codesourcery.com>
5382 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5384 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5385 nvptx_cannot_force_const_mem.
5387 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5389 * opts.c (finish_options): Move test for flag_split_stack after
5390 it has been initialized.
5392 2017-06-11 Jason Merrill <jason@redhat.com>
5394 * tree.h (id_equal): New.
5395 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5396 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5397 instead of strcmp of IDENTIFIER_POINTER.
5399 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5401 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5402 (mark_all_inlined_calls_cdtor): Fix formating.
5403 (inline_transform): Rescale profile before inlining.
5405 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5407 * cgraph.h (cgraph_edge::clone): Update prototype.
5408 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5409 (cgraph_node::create_clone): Update.
5410 (cgraph_node::create_version_clone): Update.
5411 * tree-inline.c (copy_bb): Update.
5412 (expand_call_inline): Update.
5414 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5416 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5417 factored out from ...
5418 (rs6000_emit_prologue): ... here.
5420 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5422 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5423 factored out from ...
5424 (rs6000_emit_prologue): ... here.
5426 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5428 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5430 (handle_missing_profiles): Fix computation of tp_first_run.
5431 (counts_to_freqs): Do not touch freqs when count is 0.
5433 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5435 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5438 2017-06-10 Tom de Vries <tom@codesourcery.com>
5440 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5441 attributes): Document signal effective target.
5443 2017-06-10 Tom de Vries <tom@codesourcery.com>
5445 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5446 Document effective target stack_size.
5448 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5450 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5451 to the edit_context if they can be auto-applied.
5453 2017-06-9 Ian Lance Taylor <iant@golang.org>
5455 * opts.c (finish_options): If -fsplit-stack, disable implicit
5456 -forder-blocks-and-partition.
5457 * doc/invoke.texi (Optimize Options): Document that when using
5458 -fsplit-stack -forder-blocks-and-partition is not implicitly
5461 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5463 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5464 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5465 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5466 * builtins.def (abort, trap, unreachable): Declare cold.
5467 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5468 * tree-core.h (ECF_COLD): New.
5469 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5470 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5472 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5474 * predict.c (unlikely_executed_stmt_p): Cleanup.
5476 2017-06-09 Richard Biener <rguenther@suse.de>
5478 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5479 model if the ref is always written to.
5481 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5483 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5485 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5487 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5490 2017-06-09 Tom de Vries <tom@codesourcery.com>
5493 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5494 "target cannot support label values" when encountering LABEL_REF.
5496 2017-06-09 Martin Liska <mliska@suse.cz>
5498 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5499 (gimple_gen_ic_func_profiler): Emit direct comparison
5500 of __gcov_indirect_call_callee with NULL.
5501 (gimple_gen_time_profiler): Change probability from
5502 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5504 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5506 * profile.c (edge_gcov_counts): Turn to pointer.
5507 (compute_branch_probabilities, compute_branch_probabilities): Update.
5508 (branch_prob): Do not clear edge_gcov_count.
5509 * profile.h (edge_gcov_counts): Turn to pointer.
5510 (edge_gcov_count): Update.
5512 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5514 * gimple.h (gimple_check_failed): Mark cold.
5516 2017-06-09 Richard Biener <rguenther@suse.de>
5518 PR tree-optimization/66623
5519 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5520 refactor check_reduction into two parts, properly computing
5521 whether we have to check reduction validity for outer loop
5524 2017-06-09 Richard Biener <rguenther@suse.de>
5526 PR tree-optimization/79483
5527 * graphite-scop-detection.c (order): New global.
5528 (get_order): Compute bb to order mapping that satisfies code
5529 generation constraints.
5530 (cmp_pbbs): New helper.
5531 (build_scops): Start domwalk at entry block, sort generated
5534 2017-06-09 Richard Biener <rguenther@suse.de>
5537 * ipa-polymorphic-call.c
5538 (ipa_polymorphic_call_context::restrict_to_inner_class):
5539 Skip FIELD_DECLs with error_mark_node type.
5540 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5543 2017-06-09 Martin Liska <mliska@suse.cz>
5545 * predict.c (struct branch_predictor): New struct.
5546 (test_prediction_value_range): New test.
5547 (predict_c_tests): New function.
5548 * selftest-run-tests.c (selftest::run_tests): Run the function.
5549 * selftest.h: Declare new tests.
5551 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5554 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5555 gen_add3_insn did not fail.
5556 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5557 r0, construct that number in a temporary reg and add that reg to r0.
5558 If asked to put the result in r0 as well, fail.
5560 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5562 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5563 for early expansion of vec_eqv.
5565 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5568 * ubsan.c (instrument_null): Avoid pointless code temporary.
5569 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5571 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5575 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5578 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5581 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5583 * predict.c (unlikely_executed_edge_p): Move ahead.
5584 (probably_never_executed_edge_p): Use it.
5586 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5589 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5590 gimple_call_builtin_p call.
5592 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5594 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5595 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5596 rtl_check_failed_type2, rtl_check_failed_code1,
5597 rtl_check_failed_code2, rtl_check_failed_code_mode,
5598 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5599 rtvec_check_failed_bounds, rtl_check_failed_flag,
5600 _fatal_insn_not_found, _fatal_insn): Likewise.
5601 * tree.h (tree_contains_struct_check_failed,
5602 tree_check_failed, tree_not_check_failed,
5603 tree_class_check_failed, tree_range_check_failed,
5604 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5605 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5606 tree_operand_check_failed, omp_clause_check_failed,
5607 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5610 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5612 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5613 flag_branch_probabilities.
5614 * ipa-inline.c (edge_badness): Likewise.
5615 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5616 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5617 * predict.c (maybe_hot_frequency_p): Likewise.
5618 (probably_never_executed): Likewise.
5619 * sched-ebb.c (schedule_ebbs): Likewise.
5620 * sched-rgn.c (find_single_block_region): Likewise.
5621 * tracer.c (tail_duplicate): Likewise.
5623 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5625 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5626 longer requires x_flag_profile_use.
5628 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5630 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5631 instead of flag_reorder_blocks_and_partition.
5632 * dbxout.c (dbxout_function_end): Likewise.
5633 * dwarf2out.c (gen_subprogram_die): Likewise.
5634 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5635 * hw-doloop.c (reorg_loops): Likewise.
5636 * varasm.c (assemble_start_function,
5637 assemble_end_function): Likewise.
5638 (decide_function_section): Do not check for
5639 flag_reorder_blocks_and_partition.
5641 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5643 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5645 (chkp_get_hard_register_fake_addr_expr): Ditto.
5646 (chkp_build_addr_expr): Add check for hard reg case.
5647 (chkp_parse_array_and_component_ref): Ditto.
5648 (chkp_find_bounds_1): Ditto.
5649 (chkp_process_stmt): Don't generate bounds store for
5652 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5654 * predict.c (maybe_hot_bb_p): Do not check profile status.
5655 (maybe_hot_edge_p): Likewise.
5656 (probably_never_executed): Check for zero counts even if profile
5658 (unlikely_executed_edge_p): New function.
5659 (unlikely_executed_stmt_p): New function.
5660 (unlikely_executed_bb_p): New function.
5661 (set_even_probabilities): Use unlikely predicates.
5662 (combine_predictions_for_bb): Likewise.
5663 (predict_paths_for_bb): Likewise.
5664 (predict_paths_leading_to_edge): Likewise.
5665 (determine_unlikely_bbs): New function.
5666 (estimate_bb_frequencies): Use it.
5667 (compute_function_frequency): Use zero counts even if profile is
5669 * profile-count.h: Fix typo.
5671 2017-08-08 Julia Koval <julia.koval@intel.com>
5673 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5674 _mm512_mask_cvtsepi16_storeu_epi8,
5675 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5676 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5677 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5678 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5679 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5680 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5681 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5682 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5683 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5684 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5685 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5686 __builtin_ia32_pmovuswb256mem_mask,
5687 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5688 __builtin_ia32_pmovwb512mem_mask): New builtins.
5690 2017-08-08 Julia Koval <julia.koval@intel.com>
5692 PR target/73350,80862
5693 * config/i386/subst.md (round): Fix round pattern.
5694 * config/i386/i386.c (ix86_erase_embedded_rounding):
5695 Fix erasing rounding for the fixed pattern.
5697 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5699 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5701 2017-06-08 Martin Liska <mliska@suse.cz>
5703 PR gcov-profile/80911
5704 * gcov.c (block_info::block_info): New constructor.
5706 2017-06-07 Carl Love <cel@us.ibm.com>
5708 * config/rs6000/rs6000-c: The return type of the following
5709 built-in functions was implemented as int not long long. Fix sign
5710 of return value for the unsigned version of vec_mulo and vec_mule.
5711 vector unsigned long long vec_bperm (vector unsigned long long,
5712 vector unsigned char)
5713 vector signed long long vec_mule (vector signed int,
5715 vector unsigned long long vec_mule (vector unsigned int,
5716 vector unsigned int)
5717 vector signed long long vec_mulo (vector signed int,
5719 vector unsigned long long vec_mulo (vector unsigned int,
5720 vector unsigned int)
5721 * doc/extend.texi: Fix the documentation for the built-in
5724 2017-06-07 Carl Love <cel@us.ibm.com>
5727 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5730 2017-06-07 Carl Love <cel@us.ibm.com>
5732 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5733 support, Generate doublehv for signed int/float for BE case only.
5735 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5737 * doc/invoke.texi (mcx16): Rewrite.
5739 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5741 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5742 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5743 *mov<mode>_softfloat, and an anonymous splitter): Use
5744 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5746 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5748 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
5750 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5751 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
5752 (rs6000_debug_reg_global): Adjust.
5753 (rs6000_init_hard_regno_mode_ok): Adjust.
5754 (rs6000_dbx_register_number): Adjust.
5755 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
5756 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5757 Remove SPE_ACC and SPEFSCR.
5758 (REG_ALLOC_ORDER): Ditto.
5759 (FRAME_POINTER_REGNUM): Change to 111.
5760 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
5761 (REG_CLASS_NAMES): Ditto.
5762 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
5763 (REGISTER_NAMES): Ditto.
5764 (ADDITIONAL_REG_NAMES): Ditto.
5765 (rs6000_reg_names): Ditto.
5766 * config/rs6000/rs6000.md: Renumber some register number
5769 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5771 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
5773 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5774 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
5776 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
5777 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5778 Delete the SPE high registers.
5779 (REG_ALLOC_ORDER): Ditto.
5780 (enum reg_class): Remove SPE_HIGH_REGS.
5781 (REG_CLASS_NAMES): Ditto.
5782 (REG_CLASS_CONTENTS): Delete the SPE high registers.
5783 (REGISTER_NAMES): Ditto.
5784 (rs6000_reg_names): Ditto.
5785 * doc/tm.texi.in: Remove SPE as example.
5786 * doc/tm.texi: Regenerate.
5788 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5790 * config/rs6000/8540.md (ppc8540_brinc): Delete.
5791 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
5792 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
5793 * config/rs6000/rs6000.md (type): Remove "brinc".
5795 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5797 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
5798 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
5799 * config/rs6000/linuxspe.h: Delete file.
5800 * config/rs6000/rs6000.md: Don't include spe.md.
5801 * config/rs6000/spe.h: Delete file.
5802 * config/rs6000/spe.md: Delete file.
5803 * config/rs6000/t-rs6000: Remove spe.md.
5805 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5807 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
5808 (reg_or_none500mem_operand): Delete.
5809 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
5810 instead of reg_or_none500mem_operand.
5812 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5814 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5815 handling of SPE flags.
5816 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
5818 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5820 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
5822 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
5823 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
5824 paired_divv2sf3): Similar.
5825 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
5826 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
5827 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
5829 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
5830 Rename the paired_* instruction patterns.
5831 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
5833 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
5834 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
5835 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
5836 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
5837 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
5839 (struct machine_function): Delete field spe_insn_chain_scanned_p.
5840 (spe_func_has_64bit_regs_p): Delete.
5841 (spe_expand_predicate_builtin): Delete.
5842 (spe_expand_evsel_builtin): Delete.
5843 (TARGET_DWARF_REGISTER_SPAN): Do not define.
5844 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
5845 (invalid_e500_subreg): Delete.
5846 (rs6000_legitimize_address): Always force_reg op2 as well, for
5847 paired single memory accesses.
5848 (rs6000_member_type_forces_blk): Delete.
5849 (rs6000_spe_function_arg): Delete.
5850 (rs6000_expand_unop_builtin): Delete SPE handling.
5851 (rs6000_expand_binop_builtin): Ditto.
5852 (spe_expand_stv_builtin): Delete.
5853 (bdesc_2arg_spe): Delete.
5854 (spe_expand_builtin): Delete.
5855 (spe_expand_predicate_builtin): Delete.
5856 (spe_expand_evsel_builtin): Delete.
5857 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
5858 (spe_init_builtins): Delete.
5859 (spe_func_has_64bit_regs_p): Delete.
5860 (savres_routine_name): Delete "info" parameter. Adjust callers.
5861 (rs6000_emit_stack_reset): Ditto.
5862 (rs6000_dwarf_register_span): Delete.
5863 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
5864 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
5865 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
5867 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
5869 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
5870 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
5871 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
5872 mulv2sf3, divv2sf3): Delete expanders.
5874 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5876 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
5878 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5880 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
5881 * config/rs6000/rs6000.c: Ditto.
5883 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5885 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
5886 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
5887 comparison_operator.
5889 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5891 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
5892 * config/rs6000/rs6000.opt: Ditto.
5893 * config/rs6000/t-rtems: Ditto.
5895 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5897 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
5898 TARGET_E500_SINGLE by 0, simplify.
5899 * config/rs6000/rs6000.c: Ditto.
5900 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
5901 (spe_build_register_parallel): Delete.
5902 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
5903 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
5904 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
5905 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
5906 (E500_CONVERT): Delete.
5907 * config/rs6000/spe.md: Remove many patterns and all define_constants.
5909 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5911 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
5912 * config/rs6000/dfp.md: Ditto.
5913 (negdd2, *negdd2_fpr): Merge.
5914 (absdd2, *absdd2_fpr): Merge.
5915 (negtd2, *negtd2_fpr): Merge.
5916 (abstd2, *abstd2_fpr): Merge.
5917 * config/rs6000/e500.h: Delete file.
5918 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
5919 TARGET_FPRS by 1 and simplify.
5920 * config/rs6000/rs6000-c.c: Ditto.
5921 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
5923 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
5925 * config/rs6000/rs6000.md: Ditto.
5926 (floatdidf2, *floatdidf2_fpr): Merge.
5927 (move_from_CR_gt_bit): Delete.
5928 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
5929 (E500_CR_IOR_COMPARE): Delete.
5930 (All patterns that require !TARGET_FPRS): Delete.
5931 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
5933 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5935 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
5937 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5939 * graphds.c (add_edge): Intitialize edge's attached data.
5940 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
5941 pointer parameter. Call pointed function on each edge during
5942 graph traversing. Skip traversing the edge when the function
5944 (graphds_dfs, graphds_scc): Ditto.
5945 (for_each_edge): New parameter. Pass the new parameter to callback
5947 * graphds.h (skip_edge_callback): New function pointer type.
5948 (graphds_dfs, graphds_scc): New function pointer parameter.
5949 (graphds_edge_callback, for_each_edge): New parameter.
5951 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5953 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
5954 out code checking if runtime alias check is possible to below ...
5955 Call the new function.
5956 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
5957 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
5959 2017-06-07 Marek Polacek <polacek@redhat.com>
5962 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
5963 TYPE_OVERFLOW_WRAPS check.
5965 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5967 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
5968 if versioning is required.
5969 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
5970 peeling with the check for versioning.
5972 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5974 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
5975 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
5976 Set true to new parameter if new ssa variable is defined.
5977 (vect_gen_vector_loop_niters): Refactor. Set range information
5978 for the new vector loop bound variable.
5979 (vect_do_peeling): Ditto.
5981 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5983 * tree-affine.c (ssa.h): Include header file.
5984 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
5985 has wrapping overflow behavior.
5987 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5989 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
5991 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5993 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
5994 (tree_to_aff_combination): ... here.
5996 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5998 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
5999 reg_pressure model function.
6000 (ivopts_global_cost_for_size): Delete.
6001 (determine_set_costs, iv_ca_recount_cost): Call new model function
6002 ivopts_estimate_reg_pressure.
6004 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6006 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6007 expensive than udiv. Remove floating point cases from mod.
6009 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6011 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6014 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6016 * config/aarch64/aarch64.md
6017 (copysignsf3): Fix mask generation.
6019 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6021 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6023 (class dump_manager): Add register_dumps method.
6024 * dumpfile.c: Include langhooks.h.
6025 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6026 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6027 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6028 (dump_manager::dump_register): Start with 512 entries instead of 32.
6029 (dump_manager::register_dumps): New method.
6030 * toplev.c (general_init): Instead of invoking register_dumps
6031 langhook, invoke register_dumps method on the dump manager.
6032 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6035 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6037 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6038 Say that # requires an associated define_split to exist, and that
6039 the define_split must be suitable for use after register allocation.
6041 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6043 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6044 (compute_outgoing_frequencies): Also initialize zero counts.
6045 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6046 around loops; preserve more of profile when nothing changes.
6048 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6050 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6052 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6053 * config/arm/arm-cpu-cdata.h: Regenerate.
6054 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6055 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6056 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6057 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6058 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6060 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6062 * config/arm/t-rmprofile: Likewise.
6063 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6065 2017-06-06 David S. Miller <davem@davemloft.net>
6068 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6069 blockage if function uses alloca.
6071 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6073 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6074 New "uid" fields to hold pretty-print IDs of group and ref.
6075 Memory references are now identified as <group_id>:<ref_id>
6076 instead of using [random] addresses.
6077 (dump_mem_details): Simplify, no functional change.
6078 (dump_mem_ref): Simplify and make output more concise.
6079 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6080 (find_or_create_group): Initialize group uid.
6081 (record_ref): Initialize ref uid. Improve debug output.
6082 (prune_group_by_reuse, should_issue_prefetch_p,)
6083 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6084 (mark_nontemporal_store, determine_loop_nest_reuse):
6085 Improve debug output.
6087 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6089 * dbgcnt.def (prefetch): New debug counter.
6090 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6091 (schedule_prefetches): Stop issueing prefetches if debug counter
6094 2017-06-06 Tom de Vries <tom@codesourcery.com>
6096 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6097 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6099 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6101 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6102 Use aarch64_reg_or_zero predicate for operand 4.
6103 (aarch64_compare_and_swap<mode> define_insn_and_split):
6104 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6105 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6107 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6109 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6110 (arm_compute_save_core_reg_mask): This.
6111 (thumb1_compute_save_reg_mask): Rename into ...
6112 (thumb1_compute_save_core_reg_mask): This.
6113 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6114 (arm_compute_frame_layout): Likewise.
6116 2017-06-06 Richard Biener <rguenther@suse.de>
6118 PR tree-optimization/80974
6119 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6120 keep or clear leaders SSA info.
6122 2017-06-06 Tom de Vries <tom@codesourcery.com>
6124 * config/nvptx/nvptx.c (split_mode_p): New function.
6125 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6127 2017-06-06 Tom de Vries <tom@codesourcery.com>
6129 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6131 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6134 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6137 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6139 * shrink-wrap.c (handle_simple_exit): Update profile.
6140 (try_shrink_wrapping): Upate profile.
6142 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6144 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6145 (tree_guess_outgoing_edge_probabilities): New.
6146 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6147 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6149 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6151 * ipa-split.c (split_function): Initialize return bb profile.
6153 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6155 * profile.c (compute_branch_probabilities): Also initialize
6158 2017-06-06 Richard Biener <rguenther@suse.de>
6160 PR tree-optimization/80928
6161 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6162 (vect_analyze_loop_operations): Properly guard analysis for
6164 (vect_transform_loop): Likewise.
6165 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6166 (vect_model_induction_cost): Do not cost for pure SLP.
6167 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6168 of induction in inner loop vectorization.
6169 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6170 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6171 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6173 (vect_analyze_slp_cost_1): Cost induction.
6174 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6175 (vect_get_slp_vect_defs): Likewise.
6176 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6177 (vect_transform_stmt): Handle SLP reductions.
6178 * tree-vectorizer.h (vectorizable_induction): Adjust.
6180 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6182 * config/rs6000/rs6000.c (make_resolver_func): Update
6183 init_lowered_empty_function call.
6185 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6187 * doc/invoke.texi: Document the -fprofile-abs-path option.
6188 * common.opt (fprofile-abs-path): New option.
6189 * gcov-io.h (gcov_write_filename): Declare.
6190 * gcov-io.c (gcov_write_filename): New function.
6191 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6192 * profile.c (output_location): Likewise.
6194 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6196 * shring-wrap.c: Revert accidental commit.
6198 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6200 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6202 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6204 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6206 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6207 profile in callgraph edge.
6208 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6209 (apply_scale): Likewise.
6210 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6211 Also scale profile when inlining function with zero profile.
6212 (initialize_cfun): Update exit block profile even when it is zero.
6213 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6214 when profile is read.
6216 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6218 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6219 (CLONE_*): New constants to define the processors we can generate
6220 code for with the target_clone attribute.
6221 (rs6000_clone_map): New array to identify which clone processors
6222 the current program is running on.
6223 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6224 target_clone attribute.
6225 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6226 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6227 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6228 (cpu_expand_builtin): Add support for target_clone attribute.
6229 (rs6000_valid_attribute_p): Allow "default" attribute.
6230 (get_decl_name): New debug function to simplify printing the
6231 current function name in debugging statements.
6232 (rs6000_clone_priority): New functions to support the target_clone
6233 attribute, and be able to generate code to switch between ISA 2.05
6234 through ISA 3.0 (power6 through power9).
6235 (rs6000_compare_version_priority): Likewise.
6236 (rs6000_get_function_versions_dispatcher): Likewise.
6237 (make_resolver_func): Likewise.
6238 (add_condition_to_bb): Likewise.
6239 (dispatch_function_versions): Likewise.
6240 (rs6000_generate_version_dispatcher_body): Likewise.
6241 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6242 (fusion_gpr_load_p): Fix a spacing issue.
6243 * doc/extend.texi (Common Function Attributes): Document that the
6244 PowerPC supports the target_clone attribute.
6246 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6248 * config/arm/arm.h: explain F symbol found in description of ARM
6249 register allocation in its legend.
6251 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6253 * config/mips/frame-header-opt.c: Include profile-count.h.
6254 * config/riscv/riscv.c: Include profile-count.h
6256 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6258 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6260 (sm_set_flag_if_changed): Add bbs field.
6261 (execute_sm_if_changed_flag_set): Pass BBS.
6262 (execute_sm): Update.
6264 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6266 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6269 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6271 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6272 (peephole2): New peephole2 to emit the above.
6273 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6275 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6277 * config/aarch64/aarch64.c (define_peephole2 above
6278 *sub_<shift>_<mode>): New peephole.
6280 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6282 * config/i386/i386.c (make_resolver_func): Update.
6283 * Makefile.in: Add profile-count.h and profile-count.o
6284 * auto-profile.c (afdo_indirect_call): Update to new API.
6285 (afdo_set_bb_count): Update.
6286 (afdo_propagate_edge): Update.
6287 (afdo_propagate_circuit): Update.
6288 (afdo_calculate_branch_prob): Update.
6289 (afdo_annotate_cfg): Update.
6290 * basic-block.h: Include profile-count.h
6291 (struct edge_def): Turn count to profile_count.
6292 (struct basic_block_def): Likewie.
6293 (REG_BR_PROB_BASE): Move to profile-count.h
6294 (RDIV): Move to profile-count.h
6295 * bb-reorder.c (max_entry_count): Turn to profile_count.
6296 (find_traces): Update.
6297 (rotate_loop):Update.
6298 (connect_traces):Update.
6299 (sanitize_hot_paths):Update.
6300 * bt-load.c (migrate_btr_defs): Update.
6301 * cfg.c (RDIV): Remove.
6302 (init_flow): Use alloc_block.
6303 (alloc_block): Uninitialize count.
6304 (unchecked_make_edge): Uninitialize count.
6305 (check_bb_profile): Update.
6306 (dump_edge_info): Update.
6307 (dump_bb_info): Update.
6308 (update_bb_profile_for_threading): Update.
6309 (scale_bbs_frequencies_int): Update.
6310 (scale_bbs_frequencies_gcov_type): Update.
6311 (scale_bbs_frequencies_profile_count): New.
6312 * cfg.h (update_bb_profile_for_threading): Update.
6313 (scale_bbs_frequencies_profile_count): Declare.
6314 * cfgbuild.c (compute_outgoing_frequencies): Update.
6315 (find_many_sub_basic_blocks): Update.
6316 * cfgcleanup.c (try_forward_edges): Update.
6317 (try_crossjump_to_edge): Update.
6318 * cfgexpand.c (expand_gimple_tailcall): Update.
6319 (construct_exit_block): Update.
6320 * cfghooks.c (verify_flow_info): Update.
6321 (dump_bb_for_graph): Update.
6322 (split_edge): Update.
6323 (make_forwarder_block): Update.
6324 (duplicate_block): Update.
6325 (account_profile_record): Update.
6326 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6327 (get_estimated_loop_iterations): Update.
6328 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6329 (single_likely_exit): Update.
6330 * cfgloopmanip.c (scale_loop_profile): Update.
6332 (set_zero_probability): Update.
6333 (lv_adjust_loop_entry_edge): Update.
6334 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6335 (purge_dead_edges): Update.
6336 (rtl_account_profile_record): Update.
6337 * cgraph.c (cgraph_node::create): Uninitialize count.
6338 (symbol_table::create_edge): Uninitialize count.
6339 (cgraph_update_edges_for_call_stmt_node): Update.
6340 (cgraph_edge::dump_edge_flags): Update.
6341 (cgraph_node::dump): Update.
6342 (cgraph_edge::maybe_hot_p): Update.
6343 * cgraph.h: Include profile-count.h
6344 (create_clone), create_edge, create_indirect_edge): Update.
6345 (cgraph_node): Turn count to profile_count.
6346 (cgraph_edge0: Likewise.
6347 (make_speculative, clone): Update.
6348 (create_edge): Update.
6349 (init_lowered_empty_function): Update.
6350 * cgraphclones.c (cgraph_edge::clone): Update.
6351 (duplicate_thunk_for_node): Update.
6352 (cgraph_node::create_clone): Update.
6353 * cgraphunit.c (cgraph_node::analyze): Update.
6354 (cgraph_node::expand_thunk): Update.
6355 * final.c (dump_basic_block_info): Update.
6356 * gimple-streamer-in.c (input_bb): Update.
6357 * gimple-streamer-out.c (output_bb): Update.
6358 * graphite.c (print_global_statistics): Update.
6359 (print_graphite_scop_statistics): Update.
6360 * hsa-brig.c: Include basic-block.h.
6361 * hsa-dump.c: Include basic-block.h.
6362 * hsa-gen.c (T sum_slice): Update.
6363 (convert_switch_statements):Update.
6364 * hsa-regalloc.c: Include basic-block.h.
6365 * ipa-chkp.c (chkp_produce_thunks): Update.
6366 * ipa-cp.c (struct caller_statistics): Update.
6367 (init_caller_stats): Update.
6368 (gather_caller_stats): Update.
6369 (ipcp_cloning_candidate_p): Update.
6370 (good_cloning_opportunity_p): Update.
6371 (get_info_about_necessary_edges): Update.
6372 (dump_profile_updates): Update.
6373 (update_profiling_info): Update.
6374 (update_specialized_profile): Update.
6375 (perhaps_add_new_callers): Update.
6376 (decide_about_value): Update.
6377 (ipa_cp_c_finalize): Update.
6378 * ipa-devirt.c (struct odr_type_warn_count): Update.
6379 (struct decl_warn_count): Update.
6380 (struct final_warning_record): Update.
6381 (possible_polymorphic_call_targets): Update.
6382 (ipa_devirt): Update.
6383 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6384 * ipa-icf.c (sem_function::merge): Update.
6385 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6386 * ipa-inline.c (compute_uninlined_call_time): Update.
6387 (compute_inlined_call_time): Update.
6388 (want_inline_small_function_p): Update.
6389 (want_inline_self_recursive_call_p): Update.
6390 (edge_badness): Update.
6391 (lookup_recursive_calls): Update.
6392 (recursive_inlining): Update.
6393 (inline_small_functions): Update.
6394 (dump_overall_stats): Update.
6395 (dump_inline_stats): Update.
6396 * ipa-profile.c (ipa_profile_generate_summary): Update.
6397 (ipa_propagate_frequency): Update.
6398 (ipa_profile): Update.
6399 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6400 * ipa-utils.c (ipa_merge_profiles): Update.
6401 * loop-doloop.c (doloop_modify): Update.
6402 * loop-unroll.c (report_unroll): Update.
6403 (unroll_loop_runtime_iterations): Update.
6404 * lto-cgraph.c (lto_output_edge): Update.
6405 (lto_output_node): Update.
6406 (input_node): Update.
6407 (input_edge): Update.
6408 (merge_profile_summaries): Update.
6409 * lto-streamer-in.c (input_cfg): Update.
6410 * lto-streamer-out.c (output_cfg): Update.
6411 * mcf.c (create_fixup_graph): Update.
6412 (adjust_cfg_counts): Update.
6413 (sum_edge_counts): Update.
6414 * modulo-sched.c (sms_schedule): Update.
6415 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6416 * predict.c (maybe_hot_count_p): Update.
6417 (probably_never_executed): Update.
6418 (dump_prediction): Update.
6419 (combine_predictions_for_bb): Update.
6420 (propagate_freq): Update.
6421 (handle_missing_profiles): Update.
6422 (counts_to_freqs): Update.
6423 (rebuild_frequencies): Update.
6424 (force_edge_cold): Update.
6425 * predict.h: Include profile-count.h
6426 (maybe_hot_count_p, counts_to_freqs): UPdate.
6427 * print-rtl-function.c: Do not include cfg.h
6428 * print-rtl.c: Include basic-block.h
6429 * profile-count.c: New file.
6430 * profile-count.h: New file.
6431 * profile.c (is_edge_inconsistent): Update.
6432 (correct_negative_edge_counts): Update.
6433 (is_inconsistent): Update.
6434 (set_bb_counts): Update.
6435 (read_profile_edge_counts): Update.
6436 (compute_frequency_overlap): Update.
6437 (compute_branch_probabilities): Update; Initialize and deinitialize
6439 (branch_prob): Update.
6440 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6441 (edge_gcov_count): New.
6442 (bb_gcov_count): New.
6443 * shrink-wrap.c (try_shrink_wrapping): Update.
6444 * tracer.c (better_p): Update.
6445 * trans-mem.c (expand_transaction): Update.
6446 (ipa_tm_insert_irr_call): Update.
6447 (ipa_tm_insert_gettmclone_call): Update.
6448 * tree-call-cdce.c: Update.
6449 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6450 (gimple_duplicate_sese_tail): Update.
6451 (gimple_account_profile_record): Update.
6452 (execute_fixup_cfg): Update.
6453 * tree-inline.c (copy_bb): Update.
6454 (copy_edges_for_bb): Update.
6455 (initialize_cfun): Update.
6456 (freqs_to_counts): Update.
6457 (copy_cfg_body): Update.
6458 (expand_call_inline): Update.
6459 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6460 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6461 (try_unroll_loop_completely): Update.
6462 (try_peel_loop): Update.
6463 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6464 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6465 * tree-ssa-loop-split.c (connect_loops): Update.
6466 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6467 * tree-ssa-reassoc.c (branch_fixup): Update.
6468 * tree-ssa-tail-merge.c (replace_block_by): Update.
6469 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6470 (compute_path_counts): Update.
6471 (update_profile): Update.
6472 (recompute_probabilities): Update.
6473 (update_joiner_offpath_counts): Update.
6474 (estimated_freqs_path): Update.
6475 (freqs_to_counts_path): Update.
6476 (clear_counts_path): Update.
6477 (ssa_fix_duplicate_block_edges): Update.
6478 (duplicate_thread_path): Update.
6479 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6480 (struct switch_conv_info): Update.
6481 * tree-tailcall.c (decrease_profile): Update.
6482 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6483 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6484 * value-prof.c (check_counter): Update.
6485 (gimple_divmod_fixed_value): Update.
6486 (gimple_mod_pow2): Update.
6487 (gimple_mod_subtract): Update.
6488 (gimple_ic_transform): Update.
6489 (gimple_stringop_fixed_value): Update.
6490 * value-prof.h (gimple_ic): Update.
6492 2017-06-02 Carl Love <cel@us.ibm.com>
6494 * config/rs6000/rs6000-c: Add support for built-in functions
6495 vector double vec_doublee (vector signed int);
6496 vector double vec_doublee (vector unsigned int);
6497 vector double vec_doublee (vector float);
6498 vector double vec_doubleh (vector signed int);
6499 vector double vec_doubleh (vector unsigned int);
6500 vector double vec_doubleh (vector float);
6501 vector double vec_doublel (vector signed int);
6502 vector double vec_doublel (vector unsigned int);
6503 vector double vec_doublel (vector float);
6504 vector double vec_doubleo (vector signed int);
6505 vector double vec_doubleo (vector unsigned int);
6506 vector double vec_doubleo (vector float);.
6507 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6508 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6510 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6511 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6512 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6514 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6515 vec_doublel, vec_doubleh.
6516 * doc/extend.texi: Update the built-in documentation file for the
6517 new built-in functions.
6519 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6522 * ipa-inline-analysis.c (free_growth_caches): Set
6523 edge_removal_hook_holder to NULL after removing it.
6525 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6527 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6528 comparision with zero.
6530 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6531 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6532 for early expansion of vec_min and vec_max builtins.
6533 (builtin_function_type): Add min/max unsigned variants to those
6534 identified as having unsigned arguments.
6536 2017-06-02 Olivier Hainque <hainque@adacore.com>
6538 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6540 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6542 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6543 Use VALL_F16 iterator rather than VALL.
6545 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6547 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6548 Emit CBNZ inside loop when doing a strong exchange and comparing
6549 against zero. Generate the CC flags after the loop.
6551 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6553 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6554 (dl_section_ref): New.
6555 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6556 On AIX, append an expression to subtract the size of the
6557 section length to dl_section_ref.
6559 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6561 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6562 for early expansion of vector absolute builtins.
6564 2017-06-02 Richard Biener <rguenther@suse.de>
6566 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6567 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6569 2017-06-02 Richard Biener <rguenther@suse.de>
6571 PR tree-optimization/80948
6572 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6575 2017-06-02 Richard Biener <rguenther@suse.de>
6577 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6579 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6580 for inductions as relevant.
6582 2017-06-02 Richard Biener <rguenther@suse.de>
6584 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6585 (vectorizable_induction): ... this. Remove dead code.
6587 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6589 * builtins. (expand_builtin_alloca): Remove second parameter and
6590 infer its value from the first parameter instead.
6591 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6593 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6595 PR rtl-optimization/80903
6596 * loop-doloop.c (add_test): Unshare sequence.
6598 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6600 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6602 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6604 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6606 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6607 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6608 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6609 (xlogue_layout::get_stub_rtx): Make static.
6610 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6611 (xlogue_layout::compute_stub_managed_regs): Rename to...
6612 (xlogue_layout::count_stub_managed_regs): ...this.
6613 (xlogue_layout::is_stub_managed_reg): New function.
6614 (xlogue_layout::m_stub_names): Rename to...
6615 (xlogue_layout::s_stub_names): ...this, make static.
6616 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6617 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6618 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6619 xlogue_layout::s_stub_names): Instantiate statics.
6620 (stub_managed_regs): Remove.
6621 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6622 (disable_call_ms2sysv_xlogues): Rename to...
6623 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6624 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6626 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6627 change after reload_completed.
6628 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6630 (ix86_expand_prologue): Likewise.
6631 (ix86_expand_epilogue): Likewise.
6632 (ix86_expand_split_stack_prologue): Likewise.
6633 (ix86_compute_frame_layout): Remove frame parameter ...
6634 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6635 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6637 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6638 (ix86_frame): Move from here ...
6639 * config/i386/i386.h (ix86_frame): ... to here.
6640 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6641 complete ix86_frame data structure instead. Remove some_ld_name.
6643 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6645 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6646 symbols that hold a DECL_VALUE_EXPR.
6648 2017-06-01 Martin Jambor <mjambor@suse.cz>
6650 PR tree-optimization/80898
6651 * tree-sra.c (process_subtree_disqualification): Removed.
6652 (disqualify_candidate): Do not acll
6653 process_subtree_disqualification.
6654 (subtree_mark_written_and_enqueue): New function.
6655 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6656 RHS has been disqualified and re-queue LHS if necessary. Apart
6657 from that, ignore disqualified RHS.
6659 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6661 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6662 address fetch for z10 or later.
6664 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6666 * config/arc/arc.md (tst_movb): Add guard when splitting.
6668 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6670 * config/arc/arc.c (arc_can_eliminate): Test against
6671 arc_frame_pointer_needed.
6673 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6675 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6676 to prevent store reordering.
6677 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6678 (type): Add block type.
6679 (stack_tie): Define special instruction to be used in
6682 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6684 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6685 constraint. It is not valid for the pattern.
6686 (noncommutative_binary_comparison): Likewise.
6688 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6690 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6693 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6695 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6696 be used by the reg-alloc.
6698 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6700 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6701 reg-alloc when having mul64 or mul32x16 instructions.
6702 (mulsidi3): Likewise.
6703 (umulsidi3): Likewise.
6704 (mulsi32x16): New pattern.
6705 (mulsi64): Likewise.
6706 (mulsidi64): Likewise.
6707 (umulsidi64): Likewise.
6708 (MUL32x16_REG): Define.
6709 (mul64_600): Use MUL32x16_REG.
6710 (mac64_600): Likewise.
6711 (umul64_600): Likewise.
6712 (umac64_600): Likewise.
6714 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6716 * config/arc/arc.md (mulsi3_700): Make it commutative.
6718 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6720 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6722 (*sign_extendsidi2_insn): Likewise for movstosw.
6724 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6726 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6727 the type of the input discriminant value. Convert the
6728 discriminant value of signedness vary.
6730 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6732 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6733 Add to -Wall section.
6735 2017-06-01 Richard Biener <rguenther@suse.de>
6738 * fold-const.c (fold_plusminus_mult_expr): If the factored
6739 factor may be zero use a wrapping type for the inner operation.
6740 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6741 and handle moved defs.
6742 (process_assignment): Properly guard the unary op case. Return a
6743 tri-state indicating that moving the stmt before the call may allow
6744 to continue. Pass through to_move.
6745 (find_tail_calls): Handle moving unrelated defs before
6748 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6751 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
6752 splitter result in the canonical way.
6754 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
6756 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
6757 also for 32bit target. Update insn attributes.
6758 (zero-extendsidi2 splitter): Allow all registers for operand 1.
6760 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
6762 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
6763 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
6764 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
6765 (_mm_maskz_min_ss): New intrinsics.
6767 2017-05-31 Martin Liska <mliska@suse.cz>
6769 * tree-vect-loop.c (vect_create_epilog_for_reduction):
6770 Change comment style to one we normally use.
6771 (vectorizable_reduction): Likewise.
6772 (vectorizable_induction): Likewise.
6773 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6774 (vectorizable_call): Likewise.
6775 (vectorizable_simd_clone_call): Likewise.
6776 (vectorizable_conversion): Likewise.
6777 (vectorizable_assignment): Likewise.
6778 (vectorizable_shift): Likewise.
6779 (vectorizable_operation): Likewise.
6780 (vectorizable_store): Likewise.
6781 (vectorizable_load): Likewise.
6782 * tree-vectorizer.h: Likewise.
6784 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
6786 * passes.c (emergency_dump_function): New.
6787 * tree-pass.h (emergency_dump_function): Declare.
6788 * plugin.c (plugins_internal_error_function): Remove.
6789 * plugin.h (plugins_internal_error_function): Remove declaration.
6790 * toplev.c (internal_error_function): New static function. Use it...
6791 (general_init): ...here.
6793 2017-05-31 Graham Markall <graham.markall@embecosm.com>
6795 * config/arc/arc.c (arc_print_operand): Handle constant operands.
6796 (arc_rtx_costs): Add costs for new patterns.
6797 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
6798 * config/arc/predicates.md: Add _1_2_3_operand predicate.
6800 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6802 * tree-ssa-strlen.c (get_next_strinfo): New function.
6803 (get_stridx_plus_constant): Use it.
6804 (zero_length_string): Likewise.
6805 (adjust_related_strinfos): Likewise.
6806 (adjust_last_stmt): Likewise.
6808 2017-05-31 Richard Biener <rguenther@suse.de>
6811 * config/i386/i386.c (ix86_expand_builtin): Remove assert
6812 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
6814 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6816 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
6817 loop_vinfo argument and use of dependence distance vectors.
6818 Check instead whether the two references differ only in their
6819 initial value and assume that they have the same alignment if the
6820 difference is a multiple of the vector alignment.
6821 (vect_analyze_data_refs_alignment): Update call accordingly.
6823 2017-05-31 Martin Liska <mliska@suse.cz>
6826 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
6828 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6830 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
6831 (create_intersect_range_checks): Move from ...
6832 * tree-data-ref.c (create_intersect_range_checks_index)
6833 (create_intersect_range_checks): ... to here.
6834 (create_runtime_alias_checks): New function factored from ...
6835 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
6836 here. Call above function.
6837 * tree-data-ref.h (create_runtime_alias_checks): New function.
6839 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6841 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
6842 segment length for dr_b and compute it in wide_int.
6844 2017-05-31 Richard Biener <rguenther@suse.de>
6846 PR tree-optimization/80906
6847 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
6848 and pass through iv_map.
6849 (copy_bb_and_scalar_dependences): Adjust.
6850 (translate_pending_phi_nodes): Likewise.
6851 (copy_loop_close_phi_args): Handle code-generating IVs instead
6854 2017-05-30 David Malcolm <dmalcolm@redhat.com>
6856 * diagnostic-color.c (color_dict): Add "type-diff".
6857 (parse_gcc_colors): Update comment.
6858 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6859 -fdiagnostics-show-template-tree and -fno-elide-type.
6860 (GCC_COLORS): Add type-diff to example.
6862 (-fdiagnostics-show-template-tree): New.
6863 (-fno-elide-type): New.
6864 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
6865 the pp_format_decoder callback. Call any m_format_postprocessor's
6867 (pretty_printer::pretty_printer): Initialize
6868 m_format_postprocessor.
6869 (pretty_printer::~pretty_printer): Delete any
6870 m_format_postprocessor.
6871 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
6872 (class format_postprocessor): New class.
6873 (struct pretty_printer::format_decoder): Document the new parameters.
6874 (struct pretty_printer::m_format_postprocessor): New field.
6875 * tree-diagnostic.c (default_tree_printer): Update for new
6876 bool and const char ** params.
6877 * tree-diagnostic.h (default_tree_printer): Likewise.
6879 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
6881 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
6882 (lwa_operand): Delete rs6000_gen_cell_microcode test.
6883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6884 rs6000_gen_cell_microcode code.
6885 (rs6000_final_prescan_insn): Delete.
6886 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
6887 "warn-cell-microcode" entries.
6888 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
6889 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
6890 throughout. Change cc_reg_not_micro_cr0_operand to
6891 cc_reg_not_cr0_operand throughout.
6892 (*extendhi<mode>2_noload): Delete.
6893 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
6894 (mwarn-cell-microcode): Delete.
6895 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
6896 -mgen-cell-microcode and -mwarn-cell-microcode.
6898 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
6901 * config/i386/constraints.md (Yd): New constraint.
6903 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
6904 and (?Yd, r) alternatives. Update insn attributes.
6905 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
6906 and (?*Yd, r) alternatives. Update insn attributes.
6907 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
6909 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
6911 * gimplify.c (gimplify_modify_expr): Don't create a
6912 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
6915 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6917 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
6919 2017-05-30 Richard Biener <rguenther@suse.de>
6921 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
6922 and reduc_def fields.
6923 (STMT_VINFO_REDUC_TYPE): New define.
6924 (STMT_VINFO_REDUC_DEF): Likewise.
6925 (vect_force_simple_reduction): Adjust prototype.
6926 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
6927 (vect_is_simple_reduction): Remove check_reduction argument.
6928 (vect_force_simple_reduction): Adjust and set
6929 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6930 (vectorizable_reduction): Do not re-do reduction analysis
6931 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6932 * tree-parloops.c (gather_scalar_reductions): Adjust.
6934 2017-05-30 Richard Biener <rguenther@suse.de>
6937 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
6940 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
6942 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
6943 Introduce unknown_misalignment parameter and remove vf.
6944 (vect_peeling_hash_get_lowest_cost):
6945 Pass unknown_misalignment parameter.
6946 (vect_enhance_data_refs_alignment):
6947 Fix unsupportable data ref treatment.
6949 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6951 * tree-vect-data-refs.c (vect_get_data_access_cost):
6952 Workaround for SLP handling.
6953 (vect_enhance_data_refs_alignment):
6954 Compute costs for doing no peeling at all, compare to the best
6955 peeling costs so far and avoid peeling if cheaper.
6957 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6959 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
6960 Return peeling info and set costs to zero for unlimited cost
6962 (vect_enhance_data_refs_alignment): Also inspect all datarefs
6963 with unknown misalignment. Compute and costs for unknown
6964 misalignment, compare them to the costs for known misalignment
6965 and choose the cheapest for peeling.
6967 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6969 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
6970 (vect_get_peeling_costs_all_drs): Create function.
6971 (vect_peeling_hash_get_lowest_cost):
6972 Use vect_get_peeling_costs_all_drs.
6973 (vect_peeling_supportable): Create function.
6974 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
6976 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6978 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
6979 DR_HAS_NEGATIVE_STEP.
6980 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
6981 (vect_enhance_data_refs_alignment): Use.
6982 (vect_duplicate_ssa_name_ptr_info): Use.
6983 * tree-vectorizer.h (dr_misalignment): Use.
6984 (known_alignment_for_access_p): Use.
6986 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6989 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
6991 (has_section_name): New function.
6993 2017-05-30 Martin Liska <mliska@suse.cz>
6996 * auto-profile.c (get_function_decl_from_block): Fix
6999 2017-05-30 Richard Biener <rguenther@suse.de>
7002 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7004 2017-05-30 Martin Liska <mliska@suse.cz>
7006 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7007 * dumpfile.h (struct dump_file_info): Remove ctors.
7009 2017-05-30 Martin Liska <mliska@suse.cz>
7011 * predict.def: Fix GNU coding style.
7013 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7015 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7016 Mark 'to' argument with ATTRIBUTE_UNUSED.
7018 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7020 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7021 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7022 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7025 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7027 * doc/install.texi (Options specification): Restore entry of
7028 --enable-sjlj-exceptions.
7030 2017-05-27 Michael Eager <eager@eagercon.com>
7033 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7035 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7037 * config/microblaze/microblaze.h
7038 (FIXED_REGISTERS): Update in macro.
7039 (CALL_USED_REGISTERS): Update in macro.
7041 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7043 * doc/install.texi: Add links to macOS binary distributions.
7045 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7049 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7051 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7053 2017-05-26 Martin Liska <mliska@suse.cz>
7055 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7057 2017-05-26 Martin Liska <mliska@suse.cz>
7059 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7060 always leading ';; '.
7061 (dump_bb_info): Likewise.
7062 (brief_dump_cfg): Likewise.
7063 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7064 * dumpfile.c: Remove usage of TDF_VERBOSE.
7065 * dumpfile.h (enum dump_kind): Likewise.
7066 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7067 * print-tree.c (debug_verbose): Remove.
7068 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7069 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7070 * tree-diagnostic.c (default_tree_printer): Replace
7071 TDF_DIAGNOSTIC with TDF_SLIM.
7073 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7075 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7076 in parameter loop, rather than loop_vinfo.
7077 (create_intersect_range_checks): Ditto.
7078 (vect_create_cond_for_alias_checks): Update call to above functions.
7080 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7082 PR tree-optimization/80815
7083 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7084 for merging runtime alias checks. Handle negative DR_STEPs.
7086 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7088 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7090 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7091 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7092 out code pruning runtime alias checks.
7093 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7094 factored out from above.
7095 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7097 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7099 (prune_runtime_alias_test_list): New decalaration.
7101 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7103 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7104 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7105 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7106 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7107 (operator==, comp_dr_with_seg_len_pair): Ditto.
7108 (vect_prune_runtime_alias_test_list): Ditto.
7110 2017-05-26 Martin Liska <mliska@suse.cz>
7113 * params.def: Bound partial-inlining-entry-probability param.
7115 2017-05-26 Marek Polacek <polacek@redhat.com>
7118 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7121 2017-05-26 Richard Biener <rguenther@suse.de>
7123 PR tree-optimization/80842
7124 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7127 2017-05-26 Richard Biener <rguenther@suse.de>
7129 PR tree-optimization/80844
7130 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7132 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7134 * doc/md.texi (Machine Constraints): Update x86 family
7135 machine constraints section to match 'config/i386/constraints.md'.
7137 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7139 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7141 2017-05-25 Nathan Sidwell <nathan@acm.org>
7143 * doc/invoke.texi (--enable-languages): Update documentation.
7145 2017-05-25 Martin Liska <mliska@suse.cz>
7147 * dumpfile.c: Add TDF_FOLDING.
7148 * dumpfile.h (enum dump_kind): Likewise.
7149 * genmatch.c (dt_simplify::gen_1): Use it.
7151 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7153 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7155 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7157 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7158 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7160 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7162 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7163 * match.pd (X == C): Rewrite it here.
7164 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7165 with_certain_nonzero_bits2): New predicates.
7166 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7168 2017-05-24 Nathan Sidwell <nathan@acm.org>
7170 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7173 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7176 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7178 * config/powerpcspe: New port. Files are copied from the rs6000
7179 port, with "rs6000" in filenames replaced by "powerpcspe".
7181 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7183 PR rtl-optimization/80754
7184 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7186 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7188 * config/sparc/sparc.md (length): Return the correct value for -mflat
7189 sibcalls to match output_sibcall.
7191 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7195 * config/rs6000/rs6000.c (struct machine_function): Add new field
7197 (rs6000_get_separate_components): Init that field, use it.
7198 (rs6000_components_for_bb): Use the field.
7200 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7202 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7204 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7207 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7209 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7212 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7213 address against address_operand predicate.
7214 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7216 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7218 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7219 parameters passed indirectly.
7221 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7223 * config/i386/i386.md (*movdi_internal): Remove SSE4
7224 alternative 18 (?r, *v). Update insn attributes.
7225 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7226 Update insn attributes.
7227 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7228 Update insn attributes.
7229 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7230 alternative 1 (r, v). Remove isa attribute.
7231 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7232 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7233 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7235 2017-05-23 Tom de Vries <tom@codesourcery.com>
7237 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7240 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7242 * cgraphunit.c (symbol_table::process_new_functions): Update.
7243 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7244 (inline_generate_summary): Rename to ...
7245 (ipa_fn_summary_generate): ... this one.
7246 (inline_read_summary): Rename to ...
7247 (ipa_fn_summary_read): ... this one.
7248 (inline_write_summary): Rename to ...
7249 (ipa_fn_summary_write): ... this one.
7250 (inline_free_summary): Rename to ...
7251 (ipa_free_fn_summary): ... this one.
7252 (pass_data_local_fn_summary, pass_local_fn_summary,
7253 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7254 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7255 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7256 make_pass_ipa_fn_summary): New.
7257 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7258 inline_write_summary, inline_free_summary): Remove.
7259 (ipa_free_fn_summary) : New.
7260 * ipa-inline.c (ipa_inline): Update.
7261 (pass_ipa_inline): Do not generate summaries.
7262 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7264 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7265 and add pass_ipa_fn_summary.
7266 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7268 (make_pass_inline_parameters): Remove.
7270 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7272 * omp-low.c (struct omp_context): Remove "default_kind" member.
7275 * omp-offload.c (execute_oacc_device_lower): Remove the
7276 parallelism dimensions function attributes for unparallelized
7277 OpenACC kernels constructs.
7279 2017-05-23 Martin Liska <mliska@suse.cz>
7281 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7283 (cgraph_edge::make_speculative): Likewise.
7284 (cgraph_edge::resolve_speculation): Likewise.
7285 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7286 (cgraph_node::dump): Likewise.
7287 * cgraph.h: Likewise.
7288 * cgraphunit.c (analyze_functions): Likewise.
7289 (symbol_table::compile): Likewise.
7290 * ipa-cp.c (print_all_lattices): Likewise.
7291 (determine_versionability): Likewise.
7292 (initialize_node_lattices): Likewise.
7293 (ipcp_verify_propagated_values): Likewise.
7294 (estimate_local_effects): Likewise.
7295 (update_profiling_info): Likewise.
7296 (create_specialized_node): Likewise.
7297 (perhaps_add_new_callers): Likewise.
7298 (decide_about_value): Likewise.
7299 (decide_whether_version_node): Likewise.
7300 (identify_dead_nodes): Likewise.
7301 (ipcp_store_bits_results): Likewise.
7302 * ipa-devirt.c (dump_targets): Likewise.
7303 (ipa_devirt): Likewise.
7304 * ipa-icf.c (sem_item::dump): Likewise.
7305 (sem_function::equals): Likewise.
7306 (sem_variable::equals): Likewise.
7307 (sem_item_optimizer::read_section): Likewise.
7308 (sem_item_optimizer::execute): Likewise.
7309 (congruence_class::dump): Likewise.
7310 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7311 (dump_inline_summary): Likewise.
7312 (estimate_node_size_and_time): Likewise.
7313 (inline_analyze_function): Likewise.
7314 * ipa-inline-transform.c (inline_call): Likewise.
7315 * ipa-inline.c (report_inline_failed_reason): Likewise.
7316 (want_early_inline_function_p): Likewise.
7317 (edge_badness): Likewise.
7318 (update_edge_key): Likewise.
7319 (inline_small_functions): Likewise.
7320 * ipa-profile.c (ipa_profile): Likewise.
7321 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7322 (ipa_make_edge_direct_to_target): Likewise.
7323 (remove_described_reference): Likewise.
7324 (ipa_impossible_devirt_target): Likewise.
7325 (propagate_controlled_uses): Likewise.
7326 (ipa_print_node_params): Likewise.
7327 (ipcp_transform_function): Likewise.
7328 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7329 (propagate_pure_const): Likewise.
7330 * ipa-reference.c (generate_summary): Likewise.
7331 (read_write_all_from_decl): Likewise.
7332 (propagate): Likewise.
7333 (ipa_reference_read_optimization_summary): Likewise.
7334 * ipa-utils.c (ipa_merge_profiles): Likewise.
7335 * ipa.c (walk_polymorphic_call_targets): Likewise.
7336 (symbol_table::remove_unreachable_nodes): Likewise.
7337 (ipa_single_use): Likewise.
7338 * passes.c (execute_todo): Likewise.
7339 * predict.c (drop_profile): Likewise.
7340 * symtab.c (symtab_node::get_dump_name): New function.
7341 (symtab_node::dump_name): Likewise.
7342 (symtab_node::dump_asm_name): Likewise.
7343 (symtab_node::dump_references): Likewise.
7344 (symtab_node::dump_referring): Likewise.
7345 (symtab_node::dump_base): Likewise.
7346 (symtab_node::debug_symtab): Likewise.
7347 * tree-sra.c (convert_callers_for_node): Likewise.
7348 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7349 * value-prof.c (init_node_map): Likewise.
7351 2017-05-23 Martin Liska <mliska@suse.cz>
7353 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7354 and symtab_node::debug_symtab to symbol_table::debug.
7355 * cgraphunit.c (analyze_functions): Use the renamed function.
7356 (symbol_table::compile): Likewise.
7357 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7358 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7359 * passes.c (execute_todo): Likewise.
7360 * symtab.c (symbol_table::dump): New function.
7361 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7363 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7365 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7366 that nonconst implies exec.
7368 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7370 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7371 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7372 (inline_edge_summary_vec): Turn into ...
7373 (ipa_call_summaries): ... this one.
7374 (redirect_to_unreachable, edge_set_predicate,
7375 evaluate_properties_for_edge, inline_summary_alloc,
7376 reset_ipa_call_summary, reset_inline_summary,
7377 inline_summary_t::duplicate): Update.
7378 (inline_edge_duplication_hook): Turn to ...
7379 (ipa_call_summary_t::duplicate): ... this one.
7380 (inline_edge_removal_hook): Turn to ...
7381 (ipa_call_summary_t::remove): ... this one.
7382 (dump_inline_edge_summary): Turn to ...
7383 (dump_ipa_call_summary): ... this one.
7384 (estimate_function_body_sizes): Update.
7385 (inline_update_callee_summaries): Update.
7386 (remap_edge_change_prob): Update.
7387 (remap_edge_summaries): Update.
7388 (inline_merge_summary): Update.
7389 (do_estimate_edge_time): Update.
7390 (inline_generate_summary): Update.
7391 (inline_read_section): Update.
7392 (inline_read_summary): Update.
7393 (inline_free_summary): Update.
7394 * ipa-inline.c (can_inline_edge_p): Update.
7395 (compute_inlined_call_time): Update.
7396 (want_inline_small_function_p): Update.
7397 (edge_badness): Update.
7398 (early_inliner): Update.
7399 * ipa-inline.h (inline_edge_summary): Turn to ...
7400 (ipa_call_summary): ... this one.
7401 (ipa_call_summary_t): New class.
7402 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7403 (ipa_call_summaries): New.
7404 (inline_edge_summary): Remove.
7405 (estimate_edge_growth): Update.
7406 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7407 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7408 * ipa-split.c (execute_split_functions): Update.
7409 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7411 2017-05-23 Tom de Vries <tom@codesourcery.com>
7413 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7414 attributes): Document rdrand effective target.
7416 2017-05-23 Tom de Vries <tom@codesourcery.com>
7418 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7419 attributes): Sort alphabetically.
7421 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7423 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7425 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7428 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7429 V2DF/V2DI splat into two separate patterns, one that handles
7430 registers, and the other that only handles memory. Drop support
7431 for splatting from a GPR on ISA 2.07 and then splitting the
7432 splat into direct move and splat.
7433 (vsx_splat_<mode>_reg): Likewise.
7434 (vsx_splat_<mode>_mem): Likewise.
7436 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7438 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7440 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7443 * omp-low.c (finish_taskreg_remap): New function.
7444 (finish_taskreg_scan): If unit size of ctx->record_type
7445 is non-constant, unshare the size expression and replace
7446 decls in it with possible outer var refs.
7449 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7450 GOVD_SHARED rather than GOVD_PRIVATE with it.
7451 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7452 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7455 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7456 as last argument to build_outer_var_ref for pointer bases of array
7459 2017-05-19 Martin Sebor <msebor@redhat.com>
7461 * print-tree.c (print_node): Print DECL_READ_P flag.
7463 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7465 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7466 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7467 * cgraph.c: Likewise.
7468 * cgraphunit.c: Likewise.
7469 * gengtype.c: Likewise.
7470 * ipa-cp.c: Likewise.
7471 * ipa-devirt.c: Likewise.
7472 * ipa-icf.c: Likewise.
7473 * ipa-predicate.c: Likewise.
7474 * ipa-profile.c: Likewise.
7475 * ipa-prop.c: Likewise.
7476 * ipa-split.c: Likewise.
7478 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7479 edge_predicate_pool, dump_inline_hints,
7480 inline_summary::account_size_time, redirect_to_unreachable,
7481 edge_set_predicate, set_hint_predicate,
7482 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7483 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7484 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7485 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7486 ipa_call_summary_t::remove, initialize_growth_caches,
7487 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7488 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7489 mark_modified, unmodified_parm_1, unmodified_parm,
7490 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7491 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7492 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7493 will_be_nonconstant_predicate, record_modified_bb_info,
7494 get_minimal_bb, record_modified, param_change_prob,
7495 phi_result_unknown_predicate, predicate_for_phi_result,
7496 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7497 estimate_function_body_sizes, compute_inline_parameters,
7498 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7499 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7500 inline_update_callee_summaries, remap_edge_change_prob,
7501 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7502 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7503 inline_analyze_function, inline_summary_t::insert,
7504 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7505 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7506 inline_free_summary): Move to ipa-fnsummary.h
7507 (predicate_t): Remove.
7508 * ipa-fnsummary.c: New file.
7509 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7510 (enum inline_hints_vals, inline_hints, agg_position_info,
7511 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7512 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7513 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7514 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7515 inline_read_summary, inline_write_summary, inline_free_summary,
7516 inline_analyze_function, initialize_inline_failed,
7517 inline_merge_summary, inline_update_overall_summary,
7518 compute_inline_parameters): Move to ipa-fnsummary.h
7519 * ipa-fnsummary.h: New file.
7520 * ipa-inline-transform.h: Include ipa-inline.h.
7521 * ipa-inline.c: LIkewise.
7523 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7525 * ipa-inline.c (edge_badness): Use inlined_time instead of
7526 inline_summaries->get.
7528 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7530 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7532 2017-05-22 Nathan Sidwell <nathan@acm.org>
7534 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7535 (fdump-lang): Document 'raw' option.
7536 * dumpfile.h (TDI_tu): Delete.
7537 * dumpfile.c (dump_files): Remove translation-unit.
7538 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7540 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7542 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7543 command option from $(AWK) call.
7544 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7545 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7546 [FORMAT]: Remove handling of variable.
7547 * config/avr/t-multilib: Regenerate.
7549 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7551 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7553 (dump_inline_summary): Do not print self_time.
7554 (estimate_function_body_sizes): Do not set self_time.
7555 (compute_inline_parameters): Likewise.
7556 (inline_read_section, inline_write_summary): Do not stream self_time.
7557 * ipa-inline.h (inline_summary): Drop self_time.
7559 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7561 * ipa-inline-analysis.c (account_size_time): Rename to ...
7562 (inline_summary::account_size_time): ... this one.
7563 (reset_ipa_call_summary): Turn to ...
7564 (ipa_call_summary::reset): ... this one.
7565 (reset_inline_summary): Turn to ...
7566 (inline_summary::reset): ... this one.
7567 (inline_summary_t::remove): Update.
7568 (inline_summary_t::duplicate): Update.
7569 (ipa_call_summary_t::remove): Update.
7570 (dump_inline_summary): Update.
7571 (estimate_function_body_sizes): Update.
7572 (compute_inline_parameters): Update.
7573 (estimate_node_size_and_time): Update.
7574 (inline_merge_summary): Update.
7575 (inline_update_overall_summary): Update.
7576 (inline_read_section): Update.
7577 (inline_write_summary): Update.
7578 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7579 add account_size_time and reset member functions.
7580 (ipa_call_summary): Add reset function.
7581 * ipa-predicate.h (predicate::operator &): Constify.
7583 2017-05-22 Richard Biener <rguenther@suse.de>
7585 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7587 2017-05-19 Jason Merrill <jason@redhat.com>
7589 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7591 2017-05-19 Marek Polacek <polacek@redhat.com>
7594 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7595 TYPE_OVERFLOW_WRAPS checks.
7597 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7599 * tree-core.h (enum omp_clause_default_kind): Add
7600 "OMP_CLAUSE_DEFAULT_PRESENT".
7601 * tree-pretty-print.c (dump_omp_clause): Handle it.
7602 * gimplify.c (enum gimplify_omp_var_data): Add
7603 "GOVD_MAP_FORCE_PRESENT".
7604 (gimplify_adjust_omp_clauses_1): Map it to
7605 "GOMP_MAP_FORCE_PRESENT".
7606 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7608 * gimplify.c (oacc_default_clause): Clarify.
7610 2017-05-19 Nathan Sidwell <nathan@acm.org>
7612 LANG_HOOK_REGISTER_DUMPS
7613 * toplev.c (general_init): Call register dump lang hook.
7614 * doc/invoke.texi: Document -fdump-lang option family.
7615 * dumpfile.c (dump_files): Remove class dump here.
7616 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7617 * dumpfile.h (tree_dump_index): Remove TDI_class.
7618 * langhooks-def.h (lhd_register_dumps): Declare.
7619 (LANG_HOOKS_REGISTER_DUMPS): Define.
7620 (LANG_HOOKS_INITIALIZER): Add it.
7621 * langhooks.c (lhd_register_dumps): Define.
7622 * langhooks.h (struct lang_hooks): Add register_dumps.
7624 2017-05-19 Nathan Sidwell <nathan@acm.org>
7626 * context.h (context::set_passes): New.
7627 * context.c (context::context): Do not create pass manager.
7628 * toplev.c (general_init): Create pass manager here.
7630 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7632 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7633 use this splitter if two add or or instructions would also work for
7634 the constant we want to generate.
7636 2017-05-19 Richard Biener <rguenther@suse.de>
7639 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7640 predicate evaluation.
7642 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7644 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7646 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7649 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7651 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7652 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7653 (inline_edge_summary_vec): Turn into ...
7654 (ipa_call_summaries): ... this one.
7655 (redirect_to_unreachable, edge_set_predicate,
7656 evaluate_properties_for_edge, inline_summary_alloc,
7657 reset_ipa_call_summary, reset_inline_summary,
7658 inline_summary_t::duplicate): Update.
7659 (inline_edge_duplication_hook): Turn to ...
7660 (ipa_call_summary_t::duplicate): ... this one.
7661 (inline_edge_removal_hook): Turn to ...
7662 (ipa_call_summary_t::remove): ... this one.
7663 (dump_inline_edge_summary): Turn to ...
7664 (dump_ipa_call_summary): ... this one.
7665 (estimate_function_body_sizes): Update.
7666 (inline_update_callee_summaries): Update.
7667 (remap_edge_change_prob): Update.
7668 (remap_edge_summaries): Update.
7669 (inline_merge_summary): Update.
7670 (do_estimate_edge_time): Update.
7671 (inline_generate_summary): Update.
7672 (inline_read_section): Update.
7673 (inline_read_summary): Update.
7674 (inline_free_summary): Update.
7675 * ipa-inline.c (can_inline_edge_p): Update.
7676 (compute_inlined_call_time): Update.
7677 (want_inline_small_function_p): Update.
7678 (edge_badness): Update.
7679 (early_inliner): Update.
7680 * ipa-inline.h (inline_edge_summary): Turn to ...
7681 (ipa_call_summary): ... this one.
7682 (ipa_call_summary_t): New class.
7683 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7684 (ipa_call_summaries): New.
7685 (inline_edge_summary): Remove.
7686 (estimate_edge_growth): Update.
7687 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7688 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7689 * ipa-split.c (execute_split_functions): Update.
7690 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7692 2017-05-19 Richard Biener <rguenther@suse.de>
7695 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7697 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7699 * config/rs6000/rs6000.c (struct machine_function): Add field
7700 fpr_is_wrapped_separately.
7701 (rs6000_get_separate_components): Use 64 components. Handle the
7703 (rs6000_components_for_bb): Handle the FPR components.
7704 (rs6000_emit_prologue_components): Handle the FPR components.
7705 (rs6000_emit_epilogue_components): Handle the FPR components.
7706 (rs6000_set_handled_components): Handle the FPR components.
7707 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7708 that are already separately shrink-wrapped.
7709 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7710 that are already separately shrink-wrapped.
7712 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7715 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7718 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7719 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7720 cases where the register allocator uses a move and an offsettable
7721 memory operation to/from a FPR register on ISA 2.06/2.07.
7722 (define_peephole2 for Altivec d-form store): Likewise.
7724 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7727 * config/i386/mmx.md (*mov<mode>_internal): Enable
7728 alternatives 11, 12, 13 and 14 also for 32bit targets.
7729 Remove alternatives 15, 16, 17 and 18.
7730 * config/i386/sse.md (vec_concatv2di): Change
7731 alternative (!x, *y) to (x, ?!*Yn).
7733 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7735 * dumpfile.h (enum dump_kind): Remove stray comma.
7737 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7739 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7740 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7741 predicate::num_conditions
7742 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7743 (CHANGED): turn into predicate::changed.
7744 (agg_position_info): Move to ipa-predicate.h
7745 (add_condition, predicate::add_clause, predicate::operator &=,
7746 predicate::or_with, predicate::evaluate, predicate::probability,
7747 dump_condition, dump_clause, predicate::dump,
7748 predicate::remap_after_duplication, predicate::remap_after_inlining,
7749 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
7750 (evaluate_conditions_for_known_args): Update.
7751 (set_cond_stmt_execution_predicate): Update.
7752 * ipa-inline.h: Include ipa-predicate.h
7753 (condition, inline_param_summary, conditions, agg_position_info,
7754 predicate): Move to ipa-predicate.h
7755 * ipa-predicate.c: New file.
7756 * ipa-predicate.h: New file.
7758 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
7760 * final.c (leaf_function_p): Check we are not in a sequence.
7762 2017-05-18 Martin Liska <mliska@suse.cz>
7764 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
7765 * dumpfile.c (dump_register): Use new enum dump_kind.
7766 (get_dump_file_name): Likewise.
7767 (dump_enable_all): Likewise.
7768 (dump_switch_p_1): Likewise.
7769 (enable_rtl_dump_file): Remove usage of TDF_RTL.
7770 * dumpfile.h (enum dump_kind): New enum type.
7771 (struct dump_file_info): Create constructor and
7772 format fields and comments.
7773 * passes.c (pass_manager::register_one_dump_file):
7775 * statistics.c (statistics_early_init): Likewise.
7776 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
7777 TDF_TREE with TDF_SLIM.
7778 (gather_memory_references_ref): Likewise.
7780 2017-05-18 Martin Liska <mliska@suse.cz>
7782 * vec.h (struct vnull): Use it.
7784 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7786 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
7787 (true_predicate, false_predicate, true_predicate_p,
7788 false_predicate_p): Remove.
7789 (single_cond_predicate, not_inlined_predicate): Turn to member function
7791 (add_condition): Update.
7792 (add_clause): Turn to...
7793 (predicate::add_clause): ... this one; update; allow passing NULL
7795 (and_predicates): Turn to ...
7796 (predicate::operator &=): ... this one.
7797 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
7798 (or_predicates): Turn to ...
7799 (predicate::or_with): ... this one.
7800 (evaluate_predicate): Turn to ...
7801 (predicate::evaluate): ... this one.
7802 (predicate_probability): Turn to ...
7803 (predicate::probability): ... this one.
7804 (dump_condition): Update.
7805 (dump_predicate): Turn to ...
7806 (predicate::dump): ... this one.
7807 (account_size_time): Update.
7808 (edge_set_predicate): Update.
7809 (set_hint_predicate): UPdate.
7810 (evaluate_conditions_for_known_args): Update.
7811 (evaluate_properties_for_edge): Update.
7812 (remap_predicate_after_duplication): Turn to...
7813 (predicate::remap_after_duplication): ... this one.
7814 (remap_hint_predicate_after_duplication): Update.
7815 (inline_summary_t::duplicate): UPdate.
7816 (dump_inline_edge_summary): Update.
7817 (dump_inline_summary): Update.
7818 (set_cond_stmt_execution_predicate): Update.
7819 (set_switch_stmt_execution_predicate): Update.
7820 (compute_bb_predicates): Update.
7821 (will_be_nonconstant_expr_predicate): Update.
7822 (will_be_nonconstant_predicate): Update.
7823 (phi_result_unknown_predicate): Update.
7824 (predicate_for_phi_result): Update.
7825 (array_index_predicate): Update.
7826 (estimate_function_body_sizes): Update.
7827 (estimate_node_size_and_time): Update.
7828 (estimate_ipcp_clone_size_and_time): Update.
7829 (remap_predicate): Rename to ...
7830 (predicate::remap_after_inlining): ... this one.
7831 (remap_hint_predicate): Update.
7832 (inline_merge_summary): Update.
7833 (inline_update_overall_summary): Update.
7834 (estimate_size_after_inlining): Update.
7835 (read_predicate): Rename to ...
7836 (predicate::stream_in): ... this one.
7837 (read_inline_edge_summary): Update.
7838 (write_predicate): Rename to ...
7839 (predicate::stream_out): ... this one.
7840 (write_inline_edge_summary): Update.
7841 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
7842 (clause_t): Turn to uint32_t
7843 (predicate): Turn to class; implement constructor and operators
7845 (size_time_entry): Update.
7846 (inline_summary): Update.
7847 (inline_edge_summary): Update.
7849 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
7851 * fold-const.c (fold_binary_loc): Move transformation...
7852 * match.pd (C - X CMP X): ... here.
7854 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
7856 * config/sparc/sparc.c (sparc_option_override): Set function
7857 alignment for -mcpu=niagara7 to 64 to match the I$ line.
7858 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
7860 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
7862 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
7864 2017-05-18 Marek Polacek <polacek@redhat.com>
7867 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
7868 (pass_ubsan::execute): Call gimple_assign_single_p instead of
7869 gimple_assign_load_p.
7871 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
7874 * real.c (do_compare): Give decimal_do_compare preference over
7875 comparing just the signs.
7877 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
7879 * doc/md.texi (Canonicalization of Instructions): Describe the
7880 canonical form of instructions that inherently set a condition
7883 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
7886 * tree-cfg.c: Move deletion of unreachable case statements to after
7887 the merging of consecutive case labels.
7889 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7891 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
7892 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
7893 restoring of callee-saved registers.
7895 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
7897 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
7898 * config/visium/visium.c (single_set_and_flags): Likewise.
7899 * config/visium/visium.md (Substitutions): Likewise.
7901 2017-05-17 Martin Liska <mliska@suse.cz>
7903 * cfg.c: Introduce dump_flags_t type and
7904 use it instead of int type.
7906 * cfghooks.c: Likewise.
7907 * cfghooks.h (struct cfg_hooks): Likewise.
7908 * cfgrtl.c: Likewise.
7909 * cfgrtl.h: Likewise.
7910 * cgraph.c (cgraph_node::get_body): Likewise.
7911 * coretypes.h: Likewise.
7912 * domwalk.c: Likewise.
7913 * domwalk.h: Likewise.
7914 * dumpfile.c (struct dump_option_value_info): Likewise.
7915 (dump_enable_all): Likewise.
7916 (dump_switch_p_1): Likewise.
7917 (opt_info_switch_p): Likewise.
7918 * dumpfile.h (enum tree_dump_index): Likewise.
7919 (struct dump_file_info): Likewise.
7920 * genemit.c: Likewise.
7921 * generic-match-head.c: Likewise.
7922 * gengtype.c (open_base_files): Likewise.
7923 * gimple-pretty-print.c: Likewise.
7924 * gimple-pretty-print.h: Likewise.
7925 * graph.c (print_graph_cfg): Likewise.
7926 * graphite-scop-detection.c (dot_all_sese): Likewise.
7927 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7928 * loop-unroll.c (report_unroll): Likewise.
7929 * passes.c (pass_manager::register_one_dump_file): Likewise.
7930 * print-tree.c: Likewise.
7931 * statistics.c: Likewise.
7932 * tree-cfg.c: Likewise.
7933 * tree-cfg.h: Likewise.
7934 * tree-dfa.c: Likewise.
7935 * tree-dfa.h: Likewise.
7936 * tree-dump.c (dump_function): Likewise.
7937 * tree-dump.h (struct dump_info): Likewise.
7938 * tree-pretty-print.c: Likewise.
7939 * tree-pretty-print.h: Likewise.
7940 * tree-ssa-live.c: Likewise.
7941 * tree-ssa-live.h: Likewise.
7942 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
7943 * tree-vect-loop.c: Likewise.
7944 * tree-vect-slp.c: Likewise.
7946 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
7947 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7949 PR tree-optimization/80457
7950 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
7951 of all arguments to a statement as scalar_to_vec operations.
7952 (vectorizable_call): Adjust call to vect_model_simple_cost for
7954 (vectorizable_conversion): Likewise.
7955 (vectorizable_assignment): Likewise.
7956 (vectorizable_shift): Likewise.
7957 (vectorizable_operation): Likewise.
7958 (vectorizable_comparison): Likewise.
7959 (vect_is_simple_cond): Record the def types for operands.
7960 (vectorizable_condition): Likewise, call vect_model_simple_cost.
7961 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
7962 for statement argument count.
7964 2017-05-16 Carl Love <cel@us.ibm.com>
7966 * config/rs6000/rs6000-c: Add support for built-in functions
7967 vector unsigned long long vec_bperm (vector unsigned long long,
7968 vector unsigned char)
7969 vector signed long long vec_mule (vector signed int,
7971 vector unsigned long long vec_mule (vector unsigned int,
7972 vector unsigned int)
7973 vector signed long long vec_mulo (vector signed int,
7975 vector unsigned long long vec_mulo (vector unsigned int,
7976 vector unsigned int)
7977 vector signed char vec_sldw (vector signed char,
7980 vector unsigned char vec_sldw (vector unsigned char,
7981 vector unsigned char,
7983 vector signed short vec_sldw (vector signed short,
7984 vector signed short,
7986 vector unsigned short vec_sldw (vector unsigned short,
7987 vector unsigned short,
7989 vector signed int vec_sldw (vector signed int,
7992 vector unsigned int vec_sldw (vector unsigned int,
7993 vector unsigned int,
7995 vector signed long long vec_sldw (vector signed long long,
7996 vector signed long long,
7998 vector unsigned long long vec_sldw (vector unsigned long long,
7999 vector unsigned long long,
8001 * config/rs6000/rs6000-c: Add support for built-in functions
8002 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8003 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8004 * doc/extend.texi: Update the built-in documentation for the
8005 new built-in functions.
8007 2017-05-16 Marek Polacek <polacek@redhat.com>
8011 * tree.c (save_expr): Don't fold the expression.
8013 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8015 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8016 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8017 and (?*y,m). Update insn attributes.
8019 2017-05-16 Martin Liska <mliska@suse.cz>
8021 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8022 flags argument of print_gimple_stmt, print_gimple_expr,
8023 print_generic_stmt and print_generic_expr.
8024 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8025 * coretypes.h: Likewise.
8026 * except.c (dump_eh_tree): Likewise.
8027 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8028 * gimple-pretty-print.h: Likewise.
8029 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8030 (backprop::push_to_worklist): Likewise.
8031 (backprop::pop_from_worklist): Likewise.
8032 (backprop::process_use): Likewise.
8033 (backprop::intersect_uses): Likewise.
8034 (note_replacement): Likewise.
8035 * gimple-ssa-store-merging.c
8036 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8037 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8038 (pass_store_merging::execute): Likewise.
8039 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8040 (ssa_base_cand_dump_callback): Likewise.
8041 (dump_incr_vec): Likewise.
8042 (replace_refs): Likewise.
8043 (replace_mult_candidate): Likewise.
8044 (create_add_on_incoming_edge): Likewise.
8045 (create_phi_basis): Likewise.
8046 (insert_initializers): Likewise.
8047 (all_phi_incrs_profitable): Likewise.
8048 (introduce_cast_before_cand): Likewise.
8049 (replace_one_candidate): Likewise.
8050 * gimplify.c (gimplify_expr): Likewise.
8051 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8052 (set_rename): Likewise.
8053 (rename_uses): Likewise.
8054 (copy_loop_phi_nodes): Likewise.
8055 (add_close_phis_to_merge_points): Likewise.
8056 (copy_loop_close_phi_args): Likewise.
8057 (copy_cond_phi_args): Likewise.
8058 (graphite_copy_stmts_from_block): Likewise.
8059 (translate_pending_phi_nodes): Likewise.
8060 * graphite-poly.c (print_pdr): Likewise.
8061 (dump_gbb_cases): Likewise.
8062 (dump_gbb_conditions): Likewise.
8063 (print_scop_params): Likewise.
8064 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8065 (build_cross_bb_scalars_use): Likewise.
8066 (gather_bbs::before_dom_children): Likewise.
8067 * hsa-dump.c (dump_hsa_immed): Likewise.
8068 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8069 (get_replacement_map): Likewise.
8070 * ipa-inline-analysis.c (dump_condition): Likewise.
8071 (estimate_function_body_sizes): Likewise.
8072 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8073 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8074 * ipa-prop.c (ipa_dump_param): Likewise.
8075 (ipa_print_node_jump_functions_for_edge): Likewise.
8076 (ipa_modify_call_arguments): Likewise.
8077 (ipa_modify_expr): Likewise.
8078 (ipa_dump_param_adjustments): Likewise.
8079 (ipa_dump_agg_replacement_values): Likewise.
8080 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8081 * ipa-pure-const.c (check_stmt): Likewise.
8082 (pass_nothrow::execute): Likewise.
8083 * ipa-split.c (execute_split_functions): Likewise.
8084 * omp-offload.c (dump_oacc_loop_part): Likewise.
8085 (dump_oacc_loop): Likewise.
8086 * trans-mem.c (tm_log_emit): Likewise.
8087 (tm_memopt_accumulate_memops): Likewise.
8088 (dump_tm_memopt_set): Likewise.
8089 (dump_tm_memopt_transform): Likewise.
8090 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8091 (print_loop): Likewise.
8092 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8093 (chkp_gather_checks_info): Likewise.
8094 (chkp_get_check_result): Likewise.
8095 (chkp_remove_check_if_pass): Likewise.
8096 (chkp_use_outer_bounds_if_possible): Likewise.
8097 (chkp_reduce_bounds_lifetime): Likewise.
8098 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8099 (chkp_mark_completed_bounds): Likewise.
8100 (chkp_register_incomplete_bounds): Likewise.
8101 (chkp_mark_invalid_bounds): Likewise.
8102 (chkp_maybe_copy_and_register_bounds): Likewise.
8103 (chkp_build_returned_bound): Likewise.
8104 (chkp_get_bound_for_parm): Likewise.
8105 (chkp_build_bndldx): Likewise.
8106 (chkp_get_bounds_by_definition): Likewise.
8107 (chkp_generate_extern_var_bounds): Likewise.
8108 (chkp_get_bounds_for_decl_addr): Likewise.
8109 * tree-chrec.c (chrec_apply): Likewise.
8110 * tree-data-ref.c (dump_data_reference): Likewise.
8111 (dump_subscript): Likewise.
8112 (dump_data_dependence_relation): Likewise.
8113 (analyze_overlapping_iterations): Likewise.
8114 * tree-inline.c (expand_call_inline): Likewise.
8115 (tree_function_versioning): Likewise.
8116 * tree-into-ssa.c (dump_defs_stack): Likewise.
8117 (dump_currdefs): Likewise.
8118 (dump_names_replaced_by): Likewise.
8119 (dump_update_ssa): Likewise.
8120 (update_ssa): Likewise.
8121 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8122 * tree-parloops.c (build_new_reduction): Likewise.
8123 (try_create_reduction_list): Likewise.
8124 (ref_conflicts_with_region): Likewise.
8125 (oacc_entry_exit_ok_1): Likewise.
8126 (oacc_entry_exit_single_gang): Likewise.
8127 * tree-pretty-print.h: Likewise.
8128 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8129 (get_scalar_evolution): Likewise.
8130 (add_to_evolution): Likewise.
8131 (get_loop_exit_condition): Likewise.
8132 (analyze_evolution_in_loop): Likewise.
8133 (analyze_initial_condition): Likewise.
8134 (analyze_scalar_evolution): Likewise.
8135 (instantiate_scev): Likewise.
8136 (number_of_latch_executions): Likewise.
8137 (gather_chrec_stats): Likewise.
8138 (final_value_replacement_loop): Likewise.
8139 (scev_const_prop): Likewise.
8140 * tree-sra.c (dump_access): Likewise.
8141 (disqualify_candidate): Likewise.
8142 (create_access): Likewise.
8144 (maybe_add_sra_candidate): Likewise.
8145 (create_access_replacement): Likewise.
8146 (analyze_access_subtree): Likewise.
8147 (analyze_all_variable_accesses): Likewise.
8148 (sra_modify_assign): Likewise.
8149 (initialize_constant_pool_replacements): Likewise.
8150 (find_param_candidates): Likewise.
8151 (decide_one_param_reduction): Likewise.
8152 (replace_removed_params_ssa_names): Likewise.
8153 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8154 * tree-ssa-copy.c (dump_copy_of): Likewise.
8155 (copy_prop_visit_cond_stmt): Likewise.
8156 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8157 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8158 (record_equivalences_from_stmt): Likewise.
8159 * tree-ssa-dse.c (compute_trims): Likewise.
8160 (delete_dead_call): Likewise.
8161 (delete_dead_assignment): Likewise.
8162 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8163 (forward_propagate_into_cond): Likewise.
8164 (pass_forwprop::execute): Likewise.
8165 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8166 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8168 (move_computations_worker): Likewise.
8169 (execute_sm): Likewise.
8170 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8171 (remove_exits_and_undefined_stmts): Likewise.
8172 (remove_redundant_iv_tests): Likewise.
8173 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8174 (adjust_iv_update_pos): Likewise.
8175 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8176 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8177 (value_replacement): Likewise.
8178 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8179 * tree-ssa-pre.c (print_pre_expr): Likewise.
8180 (get_representative_for): Likewise.
8181 (create_expression_by_pieces): Likewise.
8182 (insert_into_preds_of_block): Likewise.
8183 (eliminate_insert): Likewise.
8184 (eliminate_dom_walker::before_dom_children): Likewise.
8185 (eliminate): Likewise.
8186 (remove_dead_inserted_code): Likewise.
8187 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8188 * tree-ssa-reassoc.c (get_rank): Likewise.
8189 (eliminate_duplicate_pair): Likewise.
8190 (eliminate_plus_minus_pair): Likewise.
8191 (eliminate_not_pairs): Likewise.
8192 (undistribute_ops_list): Likewise.
8193 (eliminate_redundant_comparison): Likewise.
8194 (update_range_test): Likewise.
8195 (optimize_range_tests_var_bound): Likewise.
8196 (optimize_vec_cond_expr): Likewise.
8197 (rewrite_expr_tree): Likewise.
8198 (rewrite_expr_tree_parallel): Likewise.
8199 (linearize_expr): Likewise.
8200 (break_up_subtract): Likewise.
8201 (linearize_expr_tree): Likewise.
8202 (attempt_builtin_powi): Likewise.
8203 (attempt_builtin_copysign): Likewise.
8204 (transform_stmt_to_copy): Likewise.
8205 (transform_stmt_to_multiply): Likewise.
8206 (dump_ops_vector): Likewise.
8207 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8208 (print_scc): Likewise.
8209 (set_ssa_val_to): Likewise.
8210 (visit_reference_op_store): Likewise.
8211 (visit_use): Likewise.
8212 (sccvn_dom_walker::before_dom_children): Likewise.
8213 (run_scc_vn): Likewise.
8214 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8216 (expr_hash_elt::print): Likewise.
8217 (const_and_copies::pop_to_marker): Likewise.
8218 (const_and_copies::record_const_or_copy_raw): Likewise.
8219 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8220 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8221 (dump_predicates): Likewise.
8222 (find_uninit_use): Likewise.
8223 (warn_uninitialized_phi): Likewise.
8224 (pass_late_warn_uninitialized::execute): Likewise.
8225 * tree-ssa.c (verify_vssa): Likewise.
8226 (verify_ssa): Likewise.
8227 (maybe_optimize_var): Likewise.
8228 * tree-vrp.c (dump_value_range): Likewise.
8229 (dump_all_value_ranges): Likewise.
8230 (dump_asserts_for): Likewise.
8231 (register_edge_assert_for_2): Likewise.
8232 (vrp_visit_cond_stmt): Likewise.
8233 (vrp_visit_switch_stmt): Likewise.
8234 (vrp_visit_stmt): Likewise.
8235 (vrp_visit_phi_node): Likewise.
8236 (simplify_cond_using_ranges_1): Likewise.
8237 (fold_predicate_in): Likewise.
8238 (evrp_dom_walker::before_dom_children): Likewise.
8239 (evrp_dom_walker::push_value_range): Likewise.
8240 (evrp_dom_walker::pop_value_range): Likewise.
8241 (execute_early_vrp): Likewise.
8243 2017-05-16 Richard Biener <rguenther@suse.de>
8245 * dwarf2out.c (loc_list_from_tree_1): Do not create
8246 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8248 2017-05-16 Richard Biener <rguenther@suse.de>
8250 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8252 (note_variable_value_in_expr): If we resolved the decl ref
8253 do not push to the stack.
8255 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8257 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8258 operations in fast-math mode.
8259 (vaddq_f16): Likewise.
8260 (vmul_f16): Likewise.
8261 (vmulq_f16): Likewise.
8262 (vsub_f16): Likewise.
8263 (vsubq_f16): Likewise.
8264 * config/arm/neon.md (add<mode>3): New.
8266 (fma:<VH:mode>3): New. Also remove outdated comment.
8269 2017-05-16 Martin Liska <mliska@suse.cz>
8273 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8274 (odr_types_equivalent_p): Likewise.
8276 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8278 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8280 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8283 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8284 non-interunit SSE move alternatives with '?'.
8285 (zero-extendsidi peephole2): New peephole to skip intermediate
8286 general register in SSE zero-extend sequence.
8288 2017-05-15 Jeff Law <law@redhat.com>
8290 * reorg.c (relax_delay_slots): Create a new variable to hold
8291 the temporary target rather than clobbering TARGET_LABEL.
8293 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8294 missing argument to extract_bit_field call.
8295 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8297 2017-05-15 Martin Liska <mliska@suse.cz>
8300 * gcc.c (process_command): Do not allow empty argument of -o option.
8302 2017-05-15 Renlin Li <renlin.li@arm.com>
8304 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8305 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8306 * config/aarch64/constraints.md (Usf): Add long call check.
8307 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8308 (call_value): Likewise.
8309 (sibcall): Likewise.
8310 (sibcall_value): Likewise.
8312 (call_value_insn): New.
8313 (sibcall_insn): Update rtx pattern.
8314 (sibcall_value_insn): Likewise.
8315 (call_internal): Remove.
8316 (call_value_internal): Likewise.
8317 (sibcall_internal): Likewise.
8318 (sibcall_value_internal): Likewise.
8319 (call_reg): Likewise.
8320 (call_symbol): Likewise.
8321 (call_value_reg): Likewise.
8322 (call_value_symbol): Likewise.
8324 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8327 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8329 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8331 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8332 compatible with CCGOCmode and with CCZmode.
8334 2017-05-14 Martin Sebor <msebor@redhat.com>
8337 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8338 (gimple_fold_builtin_snprintf): Same.
8339 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8340 (gimple_fold_builtin_snprintf): Same.
8341 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8343 (is_call_safe): New function.
8344 (try_substitute_return_value): Call it.
8345 (try_simplify_call): New function.
8346 (pass_sprintf_length::handle_gimple_call): Call it.
8348 2017-05-14 Martin Sebor <msebor@redhat.com>
8351 * builtins.c (expand_builtin_stpncpy): Simplify.
8353 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8355 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8356 * config/i386/i386.h
8357 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8358 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8359 (struct machine_function): Add new members call_ms2sysv,
8360 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8361 (struct machine_frame_state): New fields sp_realigned and
8362 sp_realigned_offset.
8363 * config/i386/i386.c
8364 (enum xlogue_stub): New enum.
8365 (enum xlogue_stub_sets): New enum.
8366 (class xlogue_layout): New class.
8367 (struct ix86_frame): New fields stack_realign_allocate_offset,
8368 stack_realign_offset and outlined_save_offset. Modify comments to
8369 detail stack layout when using out-of-line stubs.
8370 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8371 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8372 -mcall-ms2sysv-xlogues.
8373 (stub_managed_regs): New static variable.
8374 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8375 registers managed by out-of-line stub.
8376 (disable_call_ms2sysv_xlogues): New function.
8377 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8378 m->call_ms2sysv when appropriate and compute frame layout for
8380 (sp_valid_at, fp_valid_at): New inline functions.
8381 (choose_basereg): New function.
8382 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8384 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8385 Use align parameter of choose_baseaddr to generated aligned SSE movs
8387 (pro_epilogue_adjust_stack): Modify to track
8388 machine_frame_state::sp_realigned.
8389 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8390 (ix86_nsaved_sseregs): Likewise.
8391 (ix86_emit_save_regs): Likewise.
8392 (ix86_emit_save_regs_using_mov): Likewise.
8393 (ix86_emit_save_sse_regs_using_mov): Likewise.
8394 (get_scratch_register_on_entry): Likewise.
8395 (gen_frame_set): New function.
8396 (gen_frame_load): Likewise.
8397 (gen_frame_store): Likewise.
8398 (emit_outlined_ms2sysv_save): Likewise.
8399 (emit_outlined_ms2sysv_restore): Likewise.
8400 (ix86_expand_prologue): Modify stack re-alignment code and call
8401 emit_outlined_ms2sysv_save when appropriate.
8402 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8403 parameter rtx_insn *insn, which allows the function to be used to only
8405 (ix86_expand_epilogue): Modify validity checks of frame and stack
8406 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8407 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8408 * config/i386/predicates.md
8409 (save_multiple): New predicate.
8410 (restore_multiple): Likewise.
8411 * config/i386/sse.md
8412 (save_multiple<mode>): New pattern.
8413 (save_multiple_realign<mode>): Likewise.
8414 (restore_multiple<mode>): Likewise.
8415 (restore_multiple_and_return<mode>): Likewise.
8416 (restore_multiple_leave_return<mode>): Likewise.
8417 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8419 2017-05-14 Julia Koval <julia.koval@intel.com>
8421 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8422 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8423 (__builtin_ia32_xsetbv): New builtins.
8424 * config/i386/i386.c (ix86_expand_special_args_builtin):
8426 (ix86_expand_builtin): Special expand for new intrinsics.
8427 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8428 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8429 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8431 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8433 * cfganal.c (inverted_post_order_compute): Change argument type
8435 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8436 * df-core.c (rest_of_handle_df_initialize): Adjust.
8437 (rest_of_handle_df_finish): Likewise.
8438 (df_analyze_1): Likewise.
8439 (df_analyze): Likewise.
8440 (loop_inverted_post_order_compute): Change argument to be a vec *.
8441 (df_analyze_loop): Adjust.
8442 (df_get_n_blocks): Likewise.
8443 (df_get_postorder): Likewise.
8444 * df.h (struct df_d): Change field to be a vec.
8445 * lcm.c (compute_laterin): Adjust.
8446 (compute_available): Likewise.
8447 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8448 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8449 * tree-ssa-pre.c (compute_antic): Likewise.
8451 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8453 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8454 (depth_first_search::depth_first_search): Change structure init
8455 function to this constructor.
8456 (depth_first_search::add_bb): Rename function to this member.
8457 (depth_first_search::execute): Likewise.
8458 (flow_dfs_compute_reverse_finish): Adjust.
8460 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8462 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8463 (longest_simple_path): Likewise.
8464 * shrink-wrap.c (spread_components): Likewise.
8465 (disqualify_problematic_components): Likewise.
8466 (emit_common_heads_for_components): Likewise.
8467 (emit_common_tails_for_components): Likewise.
8468 (insert_prologue_epilogue_for_components): Likewise.
8470 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8472 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8475 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8477 * df-core.c (df_set_blocks): Start using auto_bitmap.
8478 (df_compact_blocks): Likewise.
8479 * df-problems.c (df_rd_confluence_n): Likewise.
8480 * df-scan.c (df_insn_rescan_all): Likewise.
8481 (df_process_deferred_rescans): Likewise.
8482 (df_update_entry_block_defs): Likewise.
8483 (df_update_exit_block_uses): Likewise.
8484 (df_entry_block_bitmap_verify): Likewise.
8485 (df_exit_block_bitmap_verify): Likewise.
8486 (df_scan_verify): Likewise.
8487 * lra-constraints.c (lra_constraints): Likewise.
8488 (undo_optional_reloads): Likewise.
8489 (lra_undo_inheritance): Likewise.
8490 * lra-remat.c (calculate_gen_cands): Likewise.
8491 (do_remat): Likewise.
8492 * lra-spills.c (assign_spill_hard_regs): Likewise.
8493 (spill_pseudos): Likewise.
8494 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8495 (bitmap_set_subtract_values): Likewise.
8497 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8499 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8500 management with auto_bitmap.
8501 (fix_inter_tick): Likewise.
8502 (fix_recovery_deps): Likewise.
8503 * ira.c (add_store_equivs): Likewise.
8504 (find_moveable_pseudos): Likewise.
8505 (split_live_ranges_for_shrink_wrap): Likewise.
8506 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8507 (rtx_reuse_manager::seen_def_p): Likewise.
8508 (rtx_reuse_manager::set_seen_def): Likewise.
8509 * print-rtl.h (class rtx_reuse_manager): Likewise.
8511 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8513 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8515 (migrate_btr_def): Likewise.
8516 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8517 * df-core.c (loop_post_order_compute): Likewise.
8518 (loop_inverted_post_order_compute): Likewise.
8519 * hsa-common.h: Likewise.
8520 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8521 * init-regs.c (initialize_uninitialized_regs): Likewise.
8522 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8523 (inline_small_functions): Likewise.
8524 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8525 * ira.c (combine_and_move_insns): Likewise.
8526 (build_insn_chain): Likewise.
8527 * loop-invariant.c (find_invariants): Likewise.
8528 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8529 * predict.c (tree_predict_by_opcode): Likewise.
8530 (predict_paths_leading_to): Likewise.
8531 (predict_paths_leading_to_edge): Likewise.
8532 (estimate_loops_at_level): Likewise.
8533 (estimate_loops): Likewise.
8534 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8535 (spread_components): Likewise.
8536 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8537 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8538 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8539 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8540 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8541 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8542 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8543 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8544 (mark_threaded_blocks): Likewise.
8545 (thread_through_all_blocks): Likewise.
8546 * tree-ssa.c (verify_ssa): Likewise.
8547 (execute_update_addresses_taken): Likewise.
8548 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8550 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8552 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8554 (post_order_compute): Likewise.
8555 (inverted_post_order_compute): Likewise.
8556 (pre_and_rev_post_order_compute_fn): Likewise.
8558 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8560 * genrecog.c (int_set::int_set): Explicitly construct our
8561 auto_vec base class.
8562 * vec.h (auto_vec::auto_vec): New constructor.
8564 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8566 * bitmap.h (class auto_bitmap): New constructor taking
8567 bitmap_obstack * argument.
8569 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8571 * bitmap.h (class auto_bitmap): Change type of m_bits to
8572 bitmap_head, and adjust ctor / dtor and member operators.
8574 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8576 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8577 when returned register mode doesn't match original mode.
8579 2017-05-12 Jeff Law <law@redhat.com>
8580 Jakub Jelinek <jakub@redhat.com>
8582 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8583 we look for cc setter after the compare-elim changes.
8584 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8585 within the vector to match what compare-elim now expects.
8586 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8587 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8589 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8590 after the compare-elim changes.
8591 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8592 the vector to match what compare-elim now expects.
8593 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8594 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8595 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8596 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8597 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8599 * config/visium/visium.c (single_set_and_flags): Fix where
8600 we look for cc setter after the compare-elim changes.
8601 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8602 with the vector to match what compare-elim now expects.
8603 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8604 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8605 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8606 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8607 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8608 (neg<mode>2_insn_set_overflow): Likewise.
8610 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8613 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8614 maybe_expand_insn call, set ops[0].target. If still set after call,
8615 set alt_rtl. Add extra arg to recursive calls.
8616 (extract_bit_field): Add alt_rtl argument. Pass to
8618 * expmed.h (extract_bit_field): Fix prototype.
8619 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8620 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8621 to extract_bit_field_calls.
8622 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8623 Pass alt_rtl to extract_bit_field calls.
8624 * calls.c (store_unaligned_arguments_into_psuedos)
8625 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8626 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8628 * optabs.h (struct expand_operand): Add target bitfield.
8630 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8632 * compare-elim.c (try_eliminate_compare): Canonicalize
8633 operation with embedded compare to
8634 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8635 (set (reg) (operation)].
8637 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8639 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8642 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8643 cost of adding a carry flag for ADC instruction.
8644 [case MINUS]: Ignore the cost of subtracting a carry flag
8645 for SBB instruction.
8647 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8649 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8651 * config/rs6000/bmiintrin.h: New file.
8652 * config/rs6000/bmi2intrin.h: New file.
8653 * config/rs6000/x86intrin.h: New file.
8655 2017-05-12 Jeff Law <law@redhat.com>
8657 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8660 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8663 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8665 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8667 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8668 early expansion of vector divide builtins.
8669 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8670 builtins identified as having unsigned arguments.
8672 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8674 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8675 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8676 expansion of vector logical operations (and, andc, or, xor,
8679 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8681 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8682 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8684 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8686 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8687 early GIMPLE expansion of vector multiplies.
8689 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8691 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8692 TARGET_HAVE_MOVT conditional.
8693 (movt splitter): Likewise.
8695 2017-05-12 Richard Biener <rguenther@suse.de>
8697 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8698 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8699 Fold all stmts not inplace.
8701 2017-05-12 Richard Biener <rguenther@suse.de>
8703 PR tree-optimization/80713
8704 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8705 inserted_exprs bit for not removed stmts.
8707 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8710 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8711 parallelized" attribute for parallelized OpenACC kernels.
8712 * omp-offload.c (execute_oacc_device_lower): Use it.
8714 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8715 Set "oacc kernels" attribute.
8716 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8717 parameter. Adjust all users.
8718 (oacc_fn_attrib_kernels_p): Remove function.
8719 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8720 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8721 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8722 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8723 assert "oacc kernels" attribute is set.
8725 2017-05-11 Carl Love <cel@us.ibm.com>
8727 * config/rs6000/rs6000-c: Add support for built-in functions
8728 vector unsigned char vec_popcnt (vector signed char)
8729 vector unsigned char vec_popcnt (vector unsigned char)
8730 vector unsigned short vec_popcnt (vector signed short)
8731 vector unsigned short vec_popcnt (vector unsigned short)
8732 vector unsigned int vec_popcnt (vector signed int)
8733 vector unsigned int vec_popcnt (vector unsigned int)
8734 vector unsigned long long vec_popcnt (vector signed long long)
8735 vector unsigned long long vec_popcnt (vector unsigned long long)
8736 vector signed long long vec_slo (vector signed long long,
8738 vector signed long long vec_slo (vector signed long long,
8739 vector unsigned char)
8740 vector unsigned long long vec_slo (vector unsigned long long,
8742 vector unsigned long long vec_slo (vector unsigned long long,
8743 vector unsigned char)
8744 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8745 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8746 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8747 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
8748 * doc/extend.texi: Update the built-in documentation file for the
8749 new built-in functions.
8751 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8753 * attribs.h (sorted_attr_string): Move machine independent
8754 functions for target clone support from the i386 port to common
8755 code. Rename ix86_function_versions to common_function_versions.
8756 Rename make_name to make_unique_name.
8757 (common_function_versions): Likewise.
8758 (make_unique_name): Likewise.
8759 (make_dispatcher_decl): Likewise.
8760 (is_function_default_version): Likewise.
8761 * attribs.c (attr_strcmp): Likewise.
8762 (sorted_attr_string): Likewise.
8763 (common_function_versions): Likewise.
8764 (make_unique_name): Likewise.
8765 (make_dispatcher_decl): Likewise.
8766 (is_function_default_version): Likewise.
8767 * config/i386/i386.c (attr_strcmp): Likewise.
8768 (sorted_attr_string): Likewise.
8769 (ix86_function_versions): Likewise.
8770 (make_name): Likewise.
8771 (make_dispatcher_decl): Likewise.
8772 (is_function_default_version): Likewise.
8773 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
8775 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8778 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8779 Account for direct move costs for vec_construct of integer
8782 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
8785 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
8786 (UNSPEC_STX_ATOMIC): Ditto.
8787 (loaddi_via_sse): New insn.
8788 (storedi_via_sse): Ditto.
8789 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
8790 Update corresponding peephole2 patterns.
8791 (atomic_storedi_fpu): Ditto.
8793 2017-05-11 Julia Koval <julia.koval@intel.com>
8795 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
8796 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
8798 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
8799 (__builtin_ia32_rsqrt14ss_mask): New builtins.
8800 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
8802 2017-05-11 Nathan Sidwell <nathan@acm.org>
8804 * graphite-poly.c: Include dumpfile.h.
8806 * dumpfle.h (dump_function): Declare here ...
8807 * tree-dump.h (dump_function): ... not here.
8808 * dumpfile.c: #include tree-cfg.h.
8809 (dump_function): Move here from ...
8810 * tree-dump.c (dump_function): ... here.
8811 * gimplify.c: #include splay-tree.h, not tree-dump.h.
8812 * graphite-poly.c: Don't include tree-dump.h.
8813 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
8814 * print-tree.c: Likewise.
8815 * stor-layout.c: Likewise.
8816 * tree-nested.c: Likewise.
8818 * dumpfile.c (dump_start): Use TDF_FLAGS.
8819 (dump_enable_all): Fix TDF_KIND check thinko.
8821 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8823 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8824 array entries to represent two legal parameterizations of the
8825 overloaded __builtin_cmpb function, as represented by the
8826 P6_OV_BUILTIN_CMPB constant.
8827 (altivec_resolve_overloaded_builtin): Add special case handling
8828 for the __builtin_cmpb function, as represented by the
8829 P6_OV_BUILTIN_CMPB constant.
8830 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
8831 (BU_P6_64BIT_2): New macro.
8832 (BU_P6_OVERLOAD_2): New macro
8833 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
8834 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
8835 (CMPB): Add overload support to represent both 32-bit and 64-bit
8836 compare-bytes function.
8837 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8838 support for TARGET_CMPB.
8839 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
8840 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8841 documentation of the __builtin_cmpb overloaded built-in function.
8843 2017-05-11 Richard Biener <rguenther@suse.de>
8845 PR tree-optimization/80705
8846 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
8847 bases are not vectorizable.
8849 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8851 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
8852 when counting register pressure.
8854 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8856 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
8857 (iv_ca_more_deps): Renamed to ...
8858 (iv_ca_compare_deps): ... this.
8859 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
8861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8863 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
8865 (determine_group_iv_costs): ... here.
8866 (find_inv_vars_cb): Record inv var if it's not recorded before.
8868 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8870 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
8871 (get_shiftadd_cost): Ditto.
8873 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8875 * tree-ssa-address.c: Include header file.
8876 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
8878 (add_to_parts): Refactor.
8879 (addr_to_parts): New parameter. Update use of move_hint_to_base.
8880 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
8883 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8885 PR tree-optimization/53090
8886 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
8888 (extract_cond_operands): Detect condition with IV on both sides
8889 and return COMP_IV_EXPR_2.
8890 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
8891 (rewrite_use_compare): Simplify by removing call to function
8892 extract_cond_operands.
8894 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8896 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
8897 (extract_cond_operands): Detect condition comparing against non-
8898 invariant bound and return appropriate enum value.
8899 (find_interesting_uses_cond): Update use of extract_cond_operands.
8900 Handle its return value accordingly.
8901 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
8903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8905 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
8906 nonlinear iv_use computation in loop invariant sensitive way.
8908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8910 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
8911 (find_iv_candidates): Call relate_compare_use_with_all_cands.
8913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8915 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
8916 (dump_cand): Support iv_cand.inv_exprs.
8917 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
8919 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
8922 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8924 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
8926 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
8927 as local function. Include necessary header files.
8928 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
8930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8932 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
8934 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8936 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
8937 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
8938 RSHIFT_EXPR and BIT_NOT_EXPR.
8940 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8942 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
8943 (adjust_setup_cost): New parameter supporting round up adjustment.
8944 (struct address_cost_data): Delete.
8945 (force_expr_to_var_cost): Don't bound cost with spill_cost.
8946 (split_address_cost, ptr_difference_cost): Delete.
8947 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
8948 (struct ainc_cost_data): New struct.
8949 (get_address_cost_ainc): New function.
8950 (get_address_cost, get_computation_cost): Reimplement.
8951 (determine_group_iv_cost_address): Record inv_expr for all uses of
8953 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
8954 (iv_ca_has_deps): Reimplemented to ...
8955 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
8957 (iv_ca_extend): Call iv_ca_more_deps.
8959 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8961 * tree-ssa-address.c (struct mem_address): Move to header file.
8962 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
8963 * tree-ssa-address.h (struct mem_address): Move from C file.
8964 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
8966 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8968 * tree-affine.h (aff_combination_type): New interface.
8969 (aff_combination_zero_p): Remove static.
8970 (aff_combination_const_p): New interface.
8971 (aff_combination_singleton_var_p): New interfaces.
8973 2017-05-11 Richard Biener <rguenther@suse.de>
8975 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8976 Skip unreachable blocks and destinations.
8977 (eliminate): Move stmt removal and fixup ...
8978 (fini_eliminate): ... here. Skip inserted exprs.
8979 (pass_pre::execute): Move fini_pre after fini_eliminate.
8980 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
8981 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
8982 PRE to get rid of dead code that has invalid SSA form and
8983 split critical edges again.
8985 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8987 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8989 2017-05-11 Richard Biener <rguenther@suse.de>
8991 * passes.c (execute_function_todo): Verify loops if they are
8992 said to be up-to-date.
8993 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
8994 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
8996 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
8999 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9000 handle calling assemble_external ourself.
9003 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9004 modes with zero size. Enhance comment.
9006 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9008 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9009 built-ins for vec_xl and vec_xst with short and char pointer
9012 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9014 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9015 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9016 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9017 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9018 (_mm_maskz_min_round_ss): New intrinsics.
9019 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9020 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9021 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9022 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9023 (__builtin_ia32_minss_mask_round): New builtins.
9024 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9025 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9026 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9028 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9029 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9031 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9034 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9036 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9037 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9038 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9039 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9040 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9041 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9042 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9043 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9044 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9045 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9046 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9047 (__builtin_ia32_mulss_mask_round): New builtins.
9048 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9049 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9050 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9052 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9053 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9055 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9058 2017-05-10 Julia Koval <julia.koval@intel.com>
9060 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9061 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9062 (_mm256_setr_m128i): New intrinsics.
9064 2017-05-10 Julia Koval <julia.koval@intel.com>
9066 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9067 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9068 (_mm_maskz_rcp14_ss): New intrinsics.
9069 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9070 (__builtin_ia32_rcp14ss_mask): New builtins.
9071 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9073 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9075 PR tree-optimization/51513
9076 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9077 (assert_unreachable_fallthru_edge_p): Use it.
9078 (group_case_labels_stmt): Likewise.
9079 * tree-cfg.h: Prototype it.
9080 * stmt.c: Include cfghooks.h and tree-cfg.h.
9081 (emit_case_dispatch_table) <gap_label>: New local variable.
9082 Use it to fill dispatch table gaps.
9083 Test for default_label before updating probabilities.
9084 (expand_case) <default_label>: Remove unneeded initialization.
9085 Test for unreachable default case statement and remove its edge.
9086 Set default_label accordingly.
9087 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9089 2017-05-10 Carl Love <cel@us.ibm.com>
9091 * config/rs6000/rs6000-c: Add support for built-in functions
9092 vector signed char vec_neg (vector signed char)
9093 vector signed short int vec_neg (vector short int)
9094 vector signed int vec_neg (vector signed int)
9095 vector signed long long vec_neg (vector signed long long)
9096 vector float vec_neg (vector float)
9097 vector double vec_neg (vector double)
9098 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9100 * config/rs6000/altivec.h: Add define for vec_neg
9101 * doc/extend.texi: Update the built-in documentation for the
9102 new built-in functions.
9104 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9106 PR tree-optimization/77644
9107 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9109 2017-05-10 Nathan Sidwell <nathan@acm.org>
9111 * dumpfile.h (TDI_lang_all): New.
9112 (TDF_KIND): New. Renumber others
9113 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9115 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9117 (get_dump_file_name): Adjust suffix generation.
9118 (dump_enable_all): Use TDF_KIND.
9119 * doc/invoke.texi (-fdump-lang-all): Document.
9121 * dumpfile.h: Tabify.
9123 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9126 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9127 Move member access before delete.
9129 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9131 * tree-inline.c (expand_call_inline): Split block at stmt
9134 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9137 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9138 are now unused after splitting mov{sf,sd}_hardfloat.
9139 (f32_lr2): Likewise.
9141 (f32_lm2): Likewise.
9143 (f32_li2): Likewise.
9146 (f32_sr2): Likewise.
9148 (f32_sm2): Likewise.
9150 (f32_si2): Likewise.
9153 (f32_vsx): Likewise.
9155 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9156 For movsf, order stores so the VSX stores occur before the GPR
9157 store which encourages the register allocator to use a traditional
9158 FPR instead of a GPR. For movsd, order the stores so that the GPR
9159 store comes before the VSX stores to allow the power6 to work.
9160 This is due to the power6 not having a 32-bit integer store
9161 instruction from a FPR.
9162 (movsf_hardfloat): Likewise.
9163 (movsd_hardfloat): Likewise.
9165 2017-05-09 Martin Sebor <msebor@redhat.com>
9167 PR translation/80280
9168 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9171 PR translation/80280
9172 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9173 data member added in r247778.
9174 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9176 2017-05-09 Nathan Sidwell <nathan@acm.org>
9178 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9180 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9183 2017-05-09 Marek Polacek <polacek@redhat.com>
9185 * doc/invoke.texi: Fix typo.
9187 2017-05-09 Richard Biener <rguenther@suse.de>
9189 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9190 (vrp_val_is_min): Likewise.
9191 (set_value_range_to_value): Likewise.
9192 (set_value_range_to_nonnegative): Likewise.
9193 (gimple_assign_nonzero_p): Likewise.
9194 (gimple_stmt_nonzero_p): Likewise.
9195 (vrp_int_const_binop): Likewise. Remove unreachable case.
9196 (adjust_range_with_scev): Adjust comments.
9197 (compare_range_with_value): Likewise.
9198 (extract_range_from_phi_node): Likewise.
9199 (test_for_singularity): Likewise.
9201 2017-05-09 Richard Biener <rguenther@suse.de>
9203 * tree-vrp.c (get_single_symbol): Add assert that we don't
9204 get overflowed constants as invariant part.
9205 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9206 checks. Use wi::cmp instead of recursing for integer constants.
9207 (compare_values): Just ignore whether we assumed undefined
9208 overflow instead of failing the compare.
9209 (extract_range_for_var_from_comparison_expr): Add comment before the
9210 TREE_NO_WARNING sets.
9211 (test_for_singularity): Likewise.
9212 (extract_range_from_comparison): Do not disable optimization
9213 when we assumed undefined overflow.
9214 (extract_range_basic): Remove init of unused var.
9216 2017-05-09 Richard Biener <rguenther@suse.de>
9218 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9219 (extract_range_from_multiplicative_op_1): Adjust.
9220 (extract_range_from_binary_expr_1): Use int_const_binop.
9222 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9225 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9226 rs6000_store_data_bypass_p in seven define_bypass directives and
9227 in several comments.
9228 * config/rs6000/rs6000-protos.h: Add prototype for
9229 rs6000_store_data_bypass_p function.
9230 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9231 function implements slightly different (rs6000-specific) semantics
9232 than store_data_bypass_p, returning false rather than aborting
9233 with assertion error when arguments do not satisfy the
9234 requirements of store data bypass.
9235 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9236 rs6000_store_data_bypass_p.
9238 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9240 * config/xtensa/xtensa-protos.h
9241 (xtensa_initial_elimination_offset): New declaration.
9242 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9243 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9244 macro definition, add case for FRAME_POINTER_REGNUM when
9245 FRAME_GROWS_DOWNWARD.
9246 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9247 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9248 xtensa_initial_elimination_offset.
9250 2017-05-08 Nathan Sidwell <nathan@acm.org>
9252 * doc/invoke.texi: Alphabetize -fdump options.
9254 2017-05-08 Martin Sebor <msebor@redhat.com>
9256 PR translation/80280
9257 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9259 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9261 * target.def (compute_frame_layout): New optional target hook.
9262 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9263 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9264 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9266 * reload1.c (verify_initial_elim_offsets): Likewise.
9267 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9268 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9269 (arm_get_frame_offsets): Split up into this ...
9270 (arm_compute_frame_layout): ... and this function.
9272 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9274 * config/aarch64/constraints.md (Usa): New constraint.
9275 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9277 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9279 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9280 with_multilib_list after it has been checked.
9282 2017-05-08 Richard Biener <rguenther@suse.de>
9284 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9285 (bitmap_set_subtract_values): Likewise.
9287 2017-05-08 Richard Biener <rguenther@suse.de>
9289 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9290 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9292 (gimple_stmt_nonzero_warnv_p): Rename to ...
9293 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9295 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9296 (extract_range_basic): Adjust, do not disable propagation on
9297 strict overflow sensitive simplification.
9298 (vrp_visit_cond_stmt): Likewise.
9300 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9302 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9303 body size unconditionally.
9305 2017-05-07 Jeff Law <law@redhat.com>
9308 2017-05-06 Jeff Law <law@redhat.com>
9309 PR tree-optimization/78496
9310 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9313 PR tree-optimization/78496
9314 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9315 (simplify_stmt_using_ranges): Call it.
9316 (vrp_dom_walker::before_dom_children): Extract equivalences
9317 from an ASSERT_EXPR with an equality comparison against a
9320 2017-05-06 Jeff Law <law@redhat.com>
9322 PR tree-optimization/78496
9323 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9326 PR tree-optimization/78496
9327 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9328 (simplify_stmt_using_ranges): Call it.
9329 (vrp_dom_walker::before_dom_children): Extract equivalences
9330 from an ASSERT_EXPR with an equality comparison against a
9333 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9335 * lra-constraints.c (lra_copy_reg_equiv): New function.
9336 (split_reg): Use it to copy equivalence information from the
9337 original register to the spill register.
9339 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9341 PR rtl-optimization/75964
9342 * simplify-rtx.c (simplify_const_relational_operation): Remove
9343 invalid handling of comparisons of integer ABS.
9345 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9347 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9349 (init_regs): Remove declaration.
9350 (function_arg_advance_32): Initialize error_p as boolean variable.
9352 2017-05-05 Nathan Sidwell <nathan@acm.org>
9354 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9355 lines. Use for (;;).
9357 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9359 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9360 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9361 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9362 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9363 VF=2 that require versioning.
9365 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9367 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9370 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9372 * diagnostic.h (diagnostic_override_option_index): Convert from
9373 macro to inline function.
9375 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9377 * diagnostic.c (last_module_changed_p): New function.
9378 (set_last_module): New function.
9379 (diagnostic_report_current_module): Convert macro usage to
9380 the above functions.
9381 * diagnostic.h (diagnostic_context::last_module): Strengthen
9382 from const line_map * to const line_map_ordinary *.
9383 (diagnostic_last_module_changed): Delete macro.
9384 (diagnostic_set_last_module): Delete macro.
9386 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9388 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9389 with diagnostic_report_diagnostic.
9390 (diagnostic_n_impl_richloc): Likewise.
9391 * diagnostic.h (report_diagnostic): Delete macro.
9392 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9393 with diagnostic_report_diagnostic.
9394 * substring-locations.c (format_warning_va): Likewise.
9396 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9398 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9399 save/restor of format_spec. Move option-printing code to...
9400 (print_option_information): ...this new function, and
9401 reimplement by simply printing to the pretty_printer,
9402 rather than appending to the format string.
9404 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9406 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9407 handling logic into...
9408 (update_effective_level_from_pragmas): ...this new function.
9410 2017-05-04 Andrew Waterman <andrew@sifive.com>
9412 * config/riscv/riscv.opt (mstrict-align): New option.
9413 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9414 (SLOW_UNALIGNED_ACCESS): Define.
9415 (riscv_slow_unaligned_access): Declare.
9416 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9418 (riscv_slow_unaligned_access): New variable.
9419 (rocket_tune_info): Set slow_unaligned_access to true.
9420 (optimize_size_tune_info): Set slow_unaligned_access to false.
9421 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9422 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9423 (riscv_option_override): Set riscv_slow_unaligned_access.
9424 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9426 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9428 * config/riscv/riscv.md: Unify indentation.
9430 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9435 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9437 (extendsi<mode>2): Add support for doing sign extension via
9438 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9439 don't have ISA 3.0 instructions.
9440 (extendsi<mode>2 splitter): Likewise.
9441 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9442 generate the normal insns since SImode can now go in vector
9443 registers. Disallow the special UNSPECs needed for previous
9444 machines to hide SImode being used. Add new insns
9445 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9446 (fix_trunc<mode>si2_stfiwx): Likewise.
9447 (fix_trunc<mode>si2_internal): Likewise.
9448 (fixuns_trunc<mode>si2): Likewise.
9449 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9450 (fctiw<u>z_<mode>_smallint): Likewise.
9451 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9452 of floating point to 32-bit integer from doing a direct move to
9453 the GPR registers to do a store.
9454 (fctiwz_<mode>): Break long line.
9456 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9458 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9459 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9460 (addr_list, addr_offset_valid_p): New.
9461 (split_address_groups): Check offset validity with above function.
9462 (gt-tree-ssa-loop-ivopts.h): Include header file.
9464 2017-05-05 Nathan Sidwell <nathan@acm.org>
9466 * config.gcc (arm*-*-*): Add missing 'fi'.
9468 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9470 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9471 included in -fopt-info does not matter.
9472 * doc/optinfo.texi (-fopt-info): Fix description of default
9473 behavour. Explicitly say order of options included in -fopt-info
9476 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9478 * config.gcc: Allow combinations of aprofile and rmprofile values for
9479 --with-multilib-list.
9480 * config/arm/t-multilib: New file.
9481 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9482 variables. Remove setting of ISA and floating-point ABI in
9483 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9484 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9485 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9486 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9488 * config/arm/t-rmprofile: Likewise except for the matches changes.
9489 * doc/install.texi (--with-multilib-list): Document the combination of
9490 aprofile and rmprofile values and warn about pitfalls in doing that.
9492 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9494 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9495 (movdi_aarch64): Likewise.
9497 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9499 PR tree-optimization/80632
9500 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9502 (build_arrays): Initialize it for virtual phis.
9503 (fix_phi_nodes): Use it for virtual phis.
9505 PR tree-optimization/80558
9506 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9507 [x, y] op z into [x op, y op z] for op & or | if conditions
9510 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9511 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9514 * config/arm/arm.md (use_literal_pool): Remove.
9515 (64-bit immediate split): No longer takes cost into consideration
9516 if arm_disable_literal_pool is enabled.
9517 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9518 used when arm_disable_literal_pool is enabled.
9519 (arm_max_const_double_inline_cost): Remove use of
9520 arm_disable_literal_pool.
9521 (push_minipool_fix): Add assert.
9522 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9523 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9524 (no_literal_pool_sf_immediate): New.
9526 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9528 PR tree-optimization/80613
9529 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9530 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9532 2017-05-05 Richard Biener <rguenther@suse.de>
9534 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9536 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9538 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9539 of this flag from insn conditions due to removal from r247495.
9541 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9543 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9545 (arm_early_store_addr_dep_ptr): Likewise.
9546 * config/arm/aarch-common-protos.h
9547 (arm_early_load_addr_dep_ptr): Add prototype.
9548 (arm_early_store_addr_dep_ptr): Likewise.
9549 * config/arm/cortex-a53.md: Add new bypasses.
9551 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9553 * tree.c (next_type_uid): Change type to unsigned.
9554 (type_hash_canon): Decrement back next_type_uid if
9555 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9556 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9559 2017-05-04 Martin Sebor <msebor@redhat.com>
9561 * builtins.c: Fix a trivial typo in a comment.
9564 * builtins.c (check_sizes): Adjust to handle reading past the end.
9565 Avoid printing excessive upper bound of ranges. Use %E to print
9566 tree nodes instead of converting them to %wu.
9567 (expand_builtin_memchr): New function.
9568 (compute_dest_size): Rename...
9569 (compute_objsize): ...to this.
9570 (expand_builtin_memcpy): Adjust.
9571 (expand_builtin_mempcpy): Adjust.
9572 (expand_builtin_strcat): Adjust.
9573 (expand_builtin_strcpy): Adjust.
9574 (check_strncat_sizes): Adjust.
9575 (expand_builtin_strncat): Adjust.
9576 (expand_builtin_strncpy): Adjust and simplify.
9577 (expand_builtin_memset): Adjust.
9578 (expand_builtin_bzero): Adjust.
9579 (expand_builtin_memcmp): Adjust.
9580 (expand_builtin): Handle memcmp.
9581 (maybe_emit_chk_warning): Check strncat just once.
9583 2017-05-04 Martin Sebor <msebor@redhat.com>
9585 PR preprocessor/79214
9588 * builtins.c (check_sizes): Add inlining context and issue
9589 warnings even when -Wno-system-headers is set.
9590 (check_strncat_sizes): Same.
9591 (expand_builtin_strncat): Same.
9592 (expand_builtin_memmove): New function.
9593 (expand_builtin_stpncpy): Same.
9594 (expand_builtin): Handle memmove and stpncpy.
9596 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9598 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9599 which is not used any more.
9601 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9603 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9605 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9607 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9608 (cortexa53_tunings): Likewise.
9609 (cortexa57_tunings): Likewise.
9610 (cortexa72_tunings): Likewise.
9611 (cortexa73_tunings): Likewise.
9613 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9615 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9616 Set loop alignment to 8.
9618 2017-05-04 Martin Sebor <msebor@redhat.com>
9620 PR translation/80280
9621 * builtins.c (expand_builtin_object_size): Add missing quoting to
9622 %D and like directives.
9623 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9624 (hsa_type_for_tree_type): Same.
9625 (verify_function_arguments): Same.
9626 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9627 * varasm.c (get_section): Same.
9630 2017-05-04 Martin Sebor <msebor@redhat.com>
9632 PR translation/80280
9633 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9635 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9637 * config/aarch64/aarch64.c (generic_addrcost_table):
9638 Change HI/TI mode setting.
9640 2017-05-04 Martin Jambor <mjambor@suse.cz>
9642 PR tree-optimization/80622
9643 * tree-sra.c (comes_initialized_p): New function.
9644 (build_accesses_from_assign): Only set write lazily when
9645 comes_initialized_p is false.
9646 (analyze_access_subtree): Use comes_initialized_p.
9647 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9648 instead of testing for PARM_DECL.
9650 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9652 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9653 constraint on operand 0 to allow more general addressing modes.
9654 Adjust output template.
9655 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9657 * config/aarch64/aarch64-protos.h
9658 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9659 * config/aarch64/constraints.md (Dp): New address constraint.
9660 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9663 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9665 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9666 update use of estimate_ipcp_clone_size_and_time.
9667 (estimate_local_effects): Update use of
9668 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9669 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9670 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9671 Return nonspecialized time.
9673 2017-05-04 Richard Biener <rguenther@suse.de>
9675 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9676 for the last VUSE which def dominates the PHI. Directly call
9678 (get_continuation_for_phi_1): Remove.
9680 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9682 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9683 to explain the use of truncating division. Cap the number of
9684 iterations to the maximum given by nb_iterations_upper_bound,
9687 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9689 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9690 * configure: Regenerate.
9691 * config.in: Regenerate.
9692 * config/i386/driver-mingw32.c: new file.
9693 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9694 * config.host: Link driver-mingw32.o on MinGW host.
9695 * doc/install.texi: Document new --enable-mingw-wildcard configure
9698 2017-05-04 Marek Polacek <polacek@redhat.com>
9700 PR tree-optimization/80612
9701 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9703 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9704 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9706 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9707 (movt splitter): Likewise.
9708 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9709 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9710 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9711 block for Thumb-1 with MOVT.
9712 (thumb2_legitimate_address_p): Move code block ...
9713 (can_avoid_literal_pool_for_label_p): ... into this new function.
9714 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9716 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9717 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9718 "M-profile targets with the MOVT instruction".
9720 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9722 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9723 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9724 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9726 2017-05-04 Martin Liska <mliska@suse.cz>
9728 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9731 2017-05-04 Richard Biener <rguenther@suse.de>
9733 * tree.c (array_at_struct_end_p): Handle arrays at struct
9734 end with flexarrays more conservatively. Refactor and treat
9735 arrays of arrays or aggregates more strict. Fix
9736 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9737 * tree.c (array_at_struct_end_p): Adjust prototype.
9738 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9739 * gimple-fold.c (get_range_strlen): Likewise.
9740 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9742 2017-05-04 Richard Biener <rguenther@suse.de>
9744 PR tree-optimization/31130
9745 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9747 (supports_overflow_infinity): Likewise.
9748 (is_negative_overflow_infinity): Likewise.
9749 (is_positive_overflow_infinity): Likewise.
9750 (is_overflow_infinity): Likewise.
9751 (stmt_overflow_infinity): Likewise.
9752 (overflow_infinity_range_p): Likewise.
9753 (usable_range_p): Remove as always returning true.
9754 (make_overflow_infinity): Remove.
9755 (negative_overflow_infinity): Likewise.
9756 (positive_overflow_infinity): Likewise.
9757 (avoid_overflow_infinity): Likewise.
9758 (set_value_range): Adjust accordingly.
9759 (set_value_range_to_nonnegative): Likewise, remove now unused
9760 overflow_infinity arg.
9761 (vrp_operand_equal_p): Adjust.
9762 (update_value_range): Likewise.
9763 (range_int_cst_singleton_p): Likewise.
9764 (operand_less_p): Likewise.
9765 (compare_values_warnv): Likewise.
9766 (extract_range_for_var_from_comparison_expr): Likewise.
9767 (vrp_int_const_binop): Likewise.
9768 (zero_nonzero_bits_from_vr): Likewise.
9769 (extract_range_from_multiplicative_op_1): Likewise.
9770 (extract_range_from_binary_expr_1): Likewise.
9771 (extract_range_from_unary_expr): Likewise.
9772 (extract_range_from_comparison): Likewise.
9773 (extract_range_basic): Likewise.
9774 (adjust_range_with_scev): Likewise.
9775 (compare_ranges): Likewise.
9776 (compare_range_with_value): Likewise.
9777 (dump_value_range): Likewise.
9778 (test_for_singularity): Likewise, remove strict_overflow_p parameter
9780 (simplify_cond_using_ranges): Adjust.
9782 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
9784 * brig-builtins.def: Added a builtin for class_f64.
9785 * builtin-types.def: Added a builtin type needed by class_f64.
9787 2017-05-03 Jason Merrill <jason@redhat.com>
9789 * timevar.def: Add TV_CONSTEXPR.
9791 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9793 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
9795 2017-05-03 Martin Jambor <mjambor@suse.cz>
9797 * ipa-prop.c (ipa_update_after_lto_read): Removed.
9798 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
9799 * ipa-cp.c (ipcp_propagate_stage): Do not call
9800 ipa_update_after_lto_read.
9801 * ipa-inline.c (ipa_inline): Likewise.
9803 2017-05-03 Martin Jambor <mjambor@suse.cz>
9805 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
9806 tag. Added a default constructor and a destructor.
9807 (ipa_edge_args_sum_t): New class;
9808 (ipa_edge_args_sum): Declare.
9809 (ipa_edge_args_vector): Remove declaration.
9810 (IPA_EDGE_REF): Use ipa_edge_args_sum.
9811 (ipa_free_edge_args_substructures): Remove declaration.
9812 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
9813 (ipa_edge_args_info_available_for_edge_p): Likewise.
9814 * ipa-prop.c (ipa_edge_args_vector): Removed.
9815 (edge_removal_hook_holder): Likewise.
9816 (edge_duplication_hook_holder): Likewise.
9817 (ipa_edge_args_sum): New variable.
9818 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
9819 ipa_edge_args_vector.
9820 (ipa_free_edge_args_substructures): Likewise.
9821 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
9822 ipa_edge_args_vector.
9823 (ipa_edge_removal_hook): Turned into method
9824 ipa_edge_args_sum_t::remove.
9825 (ipa_edge_duplication_hook): Turned into method
9826 ipa_edge_args_sum_t::duplicate.
9827 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
9828 registering edge hooks.
9829 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
9830 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
9831 ipa_edge_args_sum instead of ipa_edge_args_vector.
9832 * ipa-profile.c (ipa_profile): Likewise.
9834 2017-05-03 Martin Jambor <mjambor@suse.cz>
9836 * symbol-summary.h (function_summary): New method exists.
9837 (function_summary::symtab_removal): Deallocate through release.
9838 (call_summary): New class.
9839 (gt_ggc_mx): New overload.
9840 (gt_pch_nx): Likewise.
9841 (gt_pch_nx): Likewise.
9843 2017-05-03 Jeff Law <law@redhat.com>
9845 PR tree-optimization/78496
9846 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
9847 from simplify_cond_using_ranges. Split off code to walk
9848 backwards through casts into ...
9849 (simplify_cond_using_ranges_2): New function.
9850 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
9851 (execute_vrp): After identifying jump threads, call
9852 simplify_cond_using_ranges_2.
9854 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
9857 * ipa-inline.h (inline_summary): Add ctor.
9858 (create_ggc): Do not use ggc_cleared_alloc.
9860 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
9861 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9863 * gcc.c (handle_braces): Support escaping in switch matching
9865 * doc/invoke.texi (Spec Files): Document it.
9866 Remove superfluous @code markup in items.
9868 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9870 * diagnostic-show-locus.c (struct column_range): New struct.
9871 (get_affected_columns): New function.
9872 (get_printed_columns): New function.
9873 (struct correction): New struct.
9874 (correction::ensure_capacity): New function.
9875 (correction::ensure_terminated): New function.
9876 (struct line_corrections): New struct.
9877 (line_corrections::~line_corrections): New dtor.
9878 (line_corrections::add_hint): New function.
9879 (layout::print_trailing_fixits): Reimplement in terms of the new
9881 (selftest::test_overlapped_fixit_printing): New function.
9882 (selftest::diagnostic_show_locus_c_tests): Call it.
9884 2017-05-03 Nathan Sidwell <nathan@acm.org>
9886 Canonicalize canonical type hashing
9887 * tree.h (type_hash_canon_hash): Declare.
9888 * tree.c (type_hash_list, attribute_hash_list): Move into
9889 type_hash_canon_hash.
9890 (build_type_attribute_qual_variant): Break out hash code calc into
9891 type_hash_canon_hash.
9892 (type_hash_canon_hash): New. Generic type hash computation.
9893 (build_range_type_1, build_array_type_1, build_function_type,
9894 build_method_type_directly, build_offset_type, build_complex_type,
9895 make_vector_type): Call it.
9897 2017-05-03 Richard Biener <rguenther@suse.de>
9899 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
9900 When all DRs have unknown misaligned do not always peel
9901 when there is a store but apply the same costing model as if
9902 there were only loads.
9904 2017-05-03 Richard Biener <rguenther@suse.de>
9907 PR tree-optimization/80492
9908 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9909 compare_base_decls returning dont-know properly.
9911 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
9913 * config/arm/iterators.md (CCSI): New mode iterator.
9914 (arch): New mode attribute.
9915 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
9916 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
9917 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
9918 code iterator for success result mode.
9919 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
9920 the corresponding new insn generators.
9922 2017-05-03 Bin Cheng <bin.cheng@arm.com>
9925 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9926 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9928 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
9930 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
9931 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
9932 (DDR_A): Wrap DDR argument in brackets.
9933 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
9934 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
9935 (DDR_REVERSED_P): Likewise.
9937 2017-05-03 Jakub Jelinek <jakub@redhat.com>
9939 PR tree-optimization/79472
9940 * tree-switch-conversion.c (struct switch_conv_info): Add
9941 contiguous_range and default_case_nonstandard fields.
9942 (collect_switch_conv_info): Compute contiguous_range and
9943 default_case_nonstandard fields, don't clear final_bb if
9944 contiguous_range and only the default case doesn't have the required
9946 (check_all_empty_except_final): Set default_case_nonstandard instead
9947 of failing if contiguous_range and the default case doesn't have empty
9949 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
9950 and only the default case doesn't have the required constants. Skip
9952 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
9953 if default_case_nonstandard.
9954 (build_constructors): Build constant 1 just once. Assert that default
9955 values aren't inserted in between cases if contiguous_range. Skip
9957 (build_arrays): Skip virtual phis.
9958 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
9959 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
9960 Handle virtual phis.
9961 (gen_inbound_check): Handle default_case_nonstandard case.
9962 (process_switch): Adjust check_final_bb caller. Call
9963 gather_default_values with the first non-default case instead of
9964 default case if default_case_nonstandard.
9966 2017-05-02 Nathan Sidwell <nathan@acm.org>
9968 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
9969 check. Fix formatting.
9971 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
9973 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
9974 errors when comparing specialized and unspecialized times.
9976 2017-05-02 David Malcolm <dmalcolm@redhat.com>
9978 * diagnostic-show-locus.c
9979 (layout::should_print_annotation_line_p): Make private.
9980 (layout::print_annotation_line): Make private.
9981 (layout::annotation_line_showed_range_p): Make private.
9982 (layout::show_ruler): Make private.
9983 (layout::print_source_line): Make private. Pass in line and
9984 line_width, rather than calling location_get_source_line. Drop
9986 (layout::print_leading_fixits): New method.
9987 (layout::print_any_fixits): Rename to...
9988 (layout::print_trailing_fixits): ...this, and make private.
9989 Don't print newline fixits.
9990 (diagnostic_show_locus): Move logic for printing one row into...
9991 (layout::print_line): ...this new function. Move the
9992 location_get_source_line call and error-handling from
9993 print_source_line to here. Call print_leading_fixits, and rename
9994 print_any_fixits to print_trailing_fixits.
9995 (selftest::test_fixit_insert_containing_newline): Update now that
9996 newlines are partially supported.
9997 (selftest::test_fixit_insert_containing_newline_2): New test.
9998 (selftest::test_fixit_replace_containing_newline): Update comments.
9999 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10000 * edit-context.c (class added_line): New class.
10001 (class edited_line): Describe newline handling in comment.
10002 (edited_line::actually_edited_p): New method.
10003 (edited_line::print_content): Delete redundant decl.
10004 (edited_line::m_predecessors): New field.
10005 (edited_file::print_content): Call edited_line::print_content.
10006 (edited_file::print_diff): Update to support newlines.
10007 (edited_file::print_diff_hunk): Likewise.
10008 (edited_file::print_run_of_changed_lines): New function.
10009 (edited_file::print_diff_line): Convert to...
10010 (print_diff_line): ...this.
10011 (edited_file::get_effective_line_count): New function.
10012 (edited_line::edited_line): Initialize new field m_predecessors.
10013 (edited_line::~edited_line): Clean up m_predecessors.
10014 (edited_line::apply_fixit): Handle newlines.
10015 (edited_line::get_effective_line_count): New function.
10016 (edited_line::print_content): New function.
10017 (edited_line::print_diff_lines): New function.
10018 (selftest::test_applying_fixits_insert_containing_newline): New
10020 (selftest::test_applying_fixits_replace_containing_newline): New
10022 (selftest::insert_line): New function.
10023 (selftest::test_applying_fixits_multiple_lines): Add example of
10025 (selftest::edit_context_c_tests): Call the new tests.
10027 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10029 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10030 parameter cand. Update dump information.
10031 (get_computation_cost): Update uses.
10033 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10035 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10036 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10037 (get_computation_at, rewrite_use_address): Update use of
10038 get_computation_aff.
10040 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10042 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10043 (get_computation): Delete.
10044 (get_computation_cost): Implement like get_computation_cost_at.
10045 Use get_computation_at.
10046 (get_computation_cost_at): Delete.
10047 (rewrite_use_nonlinear_expr): Use get_computation_at.
10048 (rewrite_use_compare, remove_unused_ivs): Ditto.
10050 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10052 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10054 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10056 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10057 (ivopts_global_cost_for_size): Rename parameter and update uses.
10058 (iv_ca_recount_cost): Update uses.
10059 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10060 candidates seperately in n_invs and n_cands.
10061 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10063 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10065 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10066 (find_inv_vars_cb): New.
10067 (find_depends): Renamed to ...
10068 (find_inv_vars): ... this.
10069 (add_candidate_1, force_var_cost): Call find_inv_vars.
10070 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10072 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10074 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10075 inv_vars. Add inv_exprs.
10076 (struct iv_cand): Rename depends_on to inv_vars.
10077 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10078 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10079 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10080 (dump_cand): Dump inv_vars.
10081 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10082 (record_invariant, find_depends, add_candidate_1): Ditto.
10083 (set_group_iv_cost, force_var_cost): Ditto.
10084 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10085 (get_computation_cost_at, get_computation_cost): Ditto.
10086 (determine_group_iv_cost_generic): Ditto.
10087 (determine_group_iv_cost_address): Ditto.
10088 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10089 (determine_group_iv_costs): Ditto.
10090 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10091 (iv_ca_set_remove_invariants): Renamed to ...
10092 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10093 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10094 (iv_ca_set_add_invariants): Renamed to ...
10095 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10096 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10097 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10098 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10099 (create_new_ivs): Remove useless dump.
10101 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10103 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10105 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10106 (iv_ca_set_no_cp, create_new_iv): Ditto.
10108 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10110 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10112 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10114 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10115 function tree_check2.
10117 2017-05-02 Martin Liska <mliska@suse.cz>
10119 * doc/gcov.texi: Add missing preposition.
10120 * gcov.c (function_info::function_info): Properly fill up
10121 all member variables.
10123 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10125 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10127 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10129 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10131 2017-05-02 Martin Liska <mliska@suse.cz>
10134 * lto-streamer-in.c (lto_read_tree_1): Remove
10135 LTO_STREAMER_DEBUG.
10136 * lto-streamer.c (struct tree_hash_entry): Likewise.
10137 (struct tree_entry_hasher): Likewise.
10138 (tree_entry_hasher::hash): Likewise.
10139 (tree_entry_hasher::equal): Likewise.
10140 (lto_streamer_init): Likewise.
10141 (lto_orig_address_map): Likewise.
10142 (lto_orig_address_get): Likewise.
10143 (lto_orig_address_remove): Likewise.
10144 * lto-streamer.h: Likewise.
10145 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10146 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10148 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10150 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10151 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10152 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10153 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10154 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10155 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10156 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10157 (mm_maskz_sub_ss): New intrinsics.
10158 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10159 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10160 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10161 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10162 (__builtin_ia32_subss_mask_round): New builtins.
10163 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10164 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10165 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10167 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10168 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10170 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10173 2017-05-02 Martin Jambor <mjambor@suse.cz>
10175 PR tree-optimization/78687
10176 * tree-sra.c (access): New field parent.
10177 (process_subtree_disqualification): New function.
10178 (disqualify_candidate): Call it.
10179 (build_accesses_from_assign): Reset write flag if creating an
10181 (build_access_subtree): Fill in parent field and also prpagate
10182 down grp_write flag.
10183 (create_artificial_child_access): New parameter set_grp_write, set
10184 grp_write to its value.
10185 (propagate_subaccesses_across_link): Also propagate grp_write flag
10187 (propagate_all_subaccesses): Push the closest parent back to work
10188 queue if add_access_to_work_queue returned true.
10190 2017-05-02 Richard Biener <rguenther@suse.de>
10192 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10193 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10194 -fstrict-overflow documentation.
10195 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10196 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10197 flag_strict_overflow.
10198 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10199 * lto-opts.c (lto_write_options): Do not stream it.
10200 * lto-wrapper.c (merge_and_complain): Do not handle it.
10201 * opts.c (default_options_table): Do not set -fstrict-overflow.
10202 (finish_options): Likewise do not clear it when sanitizing.
10203 * simplify-rtx.c (simplify_const_relational_operation): Do not
10204 test flag_strict_overflow.
10206 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10208 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10209 using enabled attribute.
10210 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10211 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10212 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10213 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10214 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10215 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10216 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10217 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10218 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10219 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10221 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10223 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10225 2017-05-02 Richard Biener <rguenther@suse.de>
10227 PR tree-optimization/80591
10229 2017-04-10 Richard Biener <rguenther@suse.de>
10231 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10234 2017-05-02 Richard Biener <rguenther@suse.de>
10236 PR tree-optimization/80549
10237 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10238 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10239 headers if they do not have a preheader.
10241 2017-05-02 Martin Liska <mliska@suse.cz>
10244 * common.opt: Fix typo.
10245 * doc/invoke.texi: Likewise.
10247 2017-05-01 Jan Beulich <jbeulich@suse.com>
10249 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10250 swapping, add (x,x,m,x,n) alternative.
10252 2017-05-01 Nathan Sidwell <nathan@acm.org>
10254 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10255 unnecessary unadjusted_alignment check.
10257 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10260 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10262 * gimplify.c (gimplify_cilk_detach): New function.
10263 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10264 * tree-core.h: Document EXPR_CILK_SPAWN.
10265 * tree.h (EXPR_CILK_SPAWN): Define.
10267 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10269 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10270 to use new fixit_hint representation, using the "replace" logic.
10271 (get_line_span_for_fixit_hint): Likewise.
10272 (layout::print_any_fixits): Likewise.
10273 (selftest::test_one_liner_many_fixits): Rename to...
10274 (selftest::test_one_liner_many_fixits_1): ...this, and update
10275 comment and expected output to reflect that the multiple fix-it
10276 hints are now consolidated into one insertion.
10277 (selftest::test_one_liner_many_fixits_2): New test.
10278 (selftest::test_diagnostic_show_locus_one_liner): Update for
10280 (selftest::test_fixit_consolidation): Update for fix-it API
10282 * diagnostic.c (print_parseable_fixits): Likewise.
10283 * edit-context.c (edited_line::m_line_events): Convert from
10284 auto_vec <line_event *> to auto_vec <line_event>.
10285 (class line_event): Convert from abstract base class to a concrete
10286 class, taking over the role of replace_event.
10287 (class insert_event): Delete.
10288 (class replace_event): Rename to class line_event. Convert to
10290 (edit_context::add_fixits): Reimplement.
10291 (edit_context::apply_insert): Delete.
10292 (edit_context::apply_replace): Rename to...
10293 (edit_context::apply_fixit): ...this. Convert to half-open range.
10294 (edited_file::apply_insert): Delete.
10295 (edited_file::apply_replace): Rename to...
10296 (edited_file::apply_fixit): ...this.
10297 (edited_line::~edited_line): Drop deletion of events.
10298 (edited_line::apply_insert): Delete.
10299 (edited_line::apply_replace): Rename to...
10300 (edited_line::apply_fixit): ...this. Convert to half-open range.
10301 Update for change to type of m_line_events.
10302 * edit-context.h (edit_context::apply_insert): Delete.
10303 (edit_context::apply_replace): Rename to...
10304 (edit_context::apply_fixit): ...this.
10306 2017-05-01 Martin Sebor <msebor@redhat.com>
10308 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10311 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10314 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10315 __get_cpuid_max returns 0.
10316 (__get_cpuid_count): Ditto.
10318 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10320 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10321 replacement expression is another instance of one of its arguments.
10323 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10326 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10327 check for stack push/pop autoinc.
10328 * config/i386/i386.c (ix86_agi_dependent): Return false
10329 if the only reason why modified_in_p returned true is that
10330 addr is SP based and set_insn is a push or pop.
10332 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10334 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10337 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10340 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10341 (account_size_time): Use two predicates - exec_pred and
10343 (evaluate_conditions_for_known_args): Compute both clause and
10345 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10346 (inline_summary_t::duplicate): Update.
10347 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10349 (compute_inline_parameters): Likewise.
10350 (estimate_edge_size_and_time): Update caluclation of time.
10351 (estimate_node_size_and_time): Compute both time and nonspecialized
10353 (estimate_ipcp_clone_size_and_time): Update.
10354 (inline_merge_summary): Update.
10355 (do_estimate_edge_time): Update.
10356 (do_estimate_edge_size): Update.
10357 (do_estimate_edge_hints): Update.
10358 (inline_read_section, inline_write_summary): Stream both new predicates.
10359 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10361 (compute_inlined_call_time): Cleanup.
10362 (big_speedup_p): Update.
10363 (edge_badness): Update.
10364 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10365 (size_time_entry): Replace predicate by exec_predicate and
10366 nonconst_predicate.
10367 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10368 (estimate_edge_time): Return also nonspec_time.
10369 (reset_edge_growth_cache): Update.
10371 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10373 PR rtl-optimization/80491
10374 * ifcvt.c (noce_process_if_block): When looking for x setter
10375 with missing else_bb, don't check only the insn right before
10376 cond_earliest, but look for the last insn that x is modified in
10377 within the same bb.
10379 PR rtl-optimization/80491
10380 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10382 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10384 PR tree-optimization/80487
10385 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10387 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10389 PR tree-optimization/79697
10390 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10391 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10392 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10394 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10395 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10397 2017-04-28 Martin Sebor <msebor@redhat.com>
10399 PR tree-optimization/80523
10400 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10401 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10403 (maybe_warn, format_directive, parse_directive): Use new functions.
10404 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10406 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10408 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10410 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10412 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10413 target_header_dir): Set correctly.
10414 * configure: Regenerated.
10415 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10416 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10417 instead of SYSTEM_HEADER_DIR.
10419 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10421 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10422 (estimate_local_effects): Likewise.
10423 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10424 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10425 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10426 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10427 do_estimate_edge_time, estimate_edge_time): Likewise.
10428 * ipa-inline-analysis.c (estimate_node_size_and_time,
10429 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10430 (estimate_time_after_inlining): Remove.
10432 2017-04-28 Martin Liska <mliska@suse.cz>
10434 * doc/gcov.texi: Enhance documentation of gcov.
10436 2017-04-28 Martin Liska <mliska@suse.cz>
10438 * doc/gcov.texi: Sort options in alphabetic order.
10439 * doc/gcov-dump.texi: Likewise.
10440 * doc/gcov-tool.texi: Likewise.
10441 * gcov.c (print_usage): Likewise.
10442 * gcov-dump.c (print_usage): Likewise.
10443 * gcov-tool.c (print_merge_usage_message): Likewise.
10444 (print_rewrite_usage_message): Likewise.
10445 (print_overlap_usage_message): Likewise.
10447 2017-04-28 Martin Liska <mliska@suse.cz>
10449 PR gcov-profile/53915
10450 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10452 2017-04-28 Martin Liska <mliska@suse.cz>
10454 PR gcov-profile/79891
10455 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10456 is marked by compiler as living on a line.
10457 (get_cycles_count): Remove usage of the union.
10458 (output_intermediate_file): Likewise.
10459 (find_source): Fix GNU coding style.
10460 (accumulate_line_counts): Remove old non-all block mode.
10461 (output_lines): Remove usage of the union.
10462 * profile.c (output_location): Include all BBs, even if
10463 belonging to a same line (and file) as a previous BB.
10465 2017-04-28 Martin Liska <mliska@suse.cz>
10467 * gcov.c (process_args): Handle new argument 'w'.
10468 (read_graph_file): Assign ID to BBs.
10469 (output_branch_count): Display BB # if verbose flag is set.
10470 (output_lines): Likewise for arcs.
10471 (print_usage): Add '--verbose' option help.
10472 * doc/gcov.texi: Document --verbose (-w) option.
10474 2017-04-28 Martin Liska <mliska@suse.cz>
10476 * gcov.c (struct block_location_info): New struct.
10477 (process_file): Fill up the new structure.
10478 (read_graph_file): Replace usage of encoding by the newly added
10480 (add_line_counts): Likewise.
10481 (accumulate_line_counts): Remove usage of the union.
10482 (function_info::function_info): New function.
10483 (function_info::~function_info): Likewise.
10484 (process_file): Call delete instead of release_function.
10485 (release_function): Release the function.
10486 (release_structures): Call delete instead of release_function.
10487 (solve_flow_graph): Replace usage of num_blocks.
10488 (find_exception_blocks): Likewise.
10489 (output_lines): Fix GNU coding style.
10491 2017-04-28 Martin Liska <mliska@suse.cz>
10494 * coverage.c (coverage_remove_note_file): New function.
10495 * coverage.h: Declare the function.
10496 * toplev.c (finalize): Clean if an error has been seen.
10498 2017-04-28 Martin Liska <mliska@suse.cz>
10500 PR gcov-profile/80031
10501 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10502 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10503 * gcov.c (read_graph_file): Read just number of blocks.
10504 * profile.c (branch_prob): Do not stream 0 flags per a basic
10507 2017-04-28 Martin Liska <mliska@suse.cz>
10509 * gcov-dump.c (tag_*): Add new argument to declarations.
10510 (dump_gcov_file): Likewise.
10511 (tag_blocks): Add and use new argument depth.
10512 (tag_arcs): Likewise.
10513 (tag_lines): Likewise.
10514 (tag_counters): Likewise.
10515 (tag_summary): Likewise.
10516 (dump_working_sets): Use depth to do a proper indentation.
10518 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10521 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10522 * symtab.c (symtab_node::debug_symtab): Move definition here.
10524 2017-04-28 Richard Biener <rguenther@suse.de>
10526 * lto-streamer.h (LTO_major_version): Bump to 7.
10528 2017-04-28 Richard Biener <rguenther@suse.de>
10530 * tree-vrp.c (assert_info): New struct.
10531 (add_assert_info): New helper.
10532 (register_edge_assert_for_2): Refactor to add asserts to a vector
10534 (register_edge_assert_for_1): Likewise.
10535 (register_edge_assert_for): Likewise.
10536 (finish_register_edge_assert_for): New helper actually registering
10537 asserts where live on edge.
10538 (find_conditional_asserts): Adjust.
10539 (find_switch_asserts): Likewise.
10540 (evrp_dom_walker::try_find_new_range): Generalize.
10541 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10543 2017-04-27 Marek Polacek <polacek@redhat.com>
10546 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10547 arg10 and arg11 to itype.
10549 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10551 * doc/extend.texi (Object Size Checking): Improve grammar.
10553 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10556 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10557 that the logic for permitting reciprocal estimates matches that
10560 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10563 * tree.c (type_cache_hasher::equal): Only compare
10564 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10565 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10566 non-aggregate element types.
10567 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10568 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10570 2017-04-27 Richard Biener <rguenther@suse.de>
10572 PR middle-end/80533
10573 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10574 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10575 keeping a reference to a trailing array.
10577 2017-04-27 Richard Biener <rguenther@suse.de>
10579 PR middle-end/80539
10580 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10581 being in loop-closed SSA form conservatively.
10582 (chrec_fold_multiply_poly_poly): Likewise.
10584 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10586 PR middle-end/79665
10587 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10588 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10590 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10593 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10594 (aarch64_function_arg_alignment): Return unsigned int again, but still
10595 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10596 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10597 Don't emit -Wpsabi note.
10598 (aarch64_function_arg_boundary): Likewise.
10599 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10602 2017-04-26 Nathan Sidwell <nathan@acm.org>
10604 * tree.h (crc32_unsigned_n): Declare.
10605 (crc32_unsigned, crc32_unsigned): Make inline.
10606 * tree.c (crc32_unsigned_bits): Replace with ...
10607 (crc32_unsigned_n): ... this.
10608 (crc32_unsigned, crc32_byte): Remove.
10609 (crc32_string): Remove unnecessary braces.
10611 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10613 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10614 * ipa-inline-analysis.c (MAX_TIME): Remove.
10615 (account_size_time): Use sreal for time.
10616 (dump_inline_summary): Update.
10617 (estimate_function_body_sizes): Update.
10618 (estimate_edge_size_and_time): Update.
10619 (estimate_calls_size_and_time): Update.
10620 (estimate_node_size_and_time): Update.
10621 (inline_merge_summary): Update.
10622 (inline_update_overall_summary): Update.
10623 (estimate_time_after_inlining): Update.
10624 (inline_read_section): Update.
10625 (inline_write_summary): Update.
10626 * ipa-inline.c (compute_uninlined_call_time): Update.
10627 (compute_inlined_call_time): Update.
10628 (recursive_inlining): Update.
10629 (inline_small_functions): Update.
10630 (dump_overall_stats): Update.
10631 * ipa-inline.h: Include sreal.h.
10632 (size_time_entry): Turn time to sreal.
10633 (inline_summary): Turn self_time nad time to sreal.
10635 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10637 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10639 (sreal::stream_out, sreal::stream_in): New.
10640 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10642 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10644 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10647 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10650 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10652 (dimode_scalar_chain::compute_convert_gain): Ditto.
10653 (dimode_scalar_chain::make_vector_copies): Ditto.
10654 (dimode_scalar_chain::convert_reg): Ditto.
10655 (dimode_scalar_chain::convert_insn): Ditto.
10656 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10657 (VI248_AVX512BW_1): New mode iterator.
10658 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10659 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10662 2017-04-25 Martin Sebor <msebor@redhat.com>
10664 PR tree-optimization/80497
10665 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10666 constants are representable in HOST_WIDE_INT.
10667 (parse_directive): Ditto.
10669 2017-04-25 Martin Sebor <msebor@redhat.com>
10672 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10673 (new_zero_array): Adjust signature.
10674 (dom_info::dom_init): Used unsigned rather that size_t.
10675 (dom_info::dom_info): Same.
10677 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10678 Jakub Jelinek <jakub@redhat.com>
10681 * config/arm/arm.c: Include gimple.h.
10682 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10683 returns negative, increment ncrn only if it returned positive.
10684 (arm_needs_doubleword_align): Return int instead of bool,
10685 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10686 members, but if there is any such non-FIELD_DECL
10687 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10688 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10689 returns negative, increment nregs only if it returned positive.
10690 (arm_setup_incoming_varargs): Likewise.
10691 (arm_function_arg_boundary): Emit -Wpsabi note if
10692 arm_needs_doubleword_align returns negative, return
10693 DOUBLEWORD_ALIGNMENT only if it returned positive.
10695 2017-04-25 Marek Polacek <polacek@redhat.com>
10698 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10699 first argument to type.
10701 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10704 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10705 type checks to test for compatibility instead of equality.
10707 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10708 Jakub Jelinek <jakub@redhat.com>
10711 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10713 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10714 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10715 the alignment computation, but return their maximum in warn_alignment.
10716 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10717 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10719 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10720 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10723 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10725 * config/arc/simdext.md (dmpyh): Fix typo.
10727 2017-04-25 Richard Biener <rguenther@suse.de>
10729 PR tree-optimization/80492
10730 * alias.c (compare_base_decls): Handle registers with asm
10731 specification conservatively.
10732 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10733 compare_base_decls returning dont-know properly.
10735 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10737 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10738 (legitimate_offset_address_p): New function.
10739 (arc_legitimate_address_p): Use above function.
10741 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10743 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10745 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10747 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
10748 ACCH registers whenever they are available.
10750 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10752 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
10753 double regs fix when not used.
10755 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10757 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
10759 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
10760 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
10762 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10764 * config/arc/arc.c (arc_output_addsi): Check for h-register class
10765 when emitting short ADD instructions.
10767 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10769 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
10771 (cmpsi_cc_c_insn): Likewise.
10772 (cbranchsi4_scratch): Compute proper instruction length using
10773 compact_hreg_operand.
10774 * config/arc/predicates.md (compact_hreg_operand): New predicate.
10776 2017-04-25 Richard Biener <rguenther@suse.de>
10778 PR middle-end/80509
10779 * passes.c (pass_manager::pass_manager): Initialize
10780 m_name_to_pass_map.
10782 2017-04-25 Richard Biener <rguenther@suse.de>
10784 PR tree-optimization/79201
10785 * tree-ssa-sink.c (statement_sink_location): Handle calls.
10787 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10790 * config/s390/vector.md: Split MEM->GPR vector moves for
10791 non-s_operand addresses.
10793 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10796 * config/s390/predicates.md (reload_const_wide_int_operand): New
10798 * config/s390/s390.md ("movti"): Remove d/P alternative.
10799 ("movti_bigconst"): New pattern definition.
10801 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10804 * s390-protos.h (s390_expand_cs_hqi): Removed.
10805 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
10806 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
10807 modes as well as CCZ1mode and CCZmode.
10808 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
10809 signature of s390_emit_compare_and_swap.
10810 (s390_expand_cs_hqi): Likewise, make static.
10811 (s390_expand_cs_tdsi): Generate an explicit compare before trying
10812 compare-and-swap, in some cases.
10813 (s390_expand_cs): Wrapper function.
10814 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
10816 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
10817 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
10818 patterns for small and large integers. Forbid symref memory operands.
10819 Move expander to s390.c. Require cc register.
10820 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
10821 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
10822 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
10823 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
10824 symref memory operands. Remove CC mode and call s390_match_ccmode
10826 ("atomic_exchange<mode>"): Allow and implement all integer modes.
10828 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10830 * config/s390/s390.md (define_peephole2): New peephole to help
10831 combining the load-and-test pattern with volatile memory.
10833 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10835 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
10836 with CCZmode for TARGET_Z196.
10838 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10840 PR rtl-optimization/80501
10841 * combine.c (make_compound_operation_int): Set subreg_code to SET
10842 even for AND with mask of the sign bit of mode.
10844 PR rtl-optimization/80500
10845 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
10846 sum's initial value.
10848 2017-04-25 Julian Brown <julian@codesourcery.com>
10849 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10851 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
10853 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
10855 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
10857 2017-04-25 Julian Brown <julian@codesourcery.com>
10858 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10860 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
10861 (thunderx2t99_sha): New Reservation.
10863 2017-04-25 Julian Brown <julian@codesourcery.com>
10864 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10866 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
10867 type for 1-element load.
10869 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
10871 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
10873 2017-04-24 Martin Jambor <mjambor@suse.cz>
10875 PR tree-optimization/80293
10876 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
10877 char arrays not totally scalarizable if it is false.
10878 (analyze_all_variable_accesses): Pass correct value in the new
10879 parameter. Add a statistics counter.
10881 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
10883 PR middle-end/79931
10884 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
10886 2017-04-24 Richard Biener <rguenther@suse.de>
10888 PR tree-optimization/80494
10889 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
10890 out for complex types.
10892 2017-04-24 Richard Biener <rguenther@suse.de>
10894 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
10895 * tree-ssa-sccvn.c (print_scc): Print SCC size.
10896 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
10897 (DFS): Adjust and never fail.
10898 (sccvn_dom_walker::fail): Remove.
10899 (sccvn_dom_walker::before_dom_children): Adjust.
10900 (run_scc_vn): Likewise and never fail.
10901 * tree-ssa-pre.c (pass_pre::execute): Adjust.
10902 (pass_fre::execute): Likewise.
10904 2017-04-24 Richard Biener <rguenther@suse.de>
10906 PR tree-optimization/79725
10907 * tree-ssa-sink.c (statement_sink_location): Return whether
10908 failure reason was zero uses. Move that check later.
10909 (sink_code_in_bb): Deal with zero uses by removing the stmt
10912 2017-04-24 Richard Biener <rguenther@suse.de>
10915 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
10916 pointer-based references.
10918 2017-04-24 Richard Biener <rguenther@suse.de>
10921 * pass_manager.h (pass_manager::operator new): Remove.
10922 (pass_manager::operator delete): Likewise.
10923 * passes.c (pass_manager::operator new): Remove.
10924 (pass_manager::operator delete): Likewise.
10925 (pass_manager::pass_manager): Zero individual pass members.
10927 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
10930 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
10931 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
10932 Check "XEXP (src, 1)" operand here.
10933 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
10934 Check "XEXP (src, 1)" operand here.
10935 (dimode_scalar_chain::make_vector_copies): Detect count register
10936 of a shift instruction. Zero extend count register from QImode
10937 to DImode to satisfy vector shift pattern count operand predicate.
10938 Substitute vector shift count operand with a DImode copy.
10939 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
10942 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
10944 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
10945 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10946 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
10947 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10948 (UNSPEC_NOREX_MEM): Remove definition.
10950 2017-04-21 Richard Biener <rguenther@suse.de>
10952 PR tree-optimization/79547
10953 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10954 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
10955 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
10956 without any constraints.
10958 2017-04-21 Richard Biener <rguenther@suse.de>
10960 PR tree-optimization/78847
10961 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
10963 2017-04-21 Richard Biener <rguenther@suse.de>
10965 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
10966 (build_distinct_type_copy): Likewise.
10967 (build_variant_type_copy): Likewise.
10968 * tree.c (build_qualified_type): Pass down mem-stat info.
10969 (build_distinct_type_copy): Likewise.
10970 (build_variant_type_copy): Likewise.
10972 2017-04-21 Richard Biener <rguenther@suse.de>
10974 PR tree-optimization/80237
10975 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
10977 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
10978 for a simplified result.
10980 2016-04-21 Richard Biener <rguenther@suse.de>
10982 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
10983 sth as strict as a simple_iv but a chrec without symbols and an
10984 operand defined in the loop we are peeling (and not some subloop).
10985 (propagate_constants_for_unrolling): Propagate all constants.
10987 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10990 * config/i386/i386.c (print_reg): Remove assert for disalowed
10991 regno values, call output_operand_lossage instead.
10993 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10996 * config/i386/constraints.md (Yc): New register constraint.
10997 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
10998 Use Yc constraint for alternative 2 of operand 0. Remove
10999 preferred_for_speed attribute.
11001 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11003 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11004 lastprivate clauses in SIMT case.
11006 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11008 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11010 2017-04-20 Richard Biener <rguenther@suse.de>
11012 PR tree-optimization/57796
11013 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11014 as N scalar stores.
11015 (vect_model_load_cost): Cost gathers as N scalar loads.
11017 2017-04-20 Richard Biener <rguenther@suse.de>
11019 * ggc-page.c (ggc_allocated_p): Rename to ...
11020 (safe_lookup_page_table_entry): ... this and return the lookup
11022 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11024 2017-04-20 Richard Biener <rguenther@suse.de>
11026 PR tree-optimization/80453
11027 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11028 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11029 from the conditions.
11030 (vn_phi_eq): Pass them down.
11031 (vn_phi_lookup): Record them.
11032 (vn_phi_insert): Likewise.
11034 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11036 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11037 uninitialized variable warning to avoid buffer overrun.
11039 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11042 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11043 is suppressed for '{ 0 }' in C.
11045 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11047 * BASE-VER: Set to 8.0.0.
11049 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11051 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11052 priority .init_array and .fini_array section with SECTION_NOTYPE
11055 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11057 PR middle-end/80423
11058 * tree.h (build_array_type): Add typeless_storage default argument.
11059 * tree.c (type_cache_hasher::equal): Also compare
11060 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11061 (build_array_type): Add typeless_storage argument, set
11062 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11064 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11065 (build_array_type): Likewise. Add typeless_storage argument.
11067 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11068 Jakub Jelinek <jakub@redhat.com>
11070 PR tree-optimization/80426
11071 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11072 operation on symbolic operands, also compute the overflow for the
11073 invariant part when the operation degenerates into a negation.
11075 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11078 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11079 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11082 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11084 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11087 * config/avr/avr.c (tree.h): Include it.
11088 (cgraph.h): Include it.
11089 (avr_encode_section_info): Don't warn for uninitialized progmem
11090 variable if it's just an alias.
11092 2017-04-19 Richard Biener <rguenther@suse.de>
11095 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11096 when needed by AutoPGO.
11098 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11101 * doc/lto.texi: Remove an extra 'that'.
11103 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11105 PR rtl-optimization/80429
11106 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11107 are only used in debug insns.
11109 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11110 Vladimir Makarov <vmakarov@redhat.com>
11112 * config/sparc/predicates.md (input_operand): Add comment. Return
11113 true for any memory operand when LRA is in progress.
11114 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11116 2017-04-18 Jeff Law <law@redhat.com>
11119 * mips.md ({return,simple_return}_internal): Do not overwrite
11122 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11124 PR tree-optimization/80443
11125 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11126 instead of adding 1, subtract -1 and similarly instead of subtracting
11129 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11131 PR rtl-optimization/80357
11132 * haifa-sched.c (tmp_bitmap): New variable.
11133 (model_recompute): Handle duplicate use records.
11134 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11135 (free_global_sched_pressure_data): Free it.
11137 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11140 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11141 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11142 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11143 instead of SYSTEM_HEADER_DIR.
11145 2017-04-18 Jeff Law <law@redhat.com>
11147 PR middle-end/80422
11148 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11149 predecessors after walking up the insn chain.
11151 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11154 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11155 sizetype type into debug info.
11157 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11160 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11161 unneeded test for TARGET_UPPER_REGS_SF.
11162 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11164 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11167 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11168 instead of gsi_after_labels.
11170 2017-04-18 Jeff Law <law@redhat.com>
11172 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11176 2017-04-13 Jeff Law <law@redhat.com>
11177 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11178 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11180 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11183 * config/avr/avr.c (intl.h): Include it.
11184 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11186 2017-04-18 Martin Liska <mliska@suse.cz>
11188 PR gcov-profile/78783
11189 * gcov-tool.c (gcov_output_files): Validate that destination
11190 file is either removed by the tool or by a user.
11192 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11193 Guy Benyei <guybe@mellanox.com>
11195 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11196 block, and do not negate it, the stored id is already negative.
11198 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11200 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11202 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11205 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11206 masks of options that should be turned off if the VSX vector
11207 options are turned off.
11208 (OTHER_P8_VECTOR_MASKS): Likewise.
11209 (OTHER_VSX_VECTOR_MASKS): Likewise.
11210 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11211 rs6000_disable_incompatible_switches to validate no type switches
11213 (rs6000_incompatible_switch): New function to disallow turning on
11214 other vector options if -mno-vsx, -mno-power8-vector, or
11215 -mno-power9-vector are specified.
11217 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11219 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11221 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11223 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11224 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11225 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11226 (ARG_POINTER_CFA_OFFSET): Likewise.
11228 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11230 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11231 conditions to take advantage of various optimizations.
11233 2017-04-13 Jeff Law <law@redhat.com>
11235 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11236 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11237 (zero_extendsidi2_dext): Likewise.
11239 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11242 * fold-const.c (fold_ternary_loc): Revert
11243 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11246 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11248 PR rtl-optimization/80343
11249 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11250 new scratch pseudo.
11252 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11255 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11256 to ubsan_encode_value.
11258 2017-04-13 Jeff Law <law@redhat.com>
11260 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11261 appearing in DEBUG_INSNs.
11263 2017-04-13 Martin Liska <mliska@suse.cz>
11265 PR gcov-profile/80413
11266 * gcov-io.c (gcov_write_string): Copy to buffer just when
11267 allocated size is greater than zero.
11269 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11272 * dwarf2out.c (decls_for_scope): Ignore declarations of
11273 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11275 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11278 * ipa-visibility.c (non_local_p): Fix typos.
11279 (localize_node): When localizing symbol in same comdat group,
11280 dissolve the group only when we know external symbols are going
11282 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11284 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11286 PR tree-optimization/79390
11287 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11288 order does not result in usable sequence, retry with reversed operand
11294 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11295 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11296 op0 instead of fold_convert_loc (loc, type, arg0).
11298 2017-04-12 Jeff Law <law@redhat.com>
11300 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11301 has a delay slot in the generated code.
11303 * config/cris/cris.md (cris_preferred_reload_class): Return
11304 GENNONACR_REGS rather than GENERAL_REGS.
11306 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11309 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11310 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11311 signedness of the result type.
11313 2017-04-12 Richard Biener <rguenther@suse.de>
11314 Jeff Law <law@redhat.com>
11316 PR tree-optimization/80359
11317 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11318 trim stores to TARGET_MEM_REFs.
11320 2017-04-12 Richard Biener <rguenther@suse.de>
11322 PR tree-optimization/79390
11323 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11324 threading case even more.
11326 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11329 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11330 for quad_address_p for TImode, instead of just not indexed_address.
11332 2017-04-12 Richard Biener <rguenther@suse.de>
11333 Bernd Edlinger <bernd.edlinger@hotmail.de>
11335 PR middle-end/79671
11336 * alias.c (component_uses_parent_alias_set_from): Handle
11337 TYPE_TYPELESS_STORAGE.
11338 (get_alias_set): Likewise.
11339 * tree-core.h (tree_type_common): Add typeless_storage flag.
11340 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11341 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11342 for types containing members with TYPE_TYPELESS_STORAGE.
11343 (place_field): Likewise.
11344 (layout_type): Likewise for ARRAY_TYPE.
11345 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11346 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11347 TYPE_TYPELESS_STORAGE.
11348 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11350 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11353 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11354 first argument to type.
11356 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11360 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11361 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11362 (rs6000_expand_binop_builtin): Likewise.
11363 (rs6000_expand_ternop_builtin): Likewise; also add missing
11364 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11365 vshasigma built-ins.
11366 * doc/extend.texi: Document that vec_xxpermdi's third argument
11367 must be a constant.
11369 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11371 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11372 Use shift_const cost parameter when calculating gain of STV shifts.
11374 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11376 PR rtl-optimization/70478
11377 * lra-constraints.c (process_alt_operands): Check memory for
11378 disfavoring memory insn operand.
11380 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11382 PR middle-end/80100
11383 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11384 left shift in unsigned HOST_WIDE_INT type.
11386 PR rtl-optimization/80385
11387 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11388 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11391 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11392 if they have any depend clauses.
11394 2017-04-11 Martin Liska <mliska@suse.cz>
11397 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11398 * ipa-split.c (split_function): Create a local comdat symbol
11399 if caller is in a comdat group.
11401 2017-04-11 Martin Liska <mliska@suse.cz>
11404 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11407 2017-04-11 Martin Sebor <msebor@redhat.com>
11409 PR middle-end/80364
11410 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11411 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11413 (directive::set_width, directive::set_precision, format_character):
11415 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11418 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11421 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11422 conflict, set target->arch_name instead of target->cpu_name.
11424 2017-04-11 Richard Biener <rguenther@suse.de>
11426 PR tree-optimization/80374
11427 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11428 build_zero_cst, remove fold_convertible_p check again.
11430 2017-04-11 Martin Liska <mliska@suse.cz>
11433 * ubsan.c (instrument_object_size): Do not instrument register
11436 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11439 * config/i386/i386-builtin-types.def
11440 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11441 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11442 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11443 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11444 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11445 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11446 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11447 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11448 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11449 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11450 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11451 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11452 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11453 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11454 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11455 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11456 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11457 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11458 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11459 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11460 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11461 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11462 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11463 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11464 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11465 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11466 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11467 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11468 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11469 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11470 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11471 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11472 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11473 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11474 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11475 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11476 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11477 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11478 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11479 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11480 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11481 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11482 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11483 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11484 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11485 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11487 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11488 flag to second_arg_count, handle 4 argument function type _COUNT
11489 aliases, handle second_arg_count on second argument rather than last.
11491 2017-04-10 Jeff Law <law@redhat.com>
11493 PR tree-optimization/80374
11494 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11495 record anything if we can not convert integer_zero_node to the
11498 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11501 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11502 Enhance special handling given to the TARGET_P9_MINMAX option in
11503 relation to certain other options.
11505 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11507 PR tree-optimization/80153
11508 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11509 remove POINTER_PLUS_EXPR's base part directly, rather than through
11512 2017-04-10 Richard Biener <rguenther@suse.de>
11513 Bin Cheng <bin.cheng@arm.com>
11515 PR tree-optimization/80153
11516 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11517 the first element of pointer type aff_tree. Build result expr in
11519 (add_elt_to_tree): Convert to type unconditionally. Remove other
11520 fold_convert calls.
11521 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11522 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11524 2017-04-10 Richard Biener <rguenther@suse.de>
11526 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11529 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11531 PR rtl-optimization/70478
11532 * lra-constraints.c (curr_small_class_check): New.
11533 (update_and_check_small_class_inputs): New.
11534 (process_alt_operands): Update curr_small_class_check. Disfavor
11535 alternative insn memory operands. Check available regs for small
11538 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11541 * config/mips/mips.opt (-mvirt): Update description.
11542 * doc/invoke.texi (-mvirt): Likewise.
11544 2017-04-10 Richard Biener <rguenther@suse.de>
11546 PR middle-end/80362
11547 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11548 looking for NEGATE_EXPR in -A / -B to A / B folding.
11550 2017-04-10 Martin Liska <mliska@suse.cz>
11552 PR gcov-profile/80224
11553 * gcov.c (print_usage): Fix usage string.
11554 (get_gcov_intermediate_filename): Remove.
11555 (output_gcov_file): Use both for normal and intermediate format.
11556 (generate_results): Do not initialize special file for
11557 intermediate format.
11559 2017-04-10 Richard Biener <rguenther@suse.de>
11561 PR tree-optimization/80304
11562 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11565 2017-04-10 Nathan Sidwell <nathan@acm.org>
11568 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11569 (rs6000_init_builtins): Use it.
11571 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11573 * config/arm/arm.md (<mrc>): Add mode to SET source.
11574 (<mrrc>): Likewise.
11576 2017-04-10 Richard Biener <rguenther@suse.de>
11578 PR middle-end/80344
11579 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11581 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11584 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11585 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11586 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11587 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11588 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11589 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11590 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11591 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11592 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11593 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11594 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11595 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11596 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11597 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11598 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11599 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11600 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11601 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11602 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11603 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11604 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11605 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11606 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11608 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11610 PR rtl-optimization/70478
11611 * lra-constraints.c: Reverse the last patch.
11613 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11615 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11616 Add comment for WCHAR_T.
11618 2017-04-08 Martin Liska <mliska@suse.cz>
11621 2017-04-07 Martin Liska <mliska@suse.cz>
11624 * ipa-split.c (split_function): Add function part to a same comdat
11627 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11630 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11632 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11634 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11635 (vec_pairing): ...to this.
11636 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11637 (rs6000_sched_init): Adjust for name change.
11638 (struct rs6000_sched_context): Likewise.
11639 (rs6000_init_sched_context): Likewise.
11640 (rs6000_set_sched_context): Likewise.
11642 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11648 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11650 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11651 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11652 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11654 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11656 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11658 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11660 PR rtl-optimization/70703
11661 * ira-color.c (update_conflict_hard_regno_costs): Use
11662 int64_t instead of HOST_WIDE_INT.
11664 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11666 PR rtl-optimization/70478
11667 * lra-constraints.c (process_alt_operands): Disfavor alternative
11668 insn memory operands.
11670 2017-04-07 Jeff Law <law@redhat.com>
11672 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11673 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11675 2017-04-07 Martin Liska <mliska@suse.cz>
11678 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11679 Show error message instead of an ICE.
11681 2017-04-07 Martin Liska <mliska@suse.cz>
11684 * ipa-split.c (split_function): Add function part to a same comdat
11687 2017-04-07 Richard Biener <rguenther@suse.de>
11689 PR middle-end/80341
11690 * tree.c (get_unwidened): Also handle ! for_type case for
11692 * convert.c (do_narrow): Split out from ...
11693 (convert_to_integer_1): ... here. Do not pass final truncation
11694 type to get_unwidened for TRUNC_DIV_EXPR.
11696 2017-04-07 Richard Biener <rguenther@suse.de>
11698 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11700 (aff_combination_const): Adjust.
11701 (aff_combination_scale): Likewise.
11702 (aff_combination_add_elt): Likewise.
11703 (aff_combination_add_cst): Likewise.
11704 (aff_combination_convert): Likewise.
11705 (add_elt_to_tree): Likewise. Remove unused argument.
11706 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11708 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11710 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11712 * config/arm/arm.c (arm_default_short_enums): Use
11713 ARM_DEFAULT_SHORT_ENUMS.
11714 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11716 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11719 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11720 members with redundant out-of-class redeclaration.
11722 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11725 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11726 * config/i386/i386.md (*zero_extendsidi2):
11727 Add (?*x,*x) and (?*v,*v) alternatives.
11729 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11732 * config/i386/i386.c (ix86_expand_builtin)
11733 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11734 mode from insn data. Convert operands to insn operand mode.
11735 Copy operands that don't satisfy insn predicate to a register.
11737 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11739 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11742 2017-04-06 Richard Biener <rguenther@suse.de>
11744 PR tree-optimization/80334
11745 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11746 preserve alignment of accesses.
11748 2017-04-06 Richard Biener <rguenther@suse.de>
11750 PR tree-optimization/80262
11751 * tree-sra.c (build_ref_for_offset): Preserve address-space
11753 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
11754 Drop useless address-space information on MEM_REF offsets.
11756 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
11758 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
11760 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11762 PR rtl-optimization/70703
11763 * ira-color.c (update_conflict_hard_regno_costs): Use
11764 HOST_WIDE_INT instead of long.
11766 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
11769 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
11770 not defined for x86_64 target. Add -mmmx target option when __SSE2__
11772 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
11773 for x86_64 target. Handle -m3dnowa option.
11775 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11777 PR rtl-optimization/70703
11778 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
11779 (update_conflict_hard_regno_costs): Use long instead of unsigned
11780 arithmetic for cost calculation.
11782 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11783 Bernd Edlinger <bernd.edlinger@hotmail.de>
11786 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
11789 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
11792 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
11793 ptr_mode with Pmode throughout.
11794 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
11795 into probe_stack_range and use DImode.
11797 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
11800 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
11801 call_eh_return is true.
11803 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11805 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
11806 Initialize last_match_fntype_index.
11808 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11811 * tree-nvr.c: Include internal-fn.h.
11812 (pass_return_slot::execute): Ignore internal calls without
11815 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11816 Richard Biener <rguenther@suse.de>
11819 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
11820 captures used multiple times, except for the last use.
11821 * generic-match-head.c: Include gimplify.h.
11823 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11825 PR tree-optimization/79390
11826 * target.h (struct noce_if_info): Declare.
11827 * targhooks.h (default_noce_conversion_profitable_p): Declare.
11828 * target.def (noce_conversion_profitable_p): New target hook.
11829 * ifcvt.h (struct noce_if_info): New type, moved from ...
11830 * ifcvt.c (struct noce_if_info): ... here.
11831 (noce_conversion_profitable_p): Renamed to ...
11832 (default_noce_conversion_profitable_p): ... this. No longer
11834 (noce_try_store_flag_constants, noce_try_addcc,
11835 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
11836 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
11837 instead of noce_conversion_profitable_p.
11838 * config/i386/i386.c: Include ifcvt.h.
11839 (ix86_option_override_internal): Don't override
11840 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
11841 (ix86_noce_conversion_profitable_p): New function.
11842 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
11843 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
11844 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
11845 * doc/tm.texi: Regenerated.
11847 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11849 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
11852 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11855 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
11856 instructions for small multiply cores.
11858 2017-04-04 Jeff Law <law@redhat.com>
11860 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
11862 (mips_expand_vec_perm_const): Initialize elements in orig_perm
11863 that are not set by the loop over the elements.
11865 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11868 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
11869 int mode, convert_modes it to mode as unsigned, otherwise use
11870 lowpart_subreg to mode rather than SImode.
11871 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
11872 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
11873 Use DImode instead of SImode for the shift count operand.
11874 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
11877 2017-04-04 Richard Biener <rguenther@suse.de>
11879 PR middle-end/80281
11880 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
11881 arithmetic done for the negate or the plus. Simplify.
11882 (A - (-B) -> A + B): Likewise.
11883 * fold-const.c (split_tree): Make sure to not negate pointers.
11885 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
11887 PR rtl-optimization/60818
11888 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
11889 a compare of comparisons with the thing compared if this results
11890 in a different machine mode.
11892 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
11894 * alias.c (base_alias_check): Fix typo in comment.
11895 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
11896 * cgraphunit.c (symbol_table::compile): Likewise.
11897 * collect2.c (maybe_run_lto_and_relink): Likewise.
11898 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
11899 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
11900 * config/avr/avr.c (avr_map_op_t): Likewise.
11901 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
11902 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
11903 * config/epiphany/epiphany.md (movcc): Likewise.
11904 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
11905 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
11907 * config/mips/mips.c (mips_save_restore_reg): Likewise.
11908 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
11909 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
11910 * config/sh/sh.c (sh_rtx_costs): Likewise.
11911 * fold-const.c (fold_truth_andor): Likewise.
11912 * genautomata.c (collapse_flag): Likewise.
11913 * gengtype.h (struct type::u::s): Likewise.
11914 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
11915 * input.c (FORMAT_AMOUNT): Likewise.
11916 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
11917 (known_aggs_to_agg_replacement_list): Likewise.
11918 * ipa-inline-analysis.c: Likewise.
11919 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
11920 * ipa-polymorphic-call.c
11921 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
11922 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
11923 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
11925 * modulo-sched.c (apply_reg_moves): Likewise.
11926 * omp-expand.c (build_omp_regions_1): Likewise.
11927 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
11928 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
11929 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
11930 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
11931 * value-prof.c: Likewise.
11932 * var-tracking.c (val_reset): Likewise.
11934 2017-04-03 Richard Biener <rguenther@suse.de>
11936 PR tree-optimization/80275
11937 * fold-const.c (split_address_to_core_and_offset): Handle
11940 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
11942 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
11943 descriptors is at least equal to that of functions.
11945 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11947 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
11949 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11952 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
11953 (mov<IMOD4:mode>): New expander.
11954 (*mov<IMOD4:mode>_internal): New insn and split pattern.
11956 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11958 PR rtl-optimization/79405
11959 * fwprop.c (propagations_left): New variable.
11960 (forward_propagate_into): Decrement it.
11961 (fwprop_init): Initialize it.
11962 (fw_prop): If the variable has reached zero, stop propagating.
11963 (fwprop_addr): Ditto.
11965 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11968 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
11969 a FUNCTION_DECL, pass it as decl instead of origin to
11972 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
11974 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
11977 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
11980 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
11981 TARGET_VSX_SMALL_INTEGER.
11983 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11985 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11986 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
11988 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11990 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
11991 extraction from odd-numbered MSA register.
11993 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11995 PR middle-end/80173
11996 * expmed.c (store_bit_field_1): Don't attempt to create
11997 a word subreg out of hard registers wider than word if they
11998 have HARD_REGNO_NREGS of 1 for their mode.
12000 PR middle-end/80163
12001 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12002 conversions to integer types wider than word and pointer.
12005 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12006 (rtx_equal_for_cselib_p): Pass 0 to it.
12007 * cselib.c (cselib_hasher::equal): Likewise.
12008 (rtx_equal_for_cselib_1): Add depth argument. If depth
12009 is 128, don't look up VALUE locs and punt. Increment
12010 depth in recursive calls when walking VALUE locs.
12012 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12014 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12015 (make_gcov_file_name): Use the canonical path name for generating
12017 (read_line): Fix handling of files with ascii null bytes.
12019 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12021 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12022 to initialise a vector register instead
12023 of using a const_int.
12025 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12027 PR translation/80189
12028 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12029 diagnostic messages.
12031 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12034 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12035 (dfp_diex_<mode>): Update mode of operand 1.
12036 * doc/extend.texi (dxex, dxexq): Document change to return type.
12037 (diex, diexq): Document change to argument type.
12039 2017-03-30 Martin Jambor <mjambor@suse.cz>
12042 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12043 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12044 it reflects the signature changes performed at the callee side.
12045 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12046 to cgraph_build_function_type_skip_args.
12047 (build_function_decl_skip_args): Adjust call to the above function.
12049 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12052 * config/i386/sse.md
12053 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12054 register as dest whenever it is a MEM not rtx_equal_p to the
12055 corresponding dup operand, and when forcing into reg move the
12056 reg into the memory afterwards.
12057 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12058 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12059 for the force_reg mode.
12060 (avx512vl_vextractf128<mode>): Use register as dest either
12061 always when a MEM, or when it is a MEM not rtx_equal_p to the
12062 corresponding dup operand, or even not when it is a CONST_VECTOR
12063 depending on the mode and lo vs. hi.
12064 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12066 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12067 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12068 Likewise. Require that operands[2] is even.
12069 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12070 Remove extraneous parens. Require that operands[2] is a multiple
12072 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12073 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12074 disallow memory then.
12076 2017-03-30 Richard Biener <rguenther@suse.de>
12078 PR tree-optimization/77498
12079 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12080 to non-constants over backedges.
12082 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12084 PR rtl-optimization/80233
12085 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12086 as last_combined_insn. Do not test for BARRIER_P separately.
12088 2017-03-29 Andreas Schwab <schwab@suse.de>
12091 * calls.c (prepare_call_address): Convert funexp to Pmode before
12092 copying to temp reg.
12094 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12096 PR tree-optimization/80158
12097 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12098 Handle possible future case of more than one alternate
12100 (replace_rhs_if_not_dup): Likewise.
12101 (replace_one_candidate): Likewise.
12103 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12105 PR rtl-optimization/80193
12106 * ira.c (ira): Do not check allocation for LRA.
12108 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12110 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12111 (nvptx_output_simt_exit): Declare.
12112 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12113 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12114 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12115 (nvptx_declare_function_name): ...here. Emit declaration of local
12116 memory space buffer for omp_simt_enter insn.
12117 (nvptx_output_unisimt_switch): New.
12118 (nvptx_output_softstack_switch): New.
12119 (nvptx_output_simt_enter): New.
12120 (nvptx_output_simt_exit): New.
12121 * config/nvptx/nvptx.h (struct machine_function): New fields
12122 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12123 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12124 (UNSPECV_SIMT_EXIT): Ditto.
12125 (omp_simt_enter_insn): New insn.
12126 (omp_simt_enter): New expansion.
12127 (omp_simt_exit): New insn.
12128 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12130 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12131 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12132 (expand_GOMP_SIMT_EXIT): New.
12133 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12134 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12135 (GOMP_SIMT_EXIT): Ditto.
12136 * target-insns.def (omp_simt_enter): New insn.
12137 (omp_simt_exit): Ditto.
12138 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12140 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12141 (lower_rec_input_clauses): Likewise.
12142 (lower_lastprivate_clauses): Handle SIMT privatization.
12144 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12145 (ompdevlow_adjust_simt_enter): New.
12146 (find_simtpriv_var_op): New.
12147 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12148 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12150 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12151 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12152 (copy_decl_for_dup_finish): Ditto.
12154 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12156 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12159 * config/i386/i386.c (ix86_option_override_internal): Always
12160 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12162 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12164 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12166 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12168 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12169 mark new edge's irreducible flag accordign to it.
12170 (vect_do_peeling): Check loop preheader edge's irreducible flag
12171 and pass it to function slpeel_add_loop_guard.
12173 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12175 PR tree-optimization/80218
12176 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12177 Update block frequencies and counts.
12179 2017-03-28 Richard Biener <rguenther@suse.de>
12181 PR tree-optimization/78644
12182 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12183 of a simplification result we may not use it at all.
12185 2017-03-28 Richard Biener <rguenther@suse.de>
12188 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12189 without arguments, generate default definition of a SSA name.
12191 2017-03-28 Richard Biener <rguenther@suse.de>
12193 PR middle-end/80222
12194 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12195 TYPE_REF_CAN_ALIAS_ALL references.
12196 * fold-const.c (fold_indirect_ref_1): Likewise.
12198 2017-03-28 Martin Liska <mliska@suse.cz>
12201 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12202 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12204 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12205 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12207 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12208 (EXTRA_SPECS): Define.
12209 (SUBTARGET_EXTRA_SPECS): Likewise.
12210 (SUBTARGET_CPP_SPEC): Likewise.
12211 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12212 SUBTARGET_EXTRA_SPECS.
12213 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12215 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12217 * config/arc/simdext.md (vst64_insn): Update pattern.
12218 (vld32wh_insn): Likewise.
12219 (vld32wl_insn): Likewise.
12220 (vld64_insn): Likewise.
12221 (vld32_insn): Likewise.
12223 2017-03-28 Marek Polacek <polacek@redhat.com>
12226 * fold-const.c (fold_comparison): Use protected_set_expr_location
12227 instead of SET_EXPR_LOCATION.
12229 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12231 * tree.c (add_expr): Avoid name lookup warning.
12233 2017-03-27 Jeff Law <law@redhat.com>
12235 PR tree-optimization/80216
12236 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12237 function name. Limit recursion depth.
12238 (record_temporary_equivalences): Corresponding changes.
12240 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12242 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12245 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12248 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12250 * cfgcleanup.c (reg_note_cfa_p): New array.
12251 (insns_have_identical_cfa_notes): New function.
12252 (old_insns_match_p): Don't cross-jump in between /f
12253 and non-/f instructions. If both i1 and i2 are frame related,
12254 verify all CFA notes, their order and content.
12256 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12259 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12260 HImode and SImode with zero extend to DImode to one insn.
12261 (bswap<mode>2_extenddi): Likewise.
12262 (bswapsi2_extenddi): Likewise.
12263 (bswaphi2_extendsi): Likewise.
12264 (bswaphi2): Combine bswap HImode and SImode into one insn.
12265 Separate memory insns from swapping register.
12266 (bswapsi2): Likewise.
12267 (bswap<mode>2): Likewise.
12268 (bswaphi2_internal): Delete, no longer used.
12269 (bswapsi2_internal): Likewise.
12270 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12271 store, and gpr<-gpr swap insns.
12272 (bswap<mode>2_store): Likewise.
12273 (bswaphi2_reg): Register only splitter, combine with the splitter.
12274 (bswaphi2 splitter): Likewise.
12275 (bswapsi2_reg): Likewise.
12276 (bswapsi2 splitter): Likewise.
12277 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12278 the insns into load, store, and register/register insns.
12279 (bswapdi2_ldbrx): Likewise.
12280 (bswapdi2_load): Likewise.
12281 (bswapdi2_store): Likewise.
12282 (bswapdi2_reg): Likewise.
12284 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12286 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12287 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12289 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12292 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12294 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12295 special handling for target option conflicts between dform
12296 options (-mpower9-dform, -mpower9-dform-vector,
12297 -mpower9-dform-scalar) and -mno-direct-move.
12299 2017-03-27 Richard Biener <rguenther@suse.de>
12301 PR tree-optimization/80181
12302 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12304 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12306 * config/arc/predicates.md (move_double_src_operand): Replace the
12307 call to move_double_src_operand with a call to address_operand.
12309 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12311 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12312 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12313 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12315 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12317 * config/arc/predicates.md (long_immediate_loadstore_operand):
12318 Consider scaled addresses cases.
12320 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12322 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12323 restored when in interrupt.
12324 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12325 doesn't have delay slot.
12327 2017-03-27 Richard Biener <rguenther@suse.de>
12330 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12331 inlined thunk clones.
12333 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12336 * asan.c (instrument_derefs): Copy over last operand from
12337 original COMPONENT_REF to the new COMPONENT_REF with
12338 DECL_BIT_FIELD_REPRESENTATIVE.
12339 * ubsan.c (instrument_object_size): Likewise.
12341 2017-03-27 Richard Biener <rguenther@suse.de>
12343 PR tree-optimization/80170
12344 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12345 sure DR/SCEV didnt fold in constants we do not see when looking
12346 at the reference base alignment.
12348 2017-03-27 Richard Biener <rguenther@suse.de>
12350 PR middle-end/80171
12351 * gimple-fold.c (fold_ctor_reference): Properly guard against
12352 NULL return value from canonicalize_constructor_val.
12354 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12357 * config/i386/i386.c (ix86_expand_builtin)
12358 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12359 flags reg setting and flags reg using instructions.
12360 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12361 clobbering instructions to zero extend op2.
12363 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12365 * doc/install.texi (Configuration) <--with-aix-soname>:
12366 Update link to AIX ld.
12368 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12370 PR rtl-optimization/80160
12371 PR rtl-optimization/80159
12372 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12373 reg_alternate_class into account.
12375 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12378 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12379 to consider in curr_insn_transform.
12381 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12383 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12384 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12385 and emit_mode_inner.
12387 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12389 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12390 argument to the overloaded builtin variants. Use the new flag to
12391 deprecate certain builtin variants.
12392 * config/s390/s390-builtin-types.def: Add new builtin types.
12393 * config/s390/s390-builtins.h: Support new flags field for
12394 overloaded builtins.
12395 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12396 (s390_macro_to_expand): Enable vector float data type.
12397 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12398 builtins by incrementing the __VEC__ version number.
12399 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12401 (s390_resolve_overloaded_builtin): Emit error messages depending
12402 on the builtin flags.
12403 * config/s390/s390.c (s390_expand_builtin): Support additional
12404 flags argument. Change error message to match the messages
12405 emitted in s390-c.c.
12406 * config/s390/s390.md: New UNSPEC_* constants.
12407 (op_type): Add new instruction types.
12408 * config/s390/vecintrin.h: Add new builtins and test data class
12410 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12411 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12412 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12413 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12414 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12415 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12417 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12418 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12419 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12420 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12422 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12423 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12424 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12425 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12427 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12428 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12429 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12430 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12431 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12432 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12433 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12435 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12436 ("vec_scatter_element<V_HW_4:mode>_DI")
12437 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12438 ("vec_fpint<mode>", "vflls")
12439 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12440 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12441 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12442 ("*vec_cmphe<mode>_cc"): ... these.
12444 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12445 mode constant instead of magic value.
12447 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12449 * config/s390/s390.c (s390_expand_vec_compare): Support other
12450 vector floating point modes than just V2DF.
12451 (s390_expand_vcond): Likewise.
12452 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12453 (s390_cannot_change_mode_class): Prevent mode changes between TF
12454 and V1TF in vector registers.
12455 * config/s390/s390.md (DF, SF): New mode attributes.
12456 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12457 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12458 SFmode support for VRs.
12459 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12461 (VFT, VF_HW): New mode iterators.
12462 (vw, sdx): New mode attributes.
12463 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12464 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12465 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12466 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12467 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12468 also the new vector floating point modes. Renaming to ...
12470 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12471 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12472 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12473 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12474 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12475 ("vec_unordered<mode>"): ... these.
12477 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12478 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12479 ("*vec_extendv2df"): New insn definitions.
12481 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12483 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12484 ("mulditi3_2", "*muldi3_sign"): New patterns.
12485 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12486 rename the pattern definition.
12488 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12490 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12492 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12494 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12496 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12497 instruction if possible.
12498 * config/s390/vector.md (vec_halfnumelts): New mode
12500 ("*vec_vllezlf<mode>"): New pattern.
12502 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12504 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12505 ("popcountv4si2", "popcountv2di2"): Rename to ...
12506 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12507 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12509 ("popcount<mode>2_vxe"): New pattern.
12511 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12513 * common/config/s390/s390-common.c (processor_flags_table): Add
12515 * config.gcc: Add arch12.
12516 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12517 Default to arch12 for unknown CPU model numbers.
12518 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12519 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12520 PROCESSOR_max sanity check.
12521 * config/s390/s390-opts.h (enum processor_type): Add
12523 * config/s390/s390.c (processor_table): Add arch12.
12524 (s390_expand_builtin): Add check for B_VXE flag.
12525 (s390_issue_rate): Add PROCESSOR_ARCH12.
12526 (s390_get_sched_attrmask): Likewise.
12527 (s390_get_unit_mask): Likewise.
12528 (s390_sched_score): Enable z13 scheduling for arch12.
12529 (s390_sched_reorder): Likewise.
12530 (s390_sched_variable_issue): Likewise.
12531 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12533 (s390_tune_attr): Use z13 scheduling also for arch12.
12534 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12535 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12536 (TARGET_VXE_P): New macros.
12537 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12538 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12539 * config/s390/s390.opt: Add arch12 as processor_type.
12541 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12543 * config/s390/s390.md
12544 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12545 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12546 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12548 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12549 Rename expanders to ...
12551 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12552 ("fixuns_truncdddi2_emu"): ... these.
12554 ("fixuns_trunc<mode>si2_emu"): New expander.
12556 ("*fixuns_truncdfdi2_z13"): Rename to ...
12557 ("*fixuns_truncdfdi2_vx"): ... this.
12559 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12561 * config/s390/2964.md: Remove the single element vector compare
12562 instructions which are no longer used.
12563 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12565 (s390_canonicalize_comparison): Remove handling of DFmode
12567 (s390_expand_vec_compare_scalar): Remove function.
12568 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12569 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12571 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12575 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12577 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12578 will anyway by matched by mov<mode>_64dfp.
12580 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12582 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12583 vlef/vstef. Add missing operand to vleif.
12585 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12587 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12588 pair for all vector types with 64 bit elements.
12589 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12590 * config/s390/vector.md (V_HW_64): ... here.
12591 (V_128_NOSINGLE): New mode iterator.
12592 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12593 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12594 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12595 ("*vec_load_pairv2di"): Change to ...
12596 ("*vec_load_pair<mode>"): ... this one.
12598 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12600 * config/s390/constraints.md: Add comments.
12601 (jKK): Reject element sizes > 8 bytes.
12602 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12604 * config/s390/s390.md: Add the s_operand checks formerly in
12605 s390_split_ok_p to various splitters where they are still
12607 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12608 for 128 bit vectors. Plus two splitters.
12610 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12612 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12615 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12618 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12619 error if the boundary argument is not constant.
12621 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12623 PR rtl-optimization/80112
12624 * loop-doloop.c (doloop_condition_get): Don't check condition
12625 if cmp isn't SET with IF_THEN_ELSE src.
12627 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12629 PR tree-optimization/80158
12630 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12631 replacing a candidate statement, also replace it for the
12632 candidate's alternate interpretation.
12633 (replace_rhs_if_not_dup): Likewise.
12634 (replace_one_candidate): Likewise.
12636 2017-03-24 Richard Biener <rguenther@suse.de>
12638 PR tree-optimization/80167
12639 * graphite-isl-ast-to-gimple.c
12640 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12642 (translate_isl_ast_to_gimple::get_rename): Likewise.
12644 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12646 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12647 handling of certain combinations of target options, including the
12648 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12649 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12651 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12654 * config/arm/arm.md (*load_multiple): Add reload_completed to
12655 matching condition.
12657 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12658 Richard Biener <rguenth@suse.de>
12660 PR tree-optimization/79908
12661 PR tree-optimization/80136
12662 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12663 been cast away, gimplify_and_add suffices.
12665 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12667 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12669 2017-03-23 Richard Biener <rguenther@suse.de>
12671 PR tree-optimization/80032
12672 * gimplify.c (gimple_push_cleanup): Forced unconditional
12673 cleanups still have to go to the conditional_cleanups
12676 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12678 PR tree-optimization/80072
12679 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12681 (next_operand_entry_id): Change type to unsigned int.
12682 (sort_by_operand_rank): Make sure to return the right return value
12683 even if unsigned fields are bigger than INT_MAX.
12684 (struct oecount): Change cnt and id type to unsigned int.
12685 (oecount_hasher::equal): Formatting fix.
12686 (oecount_cmp): Make sure to return the right return value
12687 even if unsigned fields are bigger than INT_MAX.
12688 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12691 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12692 TREE_READONLY on result if writing it more than once.
12695 * doc/invoke.texi (-fsanitize=thread): Document that with
12696 -fnon-call-exceptions atomics are not able to throw
12700 * tsan.c: Include tree-eh.h.
12701 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12702 maybe_clean_or_replace_eh_stmt where needed.
12703 (instrument_memory_accesses): Add cfg_changed argument.
12704 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12705 if it returned true.
12706 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12708 PR rtl-optimization/63191
12709 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12710 wrapper function, moved the whole old content into ...
12711 (ix86_delegitimize_address_1): ... this. New inline function.
12712 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12713 true as last argument instead of ix86_delegitimize_address.
12715 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12717 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12718 cortexa57_branch_cost.
12720 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12722 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12724 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12727 * doc/md.texi (Constraints): Document wA constraint.
12728 * config/rs6000/constraints.md (wA): New.
12729 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12730 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12731 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12732 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12734 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12737 * gimplify.c (is_oacc_declared): New function.
12738 (oacc_default_clause): Use it to set default flags for acc declared
12739 variables inside parallel regions.
12740 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12741 declared variables.
12742 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12743 declare attribute to any decl as necessary.
12745 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12748 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
12749 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
12750 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
12751 (arm_arch_lpae): This.
12752 * config/arm/arm.c (arm_arch7ve): Rename into ...
12753 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
12754 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
12757 2017-03-22 Martin Liska <mliska@suse.cz>
12760 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
12761 error message instead of an ICE.
12763 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12765 * doc/extend.texi (6.11 Additional Floating Types): Revise.
12767 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
12769 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12771 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12774 2017-03-21 Martin Sebor <msebor@redhat.com>
12776 * doc/extend.texi: Use "cannot" instead of "can't."
12777 * doc/hostconfig.texi: Same.
12778 * doc/install.texi: Same.
12779 * doc/invoke.texi: Same.
12780 * doc/loop.texi: Same.
12781 * doc/md.texi: Same.
12782 * doc/objc.texi: Same.
12783 * doc/rtl.texi: Same.
12784 * doc/tm.texi: Same.
12785 * doc/tm.texi.in: Same.
12786 * doc/trouble.texi: Same.
12788 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
12791 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
12792 (collect_checksum_attributes): Set it.
12793 (die_checksum_ordered): Use it.
12795 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12797 PR tree-optimization/79908
12798 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
12799 change: For a VA_ARG whose LHS has been cast away, use
12800 force_gimple_operand to construct the side effects.
12802 2017-03-21 David Malcolm <dmalcolm@redhat.com>
12804 PR translation/80001
12805 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
12806 more amenable to translation.
12807 (oacc_loop_auto_partitions): Likewise.
12809 2017-03-21 Marek Polacek <polacek@redhat.com>
12810 Martin Sebor <msebor@redhat.com>
12812 PR tree-optimization/80109
12813 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
12814 on INTEGRAL_TYPE_P.
12816 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12817 Segher Boessenkool <segher@kernel.crashing.org>
12820 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
12821 check reg_used_between_p between insn and one of succ or succ2
12822 depending on if succ is artificial insn not inserted into insn
12825 2017-03-21 Martin Liska <mliska@suse.cz>
12827 PR gcov-profile/80081
12828 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
12829 * doc/gcc.texi: Include gcov-dump stuff.
12830 * doc/gcov-dump.texi: New file.
12832 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
12834 PR rtl-optimization/79150
12835 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
12836 conditional jump, if the jump is the last insn of the loop.
12838 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12839 Richard Biener <rguenth@suse.de>
12841 PR tree-optimization/79908
12842 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12843 been cast away, use force_gimple_operand to construct the side
12846 2017-03-21 Martin Liska <mliska@suse.cz>
12848 PR libfortran/79956
12849 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
12852 2017-03-21 Brad Spengler <spender@grsecurity.net>
12855 * plugin.c (htab_hash_plugin): New function.
12856 (add_new_plugin): Use it and adjust.
12857 (parse_plugin_arg_opt): Adjust.
12858 (init_one_plugin): Likewise.
12860 2017-03-21 Richard Biener <rguenther@suse.de>
12862 PR tree-optimization/80032
12863 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
12864 if set force the cleanup to happen unconditionally.
12865 (gimplify_target_expr): Push inserted clobbers with force_uncond
12866 to avoid them being removed by control-dependent DCE.
12868 2017-03-21 Richard Biener <rguenther@suse.de>
12870 PR tree-optimization/80122
12871 * tree-inline.c (copy_bb): Do not expans va-arg packs or
12872 va_arg_pack_len when the inlined call stmt requires pack
12874 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
12876 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12879 * tsan.c (instrument_builtin_call): If the memory model argument
12880 is not a constant, assume it is valid.
12883 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
12886 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12888 PR rtl-optimization/79910
12889 * combine.c (can_combine_p): Do not allow combining an I0 or I1
12890 if its dest is used by an insn before I2 (other than the combined
12891 insns themselves, which are properly handled already).
12893 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12896 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12898 * combine.c (record_used_regs): New static function.
12899 (try_combine): Handle situations where there is an additional
12900 instruction between I2 and I3 which needs to have a LOG_LINK
12904 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12906 * combine.c (try_combine): Delete redundant i1 test. Call
12907 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12909 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12912 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
12913 alternatives 13/14.
12915 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12917 PR tree-optimization/80054
12918 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
12919 the optimization if a PHI or any of its arguments is not dominated
12920 by the candidate's basis. Use gphi* rather than gimple* as
12922 (replace_profitable_candidates): Clean up a gimple* variable that
12923 should be a gphi* variable.
12925 2017-03-20 Martin Sebor <msebor@redhat.com>
12928 * doc/extend.texi (attribute constructor): Document present limitation.
12930 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12933 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
12934 __POWER9_VECTOR__ #ifdef control, change template definition to
12935 use Power9-specific built-in function.
12936 (vec_any_eq): Likewise.
12937 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
12938 to control outcomes from this test.
12939 (vector_ae_<mode>p): For VEC_F modes, likewise.
12941 2017-03-20 Ian Lance Taylor <iant@google.com>
12943 * config/i386/i386.c (ix86_function_regparm): Save an extra
12944 register for -fsplit-stack with DECL_STATIC_CHAIN.
12946 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12949 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
12950 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
12952 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12954 * config/riscv/riscv.c (riscv_print_operand): Use "fence
12956 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
12959 2017-03-20 Marek Polacek <polacek@redhat.com>
12962 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
12964 2017-03-20 Richard Biener <rguenther@suse.de>
12966 PR tree-optimization/80113
12967 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
12968 allocate extra SSA name for PHI def.
12969 (add_close_phis_to_outer_loops): Likewise.
12970 (add_close_phis_to_merge_points): Likewise.
12971 (copy_loop_close_phi_args): Likewise.
12972 (copy_cond_phi_nodes): Likewise.
12974 2017-03-20 Martin Liska <mliska@suse.cz>
12976 PR middle-end/79753
12977 * tree-chkp.c (chkp_build_returned_bound): Do not build
12978 returned bounds for a LHS that's not a BOUNDED_P type.
12980 2017-03-20 Martin Liska <mliska@suse.cz>
12984 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
12985 COMPLEX_CST and VECTOR_CST.
12987 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12990 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
12991 target operand. A new splitter adds the clobber statement in case
12992 the target operand is dead anyway.
12994 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
12996 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
12997 to age-old versions of binutils and glibc.
12999 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13001 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13003 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13005 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13007 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13009 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13010 requirement for binutils 2.13.
13012 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13014 * combine.c (try_combine): Delete redundant i1 test. Call
13015 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13017 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13019 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13020 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13022 <riscv64-*-elf>: Re-arrange section
13023 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13024 <riscv32-*-linux>: Likewise.
13025 <riscv64-*-elf>: Likewise
13026 <riscv64-*-linux>: Likewise.
13028 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13031 * aarch64.opt(verbose-cost-dump): Fix typo.
13033 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13036 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13037 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13039 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13041 * reload.c (find_reloads): When reloading a nonoffsettable address,
13042 use RELOAD_OTHER for it and its address reloads.
13044 PR rtl-optimization/79910
13045 * combine.c (record_used_regs): New static function.
13046 (try_combine): Handle situations where there is an additional
13047 instruction between I2 and I3 which needs to have a LOG_LINK
13050 2017-03-17 Jeff Law <law@redhat.com>
13052 PR tree-optimization/71437
13053 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13054 conditional in the hash table first.
13055 (vrp_dom_walker::before_dom_children): Extract condition from
13056 ASSERT_EXPR. Record condition, its inverion and any implied
13057 conditions as well.
13059 2017-03-17 Marek Polacek <polacek@redhat.com>
13060 Markus Trippelsdorf <markus@trippelsdorf.de>
13062 PR tree-optimization/80079
13063 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13066 2017-03-17 Richard Biener <rguenther@suse.de>
13068 PR middle-end/80075
13069 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13070 Properly verify the LHS before the RHS possibly claims to be
13072 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13075 2017-03-17 Martin Jambor <mjambor@suse.cz>
13077 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13078 (List of -O2 options): Likewise.
13079 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13082 2017-03-17 Tom de Vries <tom@codesourcery.com>
13084 * gcov-dump.c (print_usage): Print bug_report_url.
13086 2017-03-17 Richard Biener <rguenther@suse.de>
13088 PR middle-end/80050
13089 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13090 (parser::peek): Likewise.
13092 2017-03-17 Richard Biener <rguenther@suse.de>
13094 PR tree-optimization/80048
13095 * sese.c (free_sese_info): Properly release rename_map and
13096 copied_bb_map elements.
13098 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13100 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13101 Add linked-list forward and backlinks. Insert on
13102 construction, remove on destruction.
13103 (class pass_store_merging): Add m_stores_head field.
13104 (pass_store_merging::terminate_and_process_all_chains):
13105 Iterate over m_stores_head list.
13106 (pass_store_merging::terminate_all_aliasing_chains):
13108 (pass_store_merging::execute): Check for debug stmts first.
13109 Push new chains onto the m_stores_head stack.
13111 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13114 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13115 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13116 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13118 2017-03-16 Jeff Law <law@redhat.com>
13120 PR tree-optimization/71437
13121 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13122 member function. Implementation moved into after_dom_children
13123 member function and into the threader's thread_outgoing_edges
13125 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13126 some code into new thread_outgoing_edges.
13127 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13128 definition. Simplify marker handling (do it here). Assume we always
13129 have the available expression and the const/copies tables.
13130 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13132 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13133 * tree-vrp.c (equiv_stack): No longer file scoped.
13134 (vrp_dom_walker): New class.
13135 (vrp_dom_walker::before_dom_children): New member function.
13136 (vrp_dom_walker::after_dom_children): Likewise.
13137 (identify_jump_threads): Setup domwalker. Use it rather than
13138 walking edges in a random order by hand. Simplify setup/finalization.
13139 (finalize_jump_threads): Remove.
13140 (vrp_finalize): Do not call identify_jump_threads here.
13141 (execute_vrp): Do it here instead and call thread_through_all_blocks
13144 PR tree-optimization/71437
13145 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13147 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13149 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13150 (dom_opt_dom_walker::thread_across_edge): Remove
13151 handle_dominating_asserts argument. All callers changed.
13152 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13153 changes. Remove calls to lhs_of_dominating_assert. Other
13154 uses of handle_dominating_asserts turn into unconditional code
13155 (simplify_control_stmt_condition_1): Likewise.
13156 (simplify_control_stmt_condition): Likewise.
13157 (thread_through_normal_block, thread_across_edge): Likewise.
13158 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13159 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13160 object if it is not an SSA_NAME.
13161 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13162 before calling into the VRP specific simplifiers.
13163 (identify_jump_threads): Remove handle_dominating_asserts
13166 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13169 * tree-diagnostic.c (default_tree_printer): No longer static.
13170 * tree-diagnostic.h (default_tree_printer): New prototype.
13172 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13174 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13175 Change ins into fmov.
13177 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13179 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13180 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13181 Use h_con constraint for operand 1.
13182 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13183 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13185 2017-03-15 Jeff Law <law@redhat.com>
13187 PR tree-optimization/71437
13188 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13189 (record_temporary_equivalences): Use it.
13191 PR tree-optimization/71437
13192 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13193 tree-ssa-scopedtables.
13194 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13195 (record_conditions, record_cond, vuse_eq): Likewise.
13196 (record_edge_info): Adjust to API tweak of record_conditions.
13197 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13198 (record_temporary_equivalences, optimize_stmt): Likewise.
13199 (eliminate_redundant_computations): Likewise.
13200 (record_equivalences_from_stmt): Likewise.
13201 * tree-ssa-scopedtables.c: Include options.h and params.h.
13202 (vuse_eq): New function, moved from tree-ssa-dom.c
13203 (build_and_record_new_cond): Likewise.
13204 (record_conditions): Likewise. Accept vector of conditions rather
13205 than edge_equivalence structure for first argument.
13206 for the first argument.
13207 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13208 from tree-ssa-dom.c.
13209 (avail_exprs_stack::record_cond): Likewise.
13210 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13211 from tree-ssa-dom.c.
13212 (avail_exprs_stack): Add new member functions lookup_avail_expr
13214 (record_conditions): Declare.
13216 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13219 * lra-constraints.c (process_alt_operands): Increase reject for
13220 reloading an input/output operand.
13222 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13225 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13226 insns to convert from signed/unsigned char/short to IEEE 128-bit
13228 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13230 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13233 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13234 subreg of inner mode for values already in registers.
13236 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13238 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13239 iteration reg is used after the loop.
13241 2017-03-14 Martin Sebor <msebor@redhat.com>
13243 PR tree-optimization/79800
13244 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13245 precision in negative-positive range.
13246 (format_floating): Call non-const overload with adjusted precision.
13248 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13251 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13254 2017-03-14 Martin Sebor <msebor@redhat.com>
13256 PR middle-end/80020
13257 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13258 * builtins.def (aligned_alloc): Use it.
13261 * Makefile.in (GTFILES): Add calls.c.
13262 * calls.c: Include "gt-calls.h".
13264 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13266 PR rtl-optimization/79728
13267 * regs.h (struct target_regs): New field
13268 x_contains_allocatable_regs_of_mode.
13269 (contains_allocatable_regs_of_mode): New macro.
13270 * reginfo.c (init_reg_sets_1): Initialize it, and change
13271 contains_reg_of_mode so it includes global regs as well.
13272 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13273 rather than contains_regs_of_mode.
13275 2017-03-14 Martin Liska <mliska@suse.cz>
13277 * doc/invoke.texi: Document options that can't be combined with
13278 -fcheck-pointer-bounds.
13280 2017-03-14 Martin Liska <mliska@suse.cz>
13282 PR middle-end/79831
13283 * doc/invoke.texi (-Wchkp): Document the option.
13285 2017-03-14 Martin Liska <mliska@suse.cz>
13287 * Makefile.in: Install gcov-dump.
13289 2017-03-14 Martin Liska <mliska@suse.cz>
13291 * multiple_target.c (expand_target_clones): Bail out for
13292 an invalid attribute.
13294 2017-03-14 Richard Biener <rguenther@suse.de>
13296 * alias.c (struct alias_set_entry): Pack properly.
13297 * cfgloop.h (struct loop): Likewise.
13298 * cse.c (struct set): Likewise.
13299 * ipa-utils.c (struct searchc_env): Likewise.
13300 * loop-invariant.c (struct invariant): Likewise.
13301 * lra-remat.c (struct cand): Likewise.
13302 * recog.c (struct change_t): Likewise.
13303 * rtl.h (struct address_info): Likewise.
13304 * symbol-summary.h (function_summary): Likewise.
13305 * tree-loop-distribution.c (struct partition): Likewise.
13306 * tree-object-size.c (struct object_size_info): Likewise.
13307 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13308 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13309 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13310 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13311 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13312 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13313 (struct _stmt_vec_info): Likewise.
13315 2017-03-14 Martin Liska <mliska@suse.cz>
13318 * multiple_target.c (create_dispatcher_calls): Check that
13319 a target can create a function dispatcher.
13321 2017-03-14 Martin Liska <mliska@suse.cz>
13324 * multiple_target.c (expand_target_clones): Drop local.local
13325 flag for default implementation.
13327 2017-03-14 Richard Biener <rguenther@suse.de>
13329 PR tree-optimization/80030
13330 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13332 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13334 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13335 gcc_fallthrough() instead of __attribute__((fallthrough));
13337 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13339 * doc/gcc.texi: Remove "up" link to (DIR).
13340 * doc/gccint.texi: Ditto.
13342 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13344 * doc/install.texi (Specific) <avr>: Remove reference to
13347 2017-03-13 Jeff Law <law@redhat.com>
13349 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13350 attribute rather than comments.
13352 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13353 match_scratch operand is highest.
13355 2017-03-13 Martin Liska <mliska@suse.cz>
13357 PR middle-end/78339
13358 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13359 is a CHKP clone, use original declaration.
13361 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13363 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13364 (arc_conditional_register_usage): Use a different allocation order
13365 when optimizing for size.
13366 * common/config/arc/arc-common.c (arc_option_optimization_table):
13367 Section anchors default on when optimizing for size.
13369 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13371 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13373 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13375 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13376 * config/arc/arc.md (cpu_facility): Add cd variant.
13377 (*movqi_insn): Add code density variant.
13378 (*movhi_insn): Likewise.
13379 (*movqi_insn): Likewise.
13380 (*addsi3_mixed): Likewise.
13381 (subsi3_insn): Likewise.
13383 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13385 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13387 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13389 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13390 expressions with MINUS and UNARY ops.
13392 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13395 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13397 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13398 between vec_select and vector argument.
13399 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13400 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13401 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13402 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13403 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13404 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13406 2017-03-13 Richard Biener <rguenther@suse.de>
13409 * params.def (vect-max-peeling-for-alignment): Fix typo.
13411 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13413 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13414 issue that only occurred with binutils below 2.18.
13416 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13418 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13419 refer to binutils 2.11/2.12 minimum.
13421 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13423 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13424 ftp.kernel.org and simplify binutils requirement.
13426 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13428 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13430 (Optimize Options): Ditto. Also remove redundancy.
13432 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13434 PR translation/79848
13435 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13437 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13438 to G_ to avoid double translation.
13440 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13442 PR translation/79923
13443 * auto-profile.c (get_combined_location): Convert leading
13444 character of diagnostics to lower case and remove trailing period.
13445 (read_profile): Likewise for various diagnostics.
13446 * config/arm/arm.c (arm_option_override): Remove trailing period
13447 from various diagnostics.
13448 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13449 (msp430_expand_delay_cycles): Likewise.
13451 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13454 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13455 full command-line argument, rather than just "str".
13456 (aarch64_validate_march): Likewise.
13457 (aarch64_validate_mtune): Likewise.
13459 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13461 PR rtl-optimization/78911
13462 * lra-assigns.c (must_not_spill_p): New function.
13463 (spill_for): Use it.
13465 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13467 PR tree-optimization/79981
13468 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13469 ATOMIC_COMPARE_EXCHANGE ifn result.
13470 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13471 IFN_ATOMIC_COMPARE_EXCHANGE.
13473 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13476 * opts.c (parse_sanitizer_options): Add missing question mark to
13477 "did you mean" message.
13479 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13481 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13483 (VMULEUH_UNS): Likewise.
13484 (VMULOUB_UNS): Likewise.
13485 (VMULOUH_UNS): Likewise.
13486 * config/rs6000/rs6000.c (builtin_function_type): Remove
13487 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13489 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13492 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13493 x with result of extra_parsing_for_operand_code_0.
13494 (function_reader::extra_parsing_for_operand_code_0): Convert
13495 return type from void to rtx, returning x. When reading
13496 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13497 larger size containing struct block_symbol.
13499 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13501 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13502 -mfloat128-hardware without -m64.
13504 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13507 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13508 entries to the case statement that marks unsigned arguments to
13509 overloaded functions.
13511 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13513 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13514 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13516 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13519 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13520 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13522 2017-03-10 Martin Liska <mliska@suse.cz>
13526 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13527 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13530 2017-03-10 Olivier Hainque <hainque@adacore.com>
13532 * tree-switch-conversion (array_value_type): Start by resetting
13533 candidate type to it's main variant.
13535 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13537 PR rtl-optimization/79909
13538 * combine.c (try_combine): Use simplify_replace_rtx on individual
13539 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13540 of the whole CALL_INSN_FUNCTION_USAGE.
13542 PR tree-optimization/79972
13543 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13544 get_range_info on SSA_NAMEs. Formatting fixes.
13546 2017-03-10 Richard Biener <rguenther@suse.de>
13547 Jakub Jelinek <jakub@redhat.com>
13549 PR tree-optimization/77975
13550 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13551 edge to be constant.
13552 (get_val_for): For constant x return it. Formatting fix.
13553 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13554 would use the same bases as the current one.
13556 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13558 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13559 instead of vec_select for V1TImode.
13560 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13562 (VSX_LE_128): Add V1TI to this mode iterator.
13563 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13564 (*vsx_le_perm_store_<mode>): Likewise.
13565 (pre-reload splitter for VSX stores): Likewise.
13566 (post-reload splitter for VSX stores): Likewise.
13567 (*vsx_xxpermdi2_le_<mode>): Likewise.
13568 (*vsx_lxvd2x2_le_<mode>): Likewise.
13569 (*vsx_stxvd2x2_le_<mode>): Likewise.
13571 2017-03-09 Michael Eager <eager@eagercon.com>
13573 Correct failures with --enable-checking=yes,rtl.
13575 * config/microblaze/microblaze.c (microblaze_expand_shift):
13576 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13577 test for const0_rtx.
13578 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13579 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13581 2017-03-09 Richard Biener <rguenther@suse.de>
13583 PR tree-optimization/79977
13584 * graphite-scop-detection.c (scop_detection::merge_sese):
13585 Handle the case of extra exits to blocks dominating the entry.
13587 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13589 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13592 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13594 PR rtl-optimization/79949
13595 * lra-constraints.c (process_alt_operands): Check memory when
13596 trying to predict a cycle. Print about the overall increase.
13598 2017-03-09 Richard Biener <rguenther@suse.de>
13600 PR middle-end/79971
13601 * gimple-expr.c (useless_type_conversion_p): Preserve
13602 TYPE_SATURATING for fixed-point types.
13604 2017-03-09 Richard Biener <rguenther@suse.de>
13607 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13608 alignment of BLKmode params.
13610 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13613 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13614 (VALL_NO_V2Q): Likewise.
13616 * config/aarch64/aarch64-simd.md
13617 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13619 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13620 VALL_NO_V2Q mode iterator.
13621 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13623 2017-03-09 Martin Liska <mliska@suse.cz>
13625 PR tree-optimization/79631
13626 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13627 tree_int_cst_sign_bit just for INTEGER constants.
13629 2017-03-09 Martin Liska <mliska@suse.cz>
13633 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13636 2017-03-09 Marek Polacek <polacek@redhat.com>
13639 * tree.c (inchash::add_expr): Handle TREE_VEC.
13641 2017-03-09 Martin Liska <mliska@suse.cz>
13644 (chkp_narrow_size_and_offset): New function.
13645 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13646 (void chkp_parse_bit_field_ref): New function.
13647 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13648 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13650 2017-03-09 Martin Liska <mliska@suse.cz>
13653 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13654 (chkp_find_bounds_1): Remove gcc_unreachable.
13656 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13659 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13660 BUILT_IN_SYNC*, determine the access type from the size suffix and
13661 always build a MEM_REF with that type. Handle forgotten
13662 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13665 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13666 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13667 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13668 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13669 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13670 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13671 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13672 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13673 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13674 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13675 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13676 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13677 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13678 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13679 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13680 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13681 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13682 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13683 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13684 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13685 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13686 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13687 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13688 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13689 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13690 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13691 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13692 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13693 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13694 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13695 definitions outside of __OPTIMIZE__ guarded section.
13698 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13699 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13700 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13701 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13704 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13706 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13707 ("vfenez<mode>"): Add missing constraints.
13709 2017-03-08 Martin Sebor <msebor@redhat.com>
13712 * config/nds32/nds32.c (nds32_option_override):
13713 Fix misspelled diagnostic.
13715 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13718 * gimplify.c (gimplify_omp_for): Replace index var in outer
13719 taskloop statement with an artificial variable and add
13720 OMP_CLAUSE_PRIVATE clause for it.
13722 2017-03-08 Richard Biener <rguenther@suse.de>
13724 PR tree-optimization/79955
13725 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13726 for accesses that are completely outside of the variable.
13728 2017-03-08 Andrew Haley <aph@redhat.com>
13730 PR tree-optimization/79943
13731 * tree-ssa-loop-split.c (compute_new_first_bound): When
13732 calculating the new upper bound, (END-BEG) should be added, not
13735 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13737 * config/avr/avr.md (setmemhi): Make sure match_dup
13738 operand number comes before match_scratch.
13740 2017-03-08 Richard Biener <rguenther@suse.de>
13742 PR tree-optimization/79920
13743 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13744 with ncopies == 1 to ...
13745 (vect_transform_slp_perm_load): ... here. Properly compute
13746 all element loads by iterating VF times over the group. Do
13747 not handle ncopies (computed in a broken way) in
13748 vect_create_mask_and_perm.
13750 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13753 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
13754 is a uniform vector, use uniform_vector_p return value instead of
13755 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
13757 2017-03-07 Marek Polacek <polacek@redhat.com>
13759 PR middle-end/79809
13760 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
13761 (alloca_call_type): Likewise.
13763 2017-03-07 Martin Liska <mliska@suse.cz>
13765 * gcov.c (process_args): Put comment to correct location.
13767 2017-03-07 Martin Liska <mliska@suse.cz>
13769 PR middle-end/68270
13770 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
13771 Use array_at_struct_end_p instead of DECL_CHAIN (field).
13772 (chkp_narrow_bounds_for_field): Likewise.
13773 (chkp_parse_array_and_component_ref): Pass one more argument to
13776 2017-03-07 Richard Biener <rguenther@suse.de>
13778 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
13781 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
13783 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
13784 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
13786 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13789 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
13790 to end of description.
13791 (PARAM_MAX_STORES_TO_MERGE): Likewise.
13793 2017-03-07 Jakub Jelinek <jakub@redhat.com>
13795 PR rtl-optimization/79901
13796 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
13798 (*avx512f_<code><mode>3<mask_name>): ... this.
13799 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
13800 iterator instead of VI8_AVX2_AVX512BW.
13802 PR rtl-optimization/79901
13803 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
13804 min/max expander, expand it using expand_vec_cond_expr.
13807 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
13810 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13813 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
13814 to void * for PCH reasons.
13815 * dwarf2out.c (output_loc_operands, output_die): Cast
13816 v.val_vec.array to unsigned char *.
13818 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
13821 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
13824 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
13826 PR rtl-optimization/79571
13827 * lra-constraints.c (process_alt_operands): Calculate static
13828 reject and subtract it from overall when only addresses will be
13831 2017-03-06 Julia Koval <julia.koval@intel.com>
13834 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
13835 incoming stack boundary to 128 for 64-bit targets.
13837 2017-03-06 Richard Biener <rguenther@suse.de>
13839 PR tree-optimization/79894
13840 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
13841 to NULL after folding it.
13843 2017-03-06 Richard Biener <rguenther@suse.de>
13845 PR tree-optimization/79824
13846 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
13847 check disabling peeling for gaps.
13849 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
13851 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13852 attributes): Document gettimeofday.
13854 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13856 * config/s390/s390.c (s390_option_override_internal): Set
13857 PARAM_MIN_VECT_LOOP_BOUND
13859 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13861 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
13862 * config/s390/s390.md: Likewise.
13864 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13867 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
13868 (<avx2_avx512>_perm<mode>): Rename to ...
13869 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
13871 (<avx512>_perm<mode>_mask): Rename to ...
13872 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
13874 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
13875 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
13876 instead of VI8F_256_512.
13877 (avx512f_perm<mode>): New define_expand.
13878 (avx512f_perm<mode>_mask): Likewise.
13879 (avx512f_perm<mode>_1<mask_name>): New define_insn.
13880 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
13882 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13884 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
13885 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
13887 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
13889 2017-03-06 Martin Liska <mliska@suse.cz>
13892 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
13893 when having a SSA NAME w/o VAR_DECL assigned to it.
13895 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13897 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
13898 msa_dpsub_<su>_d): Fix MODE for vec_select.
13900 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13902 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
13904 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
13906 2017-03-06 Richard Biener <rguenther@suse.de>
13908 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
13909 * plugin.c (register_plugin_info): Likewise.
13910 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
13912 2017-03-05 Jakub Jelinek <jakub@redhat.com>
13914 * config/i386/sse.md (sse_storehps, sse_storelps,
13915 avx_<castmode><avxsizesuffix>_<castmode>,
13916 avx512f_<castmode><avxsizesuffix>_<castmode>,
13917 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
13918 in condition that at least one operand is not a MEM.
13920 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13922 PR middle-end/79805
13923 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
13924 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
13926 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
13927 gimple_call_nothrow_p flag based on whether original builtin can throw.
13928 If it can, emit following stmts on the fallthrough edge.
13929 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
13930 don't create new bb if inserting just debug stmts on the edge, try to
13931 insert them on the fallthru bb or just reset debug stmts.
13933 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
13936 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
13937 restore recog_data (including the operand rtxes inside it) around
13938 the call to get_insn_template.
13940 2017-03-03 Martin Sebor <msebor@redhat.com>
13942 PR tree-optimization/79699
13943 * context.c (context::~context): Free MPFR caches to avoid
13944 a memory leak on program exit.
13946 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13948 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
13949 Use wide_int::ulow () instead of .elt (0).
13951 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13953 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
13954 (*pushxf): Limit oF constraint to 32bit targets and add oC
13955 constraint for 64bit targets.
13956 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
13957 (*pushdf): Change rmF constraint to rmC.
13959 2017-03-03 Martin Liska <mliska@suse.cz>
13961 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
13962 Remove unused variable.
13964 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13967 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
13968 is a memory operand, increase num_memory.
13969 (ix86_expand_args_builtin): Likewise.
13971 2017-03-03 Jan Hubicka <jh@suse.cz>
13974 * ipa-devirt.c (maybe_record_node): Properly handle
13975 __cxa_pure_virtual visibility.
13977 2017-03-03 Martin Liska <mliska@suse.cz>
13979 PR tree-optimization/79803
13980 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
13982 (pass_loop_prefetch::execute): Disabled optimization if an
13983 assumption about L1 cache size is not met.
13985 2017-03-03 Martin Liska <mliska@suse.cz>
13987 PR rtl-optimization/79574
13988 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
13989 (hash_scan_set): Likewise.
13990 (dump_hash_table): Likewise.
13991 (hoist_code): Likewise.
13993 2017-03-03 Richard Biener <rguenther@suse.de>
13995 * fixed-value.c (fixed_from_string): Restore use of elt (1)
13996 in place of uhigh ().
13997 (fixed_convert_from_real): Likewise.
13999 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14002 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14004 2017-03-03 Richard Biener <rguenther@suse.de>
14006 PR middle-end/79818
14007 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14008 TYPE_OVERFLOW_UNDEFINED check.
14010 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14012 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14014 (vector_ae_<mode>_p): Likewise.
14015 (vector_nez_<mode>_p): Likewise.
14016 (vector_ne_v2di_p): Likewise.
14017 (vector_ae_v2di_p): Likewise.
14018 (vector_ne_<mode>_p): Likewise.
14019 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14021 (vsx_tsqrt<mode>2_fe): Likewise.
14023 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14026 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14028 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14030 PR rtl-optimization/79780
14031 * cprop.c (one_cprop_pass): When second and further conditional trap
14032 in a single basic block is turned into an unconditional trap, turn it
14033 into a deleted note to avoid RTL verification failures.
14035 2017-03-02 Richard Biener <rguenther@suse.de>
14037 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14039 2017-03-02 Richard Biener <rguenther@suse.de>
14041 PR tree-optimization/79345
14043 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14044 param and abort the walk, returning -1 if it is hit.
14045 (walk_aliased_vdefs): Take a limit param and pass it on.
14046 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14047 defaulting to 0 and return a signed int.
14048 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14049 (check_defs): New helper.
14050 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14051 about uninitialized memory.
14052 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14053 bogus uninitialized warning.
14054 (fixed_convert_from_real): Likewise.
14056 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14058 PR tree-optimization/66768
14059 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14060 iv_use if base object can't be determined.
14062 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14064 PR tree-optimization/79345
14065 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14066 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14067 (get_pattern_stats): Initialize it.
14068 * genemit.c (gen_expand): Verify match_scratch numbers come after
14069 match_operand/match_dup numbers.
14070 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14071 match_scratch numbers.
14072 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14074 * config/s390/s390.md (trunctdsd2): Likewise.
14076 2017-03-02 Richard Biener <rguenther@suse.de>
14078 * wide-int.h (wide_int_storage::operator=): Implement in terms
14081 2017-03-02 Richard Biener <rguenther@suse.de>
14083 PR tree-optimization/79777
14084 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14085 the to insert expression to sth existing.
14087 2017-03-01 Martin Sebor <msebor@redhat.com>
14089 PR middle-end/79692
14090 * gimple-ssa-sprintf.c
14091 (directive::known_width_and_precision): New function.
14092 (format_integer): Use it.
14093 (get_mpfr_format_length): Consider the full range of precision
14094 when computing %g output with the # flag. Set the likely byte
14095 count to 3 rather than 1 when precision is indeterminate.
14096 (format_floating): Correct the lower bound of precision.
14098 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14100 * doc/invoke.texi: Document default code model for 64-bit Linux.
14102 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14105 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14106 udiv rather than div since input pattern is unsigned.
14108 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14110 * config/i386/i386.c (print_reg): Warn for values of
14111 unsupported size in integer register.
14113 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14116 * config/rs6000/predicates.md (current_file_function_operand): Do
14117 not allow self calls to be local if the function is replaceable.
14119 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14122 * config/rs6000/altivec.h (vec_ctz and others): Change the
14123 preprocessor macro that controls conditional compilation from
14124 _ARCH_PWR9 to __POWER9_VECTOR__.
14125 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14126 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14127 control (instead of _ARCH_PWR9 control) so that template
14128 definition uses power9-specific function.
14129 (vec_any_eq): Likewise.
14130 (vec_all_ne): Change macro definition to use a power9-specific
14131 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14132 _ARCH_PWR9 control).
14133 (vec_any_eq) Likewise.
14134 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14135 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14136 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14137 support for xvcmpnedp instruction.
14138 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14139 macro expansion so that Power9 implementation of vec_all_ne does
14140 not use the AltiVec predicate framework.
14141 (VCMPNEH_P): Likewise.
14142 (VCMPNEW_P): Likewise.
14143 (VCMPNED_P): Likewise.
14144 (VCMPNEFP_P): Likewise.
14145 (VCMPNEDP_P): Likewise.
14146 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14147 implementation of vec_any_eq to not use AltiVec predicate
14149 (VCMPAEH_P): Likewise.
14150 (VCMPAEW_P): Likewise.
14151 (VCMPAED_P): Likewise.
14152 (VCMPAEFP_P): Likewise.
14153 (VCMPAEDP_P): Likewise.
14154 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14155 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14156 not use the AltiVec predicate framework.
14157 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14158 of vec_any_eq to not use AltiVec predicate framework.
14159 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14160 support for predefined __POWER9_VECTOR__ macro to indicate that
14161 Power9 instruction selection is enabled.
14162 (altivec_overloaded_builtins): Remove extraneous
14163 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14164 function argument types RS6000_BTI_bool_V16QI and
14165 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14166 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14167 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14168 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14169 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14170 Power9 for implementations of vec_cmpne. Change the signature for
14171 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14172 (representing vec_all_ne) to remove the previously described first
14173 argument of type RS6000_BTI_INTSI, as this was an artifact of
14174 reliance on the AltiVec predicate framework, which is no longer
14175 used in the implementation of these functions. Add
14176 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14177 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14178 since, unlike the AltiVec predicate framework implementation, we
14179 do not share function descriptors between vec_alle and vec_anyeq.
14180 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14181 set of modes that receive special treatment even when
14182 TARGET_P9_VECTOR is true. The special treatment emits code that
14183 does not depend on Power9 instructions.
14184 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14185 define_expand to not rely on AltiVec predicate framework.
14186 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14188 (vector_ne_v2di_p): Change this define_expand to not rely on
14189 AltiVec predicate framework.
14190 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14192 (vector_ne_<mode>_p): Change this define_expand to not rely on
14193 AltiVec predicate framework.
14194 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14196 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14197 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14198 define_insn pattern.
14199 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14200 define_insn pattern because the xvcmpne<VSs>. instruction is not
14202 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14203 instruction is not supported.
14205 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14207 * config/nvptx/nvptx.c: Include intl.h.
14209 2017-03-01 Martin Jambor <mjambor@suse.cz>
14212 * ipa-prop.h (ipa_bits): Removed field known.
14213 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14214 to pointers. Adjusted their comments to warn about their sharing.
14215 (ipcp_transformation_summary): Change bits to a vector of pointers.
14216 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14217 (ipa_get_ipa_bits_for_value): Declare.
14218 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14219 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14220 (ipa_bits_hash_table): Likewise.
14221 (ipa_vr_ggc_hash_traits): Likewise.
14222 (ipa_vr_hash_table): Likewise.
14223 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14224 being pointers and vr_known being removed.
14225 (ipa_set_jf_unknown): Likewise.
14226 (ipa_get_ipa_bits_for_value): New function.
14227 (ipa_set_jfunc_bits): Likewise.
14228 (ipa_get_value_range): New overloaded functions.
14229 (ipa_set_jfunc_vr): Likewise.
14230 (ipa_compute_jump_functions_for_edge): Use the above functions to
14231 construct bits and vr parts of jump functions.
14232 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14233 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14235 (ipcp_grow_transformations_if_necessary): Also allocate
14236 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14238 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14239 them. Fix too long lines.
14240 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14241 vr_known being removed.
14242 (ipa_read_jump_function): Use new setter functions to construct bits
14243 and vr parts of jump functions or set them to NULL.
14244 (write_ipcp_transformation_info): Adjust for bits being pointers.
14245 (read_ipcp_transformation_info): Likewise.
14246 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14248 Include gt-ipa-prop.h.
14249 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14251 (ipcp_store_bits_results): Likewise.
14252 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14253 Do not write to existing jump functions but use a temporary instead.
14255 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14258 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14259 attempt to use its first operand as BIT_FIELD_REF base.
14261 2017-03-01 Richard Biener <rguenther@suse.de>
14263 PR middle-end/79721
14264 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14265 interpolating formula in wrapping arithmetic.
14266 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14268 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14270 PR tree-optimization/79734
14271 * tree-vect-generic.c (expand_vector_condition): Optimize
14272 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14273 Handle VEC_COND_EXPR where comparison has different inner width from
14274 type's inner width.
14276 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14278 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14279 markup, and similar issues. Remove @opindex entries for things
14280 that aren't options. Add missing -mmpy-option entries.
14282 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14284 PR tree-optimization/79737
14285 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14286 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14287 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14288 instead of byte_size. Formatting fix.
14289 (shift_bytes_in_array_right): Formatting fix.
14291 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14294 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14295 condition on optimize for the leaf function test.
14297 2017-02-28 Martin Liska <mliska@suse.cz>
14300 * read-rtl-function.c (function_reader::handle_unknown_directive):
14301 Bail out when one uses -flto.
14303 2017-02-28 Martin Liska <mliska@suse.cz>
14305 * common.opt: Replace space with tabular for options of <number>
14307 * config/i386/i386.opt: Show <number> value for
14308 -mlarge-data-threshold.
14309 * opts.c (print_filtered_help): Do not display number in hexadecimal
14312 2017-02-28 Martin Liska <mliska@suse.cz>
14314 * common.opt: Fix --help=option -Q for options which are of
14317 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14319 * config/i386/i386.c (print_reg): Error out for values
14320 of 8-bit size in invalid integer register.
14322 2017-02-28 Martin Sebor <msebor@redhat.com>
14324 PR tree-optimization/79691
14325 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14327 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14330 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14331 gcc_unreachable with output_operand_lossage.
14333 2017-02-28 Richard Biener <rguenther@suse.de>
14335 PR tree-optimization/79740
14336 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14338 (visit_nary_op): Insert the nary into the hashtable if we
14339 pattern-matched sth.
14340 * tree-ssa-pre.c (eliminate_insert): Robustify.
14342 2017-02-28 Richard Biener <rguenther@suse.de>
14344 PR middle-end/79731
14345 * fold-const.c (decode_field_reference): Reject out-of-bound
14348 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14350 * config/i386/i386.c: Include intl.h.
14351 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14352 instead of just cond ? "..." : "...".
14353 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14354 * coverage.c (read_counts_file): Likewise.
14355 * omp-offload.c: Include intl.h.
14356 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14357 of just cond ? "..." : "...".
14358 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14359 of just cond ? "..." : "...".
14361 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14364 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14366 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14367 'tune for' CPU name.
14368 * config/arm/arm-cpu-data.h: Regenerated.
14370 2017-02-28 Richard Biener <rguenther@suse.de>
14372 PR tree-optimization/79732
14373 * tree-inline.c (expand_call_inline): Do not shadow var.
14375 2017-02-28 Richard Biener <rguenther@suse.de>
14377 PR tree-optimization/79723
14378 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14379 address-space properly.
14381 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14383 * doc/optinfo.texi (Optimization groups): Fix option used for
14385 * doc/invoke.texi (-fopt-info): Document "omp".
14386 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14387 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14388 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14389 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14390 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14392 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14394 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14395 OPTGROUP_OMP with "omp".
14397 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14400 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14401 for arithmetic shift of unsigned V2DI.
14403 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14405 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14406 arc/linux.h headers.
14407 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14408 (LINK_SPEC): Likewise.
14409 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14410 (EXTRA_SPECS): Likewise.
14411 (STARTFILE_SPEC): Likewise.
14412 (ENDFILE_SPEC): Likewise.
14413 (LIB_SPEC): Likewise.
14414 (TARGET_SDATA_DEFAULT): Likewise.
14415 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14416 (MULTILIB_DEFAULTS): Likewise.
14417 (DWARF2_UNWIND_INFO): Likewise.
14418 * config/arc/big.h: New file.
14419 * config/arc/elf.h: Likewise.
14420 * config/arc/linux.h: Likewise.
14421 * config/arc/t-uClibc: Remove.
14423 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14425 PR tree-optimization/77536
14426 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14427 (tree_transform_and_unroll_loop): Use above function to compute the
14428 estimated niter of unrolled loop and use it when scaling profile.
14429 Also use count info rather than frequency if it's non-zero.
14430 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14431 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14432 (vect_transform_loop): Call above function.
14434 2017-02-27 Richard Biener <rguenther@suse.de>
14436 PR tree-optimization/45397
14437 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14438 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14439 (visit_nary_op): Add pattern matching for CSEing sign-changed
14440 or truncated operations with wider ones.
14442 2017-02-27 Richard Biener <rguenther@suse.de>
14444 PR tree-optimization/79690
14445 * tree-vect-stmts.c (vectorizable_store): Use vector type
14446 built from the DR with address-space.
14448 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14450 * doc/invoke.texi (Optimize Options): Refine the description
14451 of asan-use-after-return.
14453 2017-02-25 Alan Modra <amodra@gmail.com>
14455 PR rtl-optimization/79584
14456 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14457 base, not ad->base_term, the reg within base. Remove assertion
14458 that ad->base == ad->base_term. Replace gen_int_mode using
14459 bogus mode with const0_rtx.
14461 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14463 PR middle-end/79396
14464 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14465 FMA_EXPR like tcc_binary or tcc_unary.
14467 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14470 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14472 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14473 (output_loc_operands): Handle DW_OP_call_ref and
14474 DW_OP_GNU_variable_value.
14475 (struct variable_value_struct): New type.
14476 (struct variable_value_hasher): Likewise.
14477 (variable_value_hash): New variable.
14478 (string_types): Remove.
14479 (copy_loc_descr): New function.
14480 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14481 (prepend_loc_descr_to_each): New function.
14482 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14483 instead of add_loc_descr_to_each if the first argument is single
14484 location list and the second has multiple.
14485 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14486 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14487 when looking for variable value which doesn't have other location info.
14488 (loc_list_from_tree): Formatting fix.
14489 (gen_array_type_die): Simplify DW_AT_string_length handling.
14490 (adjust_string_types): Remove.
14491 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14492 string_types. Call resolve_variable_values.
14493 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14494 (resolve_addr_in_expr): Likewise. Add A argument.
14495 (copy_deref_exprloc): Remove deref argument. Adjust for the
14496 original expression being DW_OP_GNU_variable_value with optionally
14497 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14498 optionally after it.
14499 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14500 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14501 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14502 (variable_value_hasher::hash, variable_value_hasher::equal): New
14504 (resolve_variable_value_in_expr, resolve_variable_value,
14505 resolve_variable_values, note_variable_value_in_expr,
14506 note_variable_value): New functions.
14507 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14510 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14513 * opts.h (handle_generated_option): Add GENERATED_P argument.
14514 * opts-common.c (handle_option): Adjust function comment.
14515 (handle_generated_option): Add GENERATED_P argument, pass it to
14517 (control_warning_option): Pass false to handle_generated_option
14519 * opts.c (maybe_default_option): Pass true to handle_generated_option
14521 * optc-gen.awk: Likewise.
14523 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14525 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14526 a REG, look at the REG it is a SUBREG of.
14527 (splitter for cmpeqsi_t): Ditto.
14529 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14531 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14532 the special USEs with the pattern of the insn, not the insn itself.
14534 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14537 * doc/invoke.texi: Document -mload-store-pairs.
14539 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14540 Sandra Loosemore <sandra@codesourcery.com>
14542 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14543 argument isn't a CONST_INT.
14544 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14545 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14546 (nios2_validate_compare): Bypass alternate compare logic if *op2
14547 is not a CONST_INT.
14548 (ldstwm_operation_p): Return false if first_base is not a REG or
14549 if first_offset is not a CONST_INT.
14551 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14553 * config/cris/cris.md: Use correct operand in a define_peephole2.
14555 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14557 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14559 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14561 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14562 this_insn if it is an INSN or JUMP_INSN.
14563 (force_offsettable): Look at base, not at addr.
14564 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14565 on things that aren't necessarily CONST_INTs.
14567 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14569 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14570 -mfpmath=sse is the default also for x86-32 targets with SSE2
14571 instruction set when @option{-ffast-math} is enabled
14573 2017-02-24 Jeff Law <law@redhat.com>
14575 PR rtl-optimizatoin/79286
14576 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14579 2017-02-24 Richard Biener <rguenther@suse.de>
14581 PR tree-optimization/79389
14582 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14585 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14587 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14588 function comment to reflect reality.
14589 (loop_exits_before_overflow): Fix typo in function description.
14591 2017-02-24 Richard Biener <rguenther@suse.de>
14593 PR tree-optimization/79389
14594 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14595 properly that a threading opportunity exists. Detect conditional
14596 copy/constant propagation opportunities.
14598 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14600 * config/visium/visium.md (type): Add trap.
14601 (b): New mode attribute.
14602 (*btst): Rename into...
14603 (*btst<mode>): ...this and adjust.
14604 (*cbranchsi4_btst_insn): Rename into...
14605 (*cbranch<mode>4_btst_insn): ...this and adjust.
14606 (trap): New define_insn.
14608 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14610 PR tree-optimization/79389
14611 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14612 (noce_reversed_cond_code): New function.
14613 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14614 reversed_comparison_code. Formatting fix.
14615 (noce_try_store_flag): Test rev_cond != NULL in addition to
14616 reversed_comparison_code.
14617 (noce_try_store_flag_constants): Likewise.
14618 (noce_try_store_flag_mask): Likewise.
14619 (noce_try_addcc): Use rev_cond if non-NULL instead of
14620 reversed_comparison_code.
14621 (noce_try_cmove_arith): Likewise. Formatting fixes.
14622 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14623 (noce_find_if_block): Initialize rev_cond.
14624 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14625 instead of false as last argument never attempt to reverse it
14628 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14630 PR tree-optimization/79663
14631 * tree-predcom.c (combine_chains): Process refs in reverse order
14632 only for ZERO length chains, and add explaining comment.
14634 2017-02-23 Jeff Law <law@redhat.com>
14636 PR tree-optimization/79578
14637 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14638 in call to operand_equal_p.
14640 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14643 * config/i386/cpuid.h: Fix another undefined behavior.
14645 2017-02-23 Richard Biener <rguenther@suse.de>
14647 PR tree-optimization/79683
14648 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14649 vector types for data-refs.
14651 2017-02-23 Martin Liska <mliska@suse.cz>
14653 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14655 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14657 PR middle-end/79665
14658 * internal-fn.c (get_range_pos_neg): Moved to ...
14659 * tree.c (get_range_pos_neg): ... here. No longer static.
14660 * tree.h (get_range_pos_neg): New prototype.
14661 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14662 are known to be in between 0 and signed maximum inclusive, try to
14663 expand both unsigned and signed divmod and use the cheaper one from
14666 2017-02-22 Jeff Law <law@redhat.com>
14668 PR tree-optimization/79578
14669 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14670 to compare base operands.
14672 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14675 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14676 gpc_reg_operand instead of fpr_reg_operand.
14678 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14680 * config/mips/mips.c (mips_return_in_memory): Force FP
14681 vector types to be returned in memory for o32 ABI.
14683 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14685 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14686 instead of DW_TAG_member for static data member declarations and don't
14687 set no_linkage_name for static inline data members.
14688 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14691 2017-02-22 Martin Liska <mliska@suse.cz>
14693 * doc/invoke.texi: Replace inequality signs with square brackets
14696 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14698 PR tree-optimization/68644
14699 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14701 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14704 * lra-constraints.c (simplify_operand_subreg): Handle
14705 WORD_REGISTER_OPERATIONS targets.
14707 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14710 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14711 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14712 elimination by swapping fld*.
14714 2017-02-22 Richard Biener <rguenther@suse.de>
14716 PR tree-optimization/79673
14717 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14718 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14719 irrelevant address-space qualifiers and avoiding a
14720 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14722 2017-02-22 Richard Biener <rguenther@suse.de>
14724 PR tree-optimization/79666
14725 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14726 to not symbolically negate if that may introduce undefined
14729 2017-02-22 Martin Liska <mliska@suse.cz>
14732 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14733 * data-streamer-out.c (streamer_write_gcov_count_stream):
14735 * value-prof.c (stream_out_histogram_value): Make assert more
14736 precise based on type of counter.
14738 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14741 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14742 Use nonimmediate_operand instead of memory_operand for operand 1.
14743 (float-extend standard_x87sse_constant_load splitter): Ditto.
14745 2017-02-21 Jeff Law <law@redhat.com>
14747 PR tree-optimization/79621
14748 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
14749 blocks with edges to themselves.
14751 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14754 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
14755 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
14756 Use gimple_call_builtin_p.
14759 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
14760 on temporarily removed DEBUG_INSNs.
14762 PR tree-optimization/79649
14763 * tree-loop-distribution.c (classify_partition): Give up on
14764 non-generic address space loads/stores.
14766 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
14768 * doc/loop.texi (Loop manipulation): Remove nonexistent
14769 tree_ssa_loop_version from the documentation.
14770 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
14772 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14775 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
14776 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
14777 * config/rs6000/rs6000.c: Include except.h.
14778 (rs6000_expand_split_stack_prologue): Call
14779 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
14781 2017-02-21 Martin Jambor <mjambor@suse.cz>
14784 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
14785 have been analyzed.
14787 2017-02-21 Martin Jambor <mjambor@suse.cz>
14789 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
14790 for backward compatibility only.
14791 * doc/invoke.texi (Option Summary): Remove all references to
14792 -fipa-cp-alignment.
14794 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
14798 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14800 * lra-constraints.c (curr_insn_transform): Handle
14801 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14803 2017-02-21 Martin Liska <mliska@suse.cz>
14805 * config/i386/i386.opt: Replace -masm-dialect with -masm.
14807 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
14809 PR translation/79638
14810 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
14812 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
14815 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
14816 (arm_function_ok_for_sibcall): Return false for an indirect call by
14817 descriptor if all the argument registers are used.
14818 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
14819 alignment of the function.
14821 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14823 PR tree-optimization/61441
14824 * simplify-rtx.c (simplify_const_unary_operation): For
14825 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
14826 the sNaN unmodified.
14828 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14830 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14831 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14832 instead of SYSTEM_HEADER_DIR.
14834 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
14835 Martin Liška <mliska@suse.cz>
14837 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
14838 Fix typos and grammar, use active voice, and clarify.
14840 2017-02-20 Marek Polacek <polacek@redhat.com>
14842 PR middle-end/79537
14843 * gimplify.c (gimplify_expr): Handle unused *&&L;.
14846 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
14848 2017-02-20 Jakub Jelinek <jakub@redhat.com>
14851 * config/i386/i386.c (ix86_expand_builtin): Handle
14852 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
14853 ix86_builtins_isa[fcode].isa as a requirement of those
14854 flags and any other flag in the bitmask.
14855 (ix86_init_mmx_sse_builtins): Use 0 instead of
14856 ~OPTION_MASK_ISA_64BIT as mask.
14857 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
14858 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
14859 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
14860 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
14862 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14865 * lra-constraints.c (split_reg): Check requested split mode
14866 is supported by the register.
14868 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14870 * lra-constraints.c (simplify_operand_subreg): Remove early
14873 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14876 * lra-constraints.c (curr_insn_transform): Tighten condition
14877 for converting SUBREG reloads from OP_OUT to OP_INOUT.
14879 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14882 * lra-constraints.c (curr_insn_transform): Handle
14883 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14885 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
14888 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
14890 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
14892 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
14895 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14898 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14900 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
14903 2017-02-18 Jakub Jelinek <jakub@redhat.com>
14905 * final.c (last_columnnum, override_columnnum): New variables.
14906 (final_start_function): Set last_columnnum, pass it to begin_prologue
14907 hook and pass 0 to dwarf2out_begin_prologue.
14908 (final_scan_insn): Update override_columnnum. Pass last_columnnum
14909 to source_line debug hook.
14910 (notice_source_line): Compute last_columnnum and for debug_column_info
14911 return true on column changes.
14912 * debug.h (struct gcc_debug_hooks): Add column argument to
14913 source_line and begin_prologue hooks.
14914 (debug_nothing_int_charstar_int_bool): Remove prototype.
14915 (debug_nothing_int_int_charstar,
14916 debug_nothing_int_int_charstar_int_bool): New prototypes.
14917 (dwarf2out_begin_prologue): Add column argument.
14918 * debug.c (do_nothing_debug_hooks): Adjust source_line and
14919 begin_prologue hooks.
14920 (debug_nothing_int_charstar_int_bool): Remove.
14921 (debug_nothing_int_int_charstar,
14922 debug_nothing_int_int_charstar_int_bool): New functions.
14923 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
14924 through to dwarf2out_source_line.
14925 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
14926 (dwarf2out_source_line): Add column argument, emit it if requested.
14927 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
14929 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14930 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14931 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
14932 through to dwarf2out_begin_prologue.
14933 (vmsdbgout_source_line): Add column argument, pass it through to
14934 dwarf2out_source_line.
14935 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
14936 dbxout_source_line caller.
14937 (dbxout_source_line): Add column argument.
14939 * common.opt (gno-column-info, gcolumn-info): New options.
14940 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
14941 (check_die): Also test for multiple DW_AT_decl_column attributes.
14942 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
14943 DW_AT_decl_column if requested.
14944 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
14946 (gen_variable_die): Likewise.
14947 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
14948 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
14951 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
14952 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
14953 (ix86_handle_option): Handle OPT_m3dnowa.
14954 * doc/invoke.texi (-m3dnowa): Document.
14955 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
14956 -m3dnowa instead of -m3dnow -march=athlon.
14959 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
14960 instead of gcc_assert for K, r and R code checks. Formatting fixes.
14962 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14965 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
14966 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
14967 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
14968 generator for vsx_xxpermdi_<mode>_be.
14969 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
14970 force big-endian semantics.
14971 (vsx_xxpermdi_<mode>_be): New define_expand with same
14972 implementation as previous version of vsx_xxpermdi_<mode>.
14974 2017-02-17 Jakub Jelinek <jakub@redhat.com>
14976 PR tree-optimization/79327
14977 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
14978 variable, its initialization and use.
14980 2017-02-17 Julia Koval <julia.koval@intel.com>
14982 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
14983 (OPTION_MASK_ISA_PKU_UNSET): New.
14984 (ix86_handle_option): Handle -mrdpid.
14985 * config/i386/cpuid.h (bit_RDPID): New.
14986 * config/i386/driver-i386.c (host_detect_local_cpu):
14987 Detect RDPID feature.
14988 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
14989 * config/i386/i386-c.c (ix86_target_macros_internal):
14991 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
14992 (ix86_valid_target_attribute_inner_p): Add "rdpid".
14993 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14994 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
14995 * config/i386/i386.md (define_insn "rdpid"): New.
14996 * config/i386/i386.opt Add -mrdpid.
14997 * config/i386/immintrin.h (_rdpid_u32): New.
14999 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15001 PR rtl-optimization/79541
15002 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15003 instead of transforming it into USE.
15005 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15007 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15008 If HONOR_SNANS (SFmode) force the input to a register.
15009 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15010 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15011 an frsp or similar insn.
15013 2017-02-17 Martin Liska <mliska@suse.cz>
15015 PR rtl-optimization/79577
15016 * params.def (selsched-max-sched-times): Increase minimum to 1.
15018 2017-02-17 Martin Liska <mliska@suse.cz>
15020 PR rtl-optimization/79574
15021 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15023 2017-02-17 Martin Liska <mliska@suse.cz>
15025 PR tree-optimization/79529
15026 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15027 ssa_defined_default_def_p to handle cases which are implicitly
15029 * tree-ssa.c (ssa_defined_default_def_p): New function.
15030 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15031 which are implicitly defined.
15032 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15034 2017-02-17 Richard Biener <rguenther@suse.de>
15036 PR middle-end/79576
15037 * params.def (max-ssa-name-query-depth): Limit to 10.
15039 2017-02-17 Richard Biener <rguenther@suse.de>
15041 PR tree-optimization/79552
15042 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15045 2017-02-17 Richard Biener <rguenther@suse.de>
15048 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15050 2017-02-17 Marek Polacek <polacek@redhat.com>
15052 PR middle-end/79536
15053 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15054 (fold_negate_expr): New wrapper.
15056 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15058 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15059 Correct terminology and de-emphasize pre-standard behavior.
15061 2017-02-16 Alan Modra <amodra@gmail.com>
15063 PR rtl-optimization/79286
15064 * ira.c (def_dominates_uses): New function.
15065 (update_equiv_regs): Don't create an equivalence for insns that
15066 may trap where the register def does not dominate the use.
15068 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15070 PR rtl-optimization/78127
15071 * lra.c (lra): Call lra_eliminate before finish the loop after
15074 2017-02-16 Richard Biener <rguenther@suse.de>
15076 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15078 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15079 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15080 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15081 (isl_val_int_from_wi): New function.
15082 (extract_affine_gmp): Rename to ...
15083 (extract_affine_wi): ... this, take a widest_int.
15084 (extract_affine_int): Just wrap extract_affine_wi.
15085 (add_param_constraints): Use isl_val_int_from_wi.
15086 (add_loop_constraints): Likewise, and extract_affine_wi.
15088 2017-02-15 Jeff Law <law@redhat.com>
15090 PR middle-end/79521
15091 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15092 ira_init_register_move_cost_if_necessary.
15094 2017-02-15 Martin Sebor <msebor@redhat.com>
15096 PR middle-end/32003
15097 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15098 removed in a prior commit.
15100 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15102 PR tree-optimization/79347
15103 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15104 counters during peeling.
15106 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15108 * Makefile.in (site.exp): Remove "set ISLVER".
15110 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15113 * real.c (real_from_integer): Call real_convert even for decimal.
15115 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15118 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15120 2017-02-14 Andrew Pinski <apinski@cavium.com>
15122 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15123 cores and change the partno/implementer to be correct.
15124 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15125 the 'B" as the implementer.
15126 * config/aarch64/aarch64-tune.md: Regenerate.
15128 2017-02-14 Carl Love <cel@us.ibm.com>
15130 * config/rs6000/rs6000.c: Add case statement entry to make the
15131 xvcvuxdsp built-in argument unsigned.
15132 * config/rs6000/vsx.md: Fix the source and return operand types so they
15133 match the instruction definitions from the ISA document. Fix typo
15134 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15137 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15140 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15141 member early_clobber_alts.
15142 * lra-lives.c (reg_early_clobber_p): New.
15143 (process_bb_lives): Use it.
15144 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15145 (debug_operand_data): Initialize early_clobber_alts.
15146 (setup_operand_alternative): Set up early_clobber_alts.
15147 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15148 alternatives to new_insn_reg.
15149 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15151 (lra_update_insn_regno_info): Pass the new arg.
15153 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15155 PR middle-end/79505
15156 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15157 (new_oacc_loop_raw): Don't clear already cleared fields.
15160 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15161 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15162 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15164 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15167 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15169 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15172 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15173 the extra instruction to the right place to store 128-bit constant
15176 2017-02-14 Martin Sebor <msebor@redhat.com>
15178 PR middle-end/79448
15179 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15180 warning for strings of unknown length.
15182 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15184 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15186 2017-02-14 Jeff Law <law@redhat.com>
15189 * ira-costs.c (scan_one_insn): Initialize register move costs
15190 for pseudos seen in USE/CLOBBER insns.
15192 PR tree-optimization/79095
15193 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15194 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15195 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15196 if the operands are known to be not equal, then the resulting range
15198 (intersect_ranges): If the new range is ~[0,0] and the old range is
15199 wide, then prefer ~[0,0].
15200 * tree-vrp.c (overflow_comparison_p_1): New function.
15201 (overflow_comparison_p): New function.
15202 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15203 if NAME is used in an overflow test.
15204 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15205 overflow check that can be expressed as an equality test, then adjust
15206 ops to be that equality test.
15208 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15210 * config/s390/s390-builtin-types.def: Remove flags argument.
15211 * config/s390/s390.c (s390_init_builtins): Likewise.
15213 2017-02-14 Martin Liska <mliska@suse.cz>
15215 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15216 vector. Fix trailing white spaces.
15218 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15220 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15223 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15225 PR rtl-optimization/68664
15226 * config/arm/arm.c (arm_sched_can_speculate_insn):
15227 New function. Declare prototype.
15228 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15230 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15232 PR rtl-optimization/68664
15233 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15235 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15237 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15239 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15240 max skip bytes for function, loop and jump.
15242 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15244 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15245 ABS_EXPR for gimple dump.
15247 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15250 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15252 PR tree-optimization/79408
15253 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15254 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15255 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15256 also if rhs1 is INTEGER_CST.
15258 2017-02-14 Richard Biener <rguenther@suse.de>
15260 PR middle-end/79432
15261 * tree-into-ssa.c (insert_phi_nodes): When the function can
15262 have abnormal edges rewrite SSA names with broken use-def
15263 dominance out of SSA and register them for PHI insertion.
15265 2017-02-13 Martin Sebor <msebor@redhat.com>
15267 PR middle-end/79496
15268 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15269 clearing info.nowrite flag when snprintf size argument is a range.
15271 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15273 * cprop.c (cprop_jump): Add missing space in string literal.
15274 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15275 (get_constraint_for_component_ref): Likewise.
15276 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15277 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15278 * lra-constraints.c (process_alt_operands): Likewise.
15279 * ipa-inline.c (inline_small_functions): Likewise.
15280 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15281 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15282 * trans-mem.c (diagnose_tm_1_op): Likewise.
15283 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15284 (grid_parallel_clauses_gridifiable): Likewise.
15286 * config/nvptx/mkoffload.c (process): Add space in between
15289 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15290 "MOD4_SSE_REGS" and "ALL_REGS".
15292 * spellcheck.c (test_data): Add , in between "foo" and "food".
15294 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15297 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15298 boundary crossing check and subsequent code generation agree.
15300 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15302 * config/aarch64/aarch64.c (has_memory_op): Delete.
15303 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15306 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15308 PR rtl-optimization/79388
15309 PR rtl-optimization/79450
15310 * combine.c (distribute_notes): When removing TEM_INSN for which
15311 corresponding dest has last value recorded, invalidate that last
15314 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15316 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15317 of explicit '@'. Add missing assembly comment marker on branch costs
15320 2017-02-13 Nathan Sidwell <nathan@acm.org>
15322 * gengtype-lex.l (<in_struct>): Add '/'.
15324 2017-02-13 Martin Liska <mliska@suse.cz>
15327 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15329 2017-02-13 Richard Biener <rguenther@suse.de>
15331 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15333 * configure: Re-generate.
15334 * config.in: Likewise.
15335 * graphite-dependences.c: Simplify as if
15336 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15337 * graphite-isl-ast-to-gimple.c: Likewise.
15338 * graphite-optimize-isl.c: Likewise.
15339 * graphite-poly.c: Likewise.
15340 * graphite-sese-to-poly.c: Likewise.
15341 * graphite.h: Likewise.
15342 * toplev.c: Include isl/version.h and use isl_version () for
15343 printing the ISL version.
15344 * doc/install.texi: Update ISL requirement.
15346 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15348 * doc/standards.texi (Standards): Update reference to
15351 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15353 * doc/extend.texi (Named Address Spaces): sourceware.org now
15355 * doc/install.texi (Binaries): Ditto.
15358 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15360 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15361 standard terminology "stringize"/"stringizing" throughout.
15362 * doc/cppinternals.texi: Likewise.
15364 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15366 * doc/extend.texi: Fix some spelling mistakes and typos.
15367 * doc/invoke.texi: Likewise.
15369 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15372 * params.def (inline-min-speedup) Change from 10 to 8.
15374 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15376 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15379 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15382 * ipa-inline-analysis.c (get_minimal_bb): New function.
15383 (record_modified): Use it.
15384 (remap_edge_change_prob): Handle also ancestor functions.
15386 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15388 * doc/contrib.texi (Contributors): Remove broken link into
15389 the Mauve CVS repository.
15391 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15393 PR middle-end/79454
15394 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15395 result computation whenever lhs doesn't have vector mode, not
15396 just when it has BLKmode.
15398 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15400 * doc/makefile.texi (profiledbootstrap): Refer to the
15401 installation instructions only in textual form.
15403 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15406 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15408 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15410 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15411 (Specific): Update mingw-w64 reference.
15413 (Specific): Remove broken link to Renesas RX processor.
15415 2017-02-10 Richard Biener <rguenther@suse.de>
15417 * toplev.c (process_options): Do not mention obsolete graphite
15418 options when printing sorry message about missing graphite support.
15419 Mention -floop-nest-optimize.
15421 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15423 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15424 (vtst_p16): Likewise.
15425 (vtstq_p8): Likewise.
15426 (vtstq_p16): Likewise.
15428 (vtstq_p64): Likewise.
15429 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15430 (vset_lane_p64): New.
15431 (vsetq_lane_p64): New.
15433 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15435 PR tree-optimization/79411
15436 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15437 stmt operands are SSA_NAMEs used in abnormal phis.
15438 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15441 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15444 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15447 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15449 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15451 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15453 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15457 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15458 not arbitrary TREE_CONSTANT.
15461 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15462 "omp declare target link" attribute unless is_global_var.
15463 * omp-offload.c (find_link_var_op): Likewise.
15465 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15466 Chung-Lin Tang <cltang@codesourcery.com>
15468 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15470 (gimplify_adjust_omp_clauses): Don't delete TILE.
15471 (gimplify_omp_for): Deal with TILE.
15472 * internal-fn.c (expand_GOACC_TILE): New function.
15473 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15475 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15476 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15478 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15479 avoid DIV for outermost collapse var.
15480 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15481 Remove out of date comments, fix whitespace.
15482 * omp-general.c (omp_extract_for_data): Deal with tiling.
15483 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15484 adjust OLF_DIM_BASE value.
15485 (struct omp_for_data): Add tiling field.
15486 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15487 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15488 for auto loops. Remove default auto determining, moved to
15489 oacc_loop_fixed_partitions.
15490 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15491 stmts, add e_mask field.
15492 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15493 (oacc_thread_numbers): Use oacc_dim_call.
15494 (oacc_xform_tile): New.
15495 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15496 (finish_oacc_loop): Adjust for ifns vector.
15497 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15498 add case for GOACC_TILE fns.
15499 (oacc_loop_xform_loop): Delete.
15500 (oacc_loop_process): Iterate over call list directly, and add
15501 handling for GOACC_TILE fns.
15502 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15504 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15505 vector partitioning to outer loops. Assign 2 partitions to loops
15506 when available. Add TILE handling.
15507 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15508 (execite_oacc_device_lower): Process GOACC_TILE fns,
15509 ignore unknown specs.
15510 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15511 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15512 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15514 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15516 * configure.ac (ACX_BUGURL): Update.
15517 * configure: Regenerate.
15519 2017-02-09 Richard Biener <rguenther@suse.de>
15521 PR tree-optimization/69823
15522 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15523 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15525 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15527 * config/arc/arc-c.def: Add __NPS400__ definition.
15528 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15529 (TARGET_NPS400): Define.
15531 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15533 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15535 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15536 pointer, arch_info.
15537 (arc_cpu_types): Fill the arch_info field with a pointer into the
15538 arc_arch_types table.
15539 (arc_selected_cpu): Declare.
15540 * config/arc/arc.c (arc_selected_cpu): Make global.
15541 (arc_selected_arch): Delete.
15542 (arc_base_cpu): Delete.
15543 (arc_override_options): Remove references to deleted variables,
15544 update access to arch information.
15545 (ARC_OPT): Update access to arch information.
15546 (ARC_OPTX): Likewise.
15547 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15548 (TARGET_ARC600): Update access to arch information.
15549 (TARGET_ARC601): Likewise.
15550 (TARGET_ARC700): Likewise.
15551 (TARGET_EM): Likewise.
15552 (TARGET_HS): Likewise.
15553 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15556 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15559 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15560 condition/operands for integer GE/LE/GEU/LEU operations.
15562 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15564 PR translation/79397
15565 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15568 2017-02-08 Martin Jambor <mjambor@suse.cz>
15571 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15572 whether allocation happened.
15573 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15574 nothing was allocated.
15576 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15578 PR tree-optimization/79408
15579 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15580 constant, but SSA_NAME with a known integer range, use the minimum
15581 of that range instead of op1 to determine if modulo can be replaced
15582 with its first operand.
15584 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15586 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15588 2017-02-08 Richard Biener <rguenther@suse.de>
15590 PR tree-optimization/71824
15591 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15592 Check all loops contained in the merged region.
15594 2017-02-07 Andrew Pinski <apinski@cavium.com>
15596 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15598 2017-02-07 Andrew Pinski <apinski@cavium.com>
15600 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15601 (thunderxt88): Likewise.
15602 (thunderxt81): Disable LSE and change v8.1 to v8.
15603 (thunderxt83): Likewise.
15605 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15606 Richard Biener <rguenther@suse.de>
15608 PR middle-end/79399
15609 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15610 type from int to size_t.
15611 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15613 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15615 PR rtl-optimization/79386
15616 * cprop.c (bypass_conditional_jumps): Initialize
15617 bypass_last_basic_block already before splitting bbs after
15618 unconditional traps...
15619 (bypass_conditional_jumps): ... rather than here.
15622 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15623 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15624 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15625 fix -masm=intel patterns.
15627 2017-02-07 Richard Biener <rguenther@suse.de>
15629 PR tree-optimization/79256
15630 PR middle-end/79278
15631 * builtins.c (get_object_alignment_2): Use min_align_of_type
15632 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15633 and ADJUST_FIELD_ALIGN.
15635 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15637 * doc/tm.texi: Regenerate.
15638 * stor-layout.c (layout_decl): Adjust.
15639 (update_alignment_for_field): Likewise.
15640 (place_field): Likewise.
15641 (min_align_of_type): Likewise.
15642 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15643 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15644 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15645 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15646 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15647 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15648 * config/i386/i386.c (x86_field_alignment): Likewise.
15649 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15650 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15651 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15652 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15653 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15654 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15658 2017-01-30 Richard Biener <rguenther@suse.de>
15660 PR tree-optimization/79256
15661 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15662 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15665 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15667 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15668 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15669 builtins to SImode and emit a zero-extend, if necessary.
15671 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15673 * docs/invoke.texi (RISC-V Options): Alphabetize.
15675 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15677 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15680 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15682 * config/riscv/riscv.c: New file.
15683 * common/config/riscv/riscv-common.c: Likewise.
15684 * config.gcc: Likewise.
15685 * config/riscv/constraints.md: Likewise.
15686 * config/riscv/elf.h: Likewise.
15687 * config/riscv/generic.md: Likewise.
15688 * config/riscv/linux.h: Likewise.
15689 * config/riscv/multilib-generator: Likewise.
15690 * config/riscv/peephole.md: Likewise.
15691 * config/riscv/pic.md: Likewise.
15692 * config/riscv/predicates.md: Likewise.
15693 * config/riscv/riscv-builtins.c: Likewise.
15694 * config/riscv/riscv-c.c: Likewise.
15695 * config/riscv/riscv-ftypes.def: Likewise.
15696 * config/riscv/riscv-modes.def: Likewise.
15697 * config/riscv/riscv-opts.h: Likewise.
15698 * config/riscv/riscv-protos.h: Likewise.
15699 * config/riscv/riscv.h: Likewise.
15700 * config/riscv/riscv.md: Likewise.
15701 * config/riscv/riscv.opt: Likewise.
15702 * config/riscv/sync.md: Likewise.
15703 * config/riscv/t-elf-multilib: Likewise.
15704 * config/riscv/t-linux: Likewise.
15705 * config/riscv/t-linux-multilib: Likewise.
15706 * config/riscv/t-riscv: Likewise.
15707 * configure.ac: Likewise.
15708 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15709 Waterman as RISC-V maintainers.
15710 * doc/install.texi: Add RISC-V entries.
15711 * doc/invoke.texi: Add RISC-V options section.
15712 * doc/md.texi: Add RISC-V constraints section.
15713 * configure: Regenerated.
15715 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15718 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15719 false values to be constant vectors with all 0 or all 1 bits set.
15720 (vcondu<mode><mode>): Likewise.
15721 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15723 (fpmask_comparison_operator): Update comment.
15724 (vecint_comparison_operator): New predicate.
15725 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15726 vector conditionals when the true and false values are constant
15727 vectors with all 0 bits or all 1 bits set.
15729 2017-02-06 Martin Sebor <msebor@redhat.com>
15731 PR tree-optimization/79376
15732 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15734 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15736 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15737 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15738 to simplify split condition.
15740 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15742 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15743 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15746 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
15748 PR rtl-optimization/68664
15749 * target.def (can_speculate_insn): New hook.
15750 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
15751 * doc/tm.texi: Regenerate.
15752 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
15753 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
15754 (rs6000_sched_can_speculate_insn): New function.
15756 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15758 PR tree-optimization/79284
15759 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
15760 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
15761 vectorizable_mask_load_store, vectorizable_operation,
15762 vect_is_simple_cond, get_same_sized_vectype): Use it instead
15763 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
15764 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
15765 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
15766 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15767 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15768 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
15769 is_gimple_assign (stmt). Replace another such test with
15770 is_gimple_assign (stmt).
15772 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15775 * config/avr/avr.c (rtl-iter.h): Include it.
15776 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
15777 (avr_legitimate_combined_insn): ...and implementation.
15779 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
15781 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
15782 * config/s390/s390.c (s390_const_operand_ok)
15783 (s390_canonicalize_comparison, s390_extract_part)
15784 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
15785 (s390_contiguous_bitmask_p, s390_rtx_costs)
15786 (legitimize_pic_address): Likewise.
15787 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
15788 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
15789 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
15790 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
15791 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
15793 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15795 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
15796 REGNO($0) == REGNO($1).
15798 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15800 * config/s390/linux.h(SIZE_TYPE): Add comment.
15802 2017-02-06 Julian Brown <julian@codesourcery.com>
15803 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15804 Virendra Pathak <virendra.pathak@broadcom.com>
15806 * config/aarch64/aarch64-cores.def: Change the scheduler
15808 * config/aarch64/aarch64.md: Include thunderx2t99.md.
15809 * config/aarch64/thunderx2t99.md: New file.
15811 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15813 * doc/standards.texi (Go Language): Update link to language
15816 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15818 * tree-eh.c (lower_resx): Sanitize profile.
15819 (cleanup_empty_eh_move_lp): Likewise.
15821 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15824 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
15826 * cfgloopmanip.h (loop_version): Update prototype.
15827 * modulo-sched.c (sms_schedule): Update call of loop_version.
15828 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
15829 * tree-parloops.c (gen_parallel_loop): Likewise.
15830 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
15831 * tree-ssa-loop-split.c (split_loop): Likewise.
15832 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
15833 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15835 2017-02-05 Martin Liska <mliska@suse.cz>
15838 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
15840 (print_operand_address): Initialize a struct to zero.
15842 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15844 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
15845 garbage collector only in textual form.
15847 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15849 * doc/extend.texi (x86 specific memory model extensions for
15850 transactional memory): Simplify a phrase.
15852 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
15855 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
15856 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
15857 (atomic_storedi_1): Likewise.
15859 2017-02-04 Jakub Jelinek <jakub@redhat.com>
15861 PR tree-optimization/79338
15862 * tree-parloops.c (gather_scalar_reductions): Don't call
15863 vect_analyze_loop_form for loop->inner before destroying loop's
15866 2017-02-03 Martin Sebor <msebor@redhat.com>
15868 PR tree-optimization/79327
15869 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
15870 when precision has resulted in leading zeros.
15871 (format_integer): Adjust the likely counter to assume an unknown
15872 argument that may be zero is non-zero.
15874 2017-02-03 Jason Merrill <jason@redhat.com>
15877 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
15878 avoid copying non-taken branch.
15880 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15882 PR tree-optimization/79340
15883 * tree-vect-loop.c (vectorizable_reduction): Release
15884 vec_defs elements after safe_splicing them into other vectors.
15887 PR tree-optimization/79327
15888 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
15889 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
15891 (format_integer): Use wide_int_to_tree instead of build_int_cst
15892 + to_?hwi. If argmin is NULL, just set argmin and argmax to
15893 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
15894 of shortest and longest sequence.
15896 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
15898 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
15899 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
15901 2017-02-03 Walter Lee <walt@tilera.com>
15904 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
15905 after initial stackframe link reg save.
15906 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
15908 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15911 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
15912 wu for stxssp alternative.
15914 2017-02-03 Martin Sebor <msebor@redhat.com>
15916 PR tree-optimization/79352
15917 * gimple-fold.c (get_range_strlen): Add argument.
15918 (get_range_strlen): Change return type to bool.
15919 (get_maxval_strlen): Pass in a dummy argument.
15920 * gimple-fold.h (get_range_strlen): Change return type to bool.
15921 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
15922 * tree.h (array_at_struct_end_p): Add argument.
15923 * tree.c (array_at_struct_end_p): Handle it.
15925 2017-02-03 Martin Liska <mliska@suse.cz>
15928 * multiple_target.c (create_dispatcher_calls): Redirect edge
15929 from a caller of a dispatcher.
15930 (expand_target_clones): Make the clones local.
15931 (ipa_target_clone): Do both target clones and resolvers.
15932 (ipa_dispatcher_calls): Remove the pass.
15933 (pass_dispatcher_calls::gate): Likewise.
15934 (make_pass_dispatcher_calls): Likewise.
15935 * passes.def (pass_target_clone): Put as very first IPA early
15938 2017-02-03 Martin Liska <mliska@suse.cz>
15940 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
15941 in case of a function with ifunc attribute.
15943 2017-02-03 Martin Liska <mliska@suse.cz>
15945 * cgraph.c (cgraph_node::dump): Dump function version info.
15946 * symtab.c (symtab_node::dump_base): Add missing new line.
15948 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15950 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
15951 (ifcombine_ifandif): Use it.
15953 2017-02-03 Martin Liska <mliska@suse.cz>
15955 * doc/invoke.texi: Document default value for
15956 use-after-scope-direct-emission-threshold.
15958 2017-02-03 Martin Liska <mliska@suse.cz>
15960 PR tree-optimization/79339
15961 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
15962 (format_floating): Likewise.
15964 2017-02-03 Martin Liska <mliska@suse.cz>
15967 * ipa-prop.c (ipa_node_params_t::insert): Remove current
15969 (ipa_node_params_t::remove): Likewise.
15970 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
15971 initialization from removed ipa_node_params_t::insert.
15972 (ipa_node_params::~ipa_node_params): Move from removed
15973 ipa_node_params_t::release.
15974 * symbol-summary.h (symbol_summary::m_released): New member.
15975 Do not release a summary twice. Do not allow to call finalizer
15976 for types of a summary that live in GGC memory.
15978 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15980 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
15983 2017-02-02 Martin Sebor <msebor@redhat.com>
15985 PR middle-end/79275
15986 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
15987 (format_string): Tighten up the range of output for non-constant
15988 strings and correct the expected range for wide non-constant strings.
15990 2017-02-02 Martin Sebor <msebor@redhat.com>
15992 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
15994 PR middle-end/32003
15995 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
15997 (-fdump-tree-@var): Add to index and document how to come up
15998 with pass-specific option and dump file names.
15999 (-fdump-passes): Clarify where to look for output.
16001 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16003 PR middle-end/77445
16004 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16005 statistics of the analyzed path; allow threading for speed when
16006 any of BBs along the path are optimized for speed.
16008 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16010 PR middle-end/78468
16011 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16012 settings of the virtual registers.
16015 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16017 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16018 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16021 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16023 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16024 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16026 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16028 * config/s390/s390.md: Add missing comments with the expanded
16030 * config/s390/vector.md: Likewise.
16031 * config/s390/vx-builtins.md: Likewise.
16033 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16036 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16037 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16038 conditions on a single line.
16040 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16042 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16043 __S390_VX__ to __VX__.
16045 2017-02-01 Andrew Pinski <apinski@cavium.com>
16047 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16048 stmt_info to record_stmt_cost.
16049 (vect_get_known_peeling_cost): Pass stmt_info if known to
16051 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16052 cpu_vector_cost field into
16053 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16054 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16055 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16056 splitting of scalar_stmt_cost and vec_stmt_cost.
16057 (thunderx_vector_cost): Likewise.
16058 (cortexa57_vector_cost): LIkewise.
16059 (exynosm1_vector_cost): Likewise.
16060 (xgene1_vector_cost): Likewise.
16061 (thunderx2t99_vector_cost): Improve after the splitting of the two
16063 (aarch64_builtin_vectorization_cost): Update for the splitting of
16064 scalar_stmt_cost and vec_stmt_cost.
16066 2017-02-01 Torvald Riegel <triegel@redhat.com>
16067 Richard Henderson <rth@redhat.com>
16069 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16070 conditional on existance of a fast atomic load.
16071 * optabs-query.c (can_atomic_load_p): New function.
16072 * optabs-query.h (can_atomic_load_p): Declare it.
16073 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16074 no fast atomic load is available for the particular size of access.
16075 (expand_atomic_compare_and_swap): Likewise.
16076 (expand_atomic_load): Likewise.
16077 (expand_atomic_store): Likewise.
16078 (expand_atomic_fetch_op): Likewise.
16079 * testsuite/lib/target-supports.exp
16080 (check_effective_target_sync_int_128): Remove x86 because it provides
16081 no fast atomic load.
16082 (check_effective_target_sync_int_128_runtime): Likewise.
16084 2017-02-01 Richard Biener <rguenther@suse.de>
16086 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16087 (graphite_transform_loops): Provide opt-info for optimized nests.
16088 * tree-parloop.c (parallelize_loops): Provide opt-info for
16089 parallelized loops.
16091 2017-02-01 Richard Biener <rguenther@suse.de>
16093 PR middle-end/79315
16094 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16095 was not set before.
16097 2017-02-01 Richard Biener <rguenther@suse.de>
16099 PR tree-optimization/71824
16100 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16101 Verify the loops are valid in the merged SESE region.
16102 (scop_detection::can_represent_loop_1): Check analyzing the
16103 evolution of the number of iterations in the region succeeds.
16105 2017-01-31 Ian Lance Taylor <iant@golang.org>
16107 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16108 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16110 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16112 PR preprocessor/79210
16113 * input.c (get_substring_ranges_for_loc): Replace line_width
16114 assertion with error-handling.
16116 2017-01-31 Richard Biener <rguenther@suse.de>
16118 PR tree-optimization/77318
16119 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16120 (create_pw_aff_from_tree): Take loop parameter.
16121 (add_condition_to_pbb): Pass loop of the condition to
16122 create_pw_aff_from_tree.
16124 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16126 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16127 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16129 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16133 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16135 (convert_int_to_float128): Likewise.
16136 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16137 (convert_int_to_float128): Likewise.
16138 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16139 (UNSPEC_IEEE128_CONVERT): Likewise.
16140 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16141 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16142 Use local variables for IBM extended format.
16143 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16144 (fix_trunc<mode>si2_fprs): Likewise.
16145 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16146 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16147 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16148 to know that we can now have integers of all sizes in vector
16150 (fix<uns>_<mode>di2_hw): Likewise.
16151 (float<uns>_<mode>si2_hw): Likewise.
16152 (fix_<mode>si2_hw): Likewise.
16153 (fixuns_<mode>si2_hw): Likewise.
16154 (float<uns>_<mode>di2_hw): Likewise.
16155 (float_<mode>di2_hw): Likewise.
16156 (float_<mode>si2_hw): Likewise.
16157 (floatuns_<mode>di2_hw): Likewise.
16158 (floatuns_<mode>si2_hw): Likewise.
16159 (xscvqp<su>wz_<mode>): Delete, no longer used.
16160 (xscvqp<su>dz_<mode>): Likewise.
16161 (xscv<su>dqp_<mode>): Likewise.
16162 (ieee128_mfvsrd_64bit): Likewise.
16163 (ieee128_mfvsrd_32bit): Likewise.
16164 (ieee128_mfvsrwz): Likewise.
16165 (ieee128_mtvsrw): Likewise.
16166 (ieee128_mtvsrd_64bit): Likewise.
16167 (ieee128_mtvsrd_32bit): Likewise.
16169 2017-01-31 Martin Liska <mliska@suse.cz>
16172 * ipa-prop.c (ipa_free_all_node_params): Call release method
16173 instead of ~sumbol_summary to not to trigger double times
16176 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16178 PR tree-optimization/71691
16179 * bitmap.h (class auto_bitmap): New.
16180 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16181 is_maybe_undefined instead of ssa_undefined_value_p.
16183 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16185 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16186 __S390_ARCH_LEVEL__ to __ARCH__.
16188 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16190 PR tree-optimization/79267
16191 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16192 if should_remove_lhs_p is true.
16194 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16197 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16198 (add_alignment_attribute): New.
16199 (base_type_die): Add alignment attribute.
16200 (subrange_type_die): Likewise.
16201 (modified_type_die): Likewise.
16202 (gen_array_type_die): Likewise.
16203 (gen_descr_array_type_die: Likewise.
16204 (gen_enumeration_type_die): Likewise.
16205 (gen_subprogram_die): Likewise.
16206 (gen_variable_die): Likewise.
16207 (gen_field_die): Likewise.
16208 (gen_ptr_to_mbr_type_die): Likewise.
16209 (gen_struct_or_union_type_die): Likewise.
16210 (gen_subroutine_type_die): Likewise.
16211 (gen_typedef_die): Likewise.
16212 (base_type_cmp): Compare alignment attribute.
16214 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16217 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16218 (setb_unsigned) New pattern for setb with CCUNS.
16219 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16220 subfc./subfe sequence to avoid overflow problems. Generate a
16221 shorter sequence with cmpld/setb for power9.
16222 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16223 for generating subfc. instruction.
16224 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16225 now uses this instruction.
16227 2017-01-30 Ian Lance Taylor <iant@google.com>
16230 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16231 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16233 2017-01-30 Martin Sebor <msebor@redhat.com>
16235 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16236 Move constant to the right of a relational operator.
16237 (get_mpfr_format_length, format_character, format_string): Ditto.
16238 (should_warn_p, maybe_warn): Same.
16240 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16242 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16245 * asan.c (get_translation_unit_decl): Remove function.
16246 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16248 2017-01-30 Martin Liska <mliska@suse.cz>
16250 PR gcov-profile/79259
16251 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16252 -fprofile-generate.
16254 2017-01-30 Martin Liska <mliska@suse.cz>
16257 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16258 Initialize variables with NULL value.
16260 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16263 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16265 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16267 2017-01-30 Richard Biener <rguenther@suse.de>
16269 PR tree-optimization/79276
16270 * tree-vrp.c (process_assert_insertions): Properly adjust common
16271 when removing a duplicate.
16273 * gcc.dg/torture/pr79276.c: New testcase.
16275 2017-01-30 Richard Biener <rguenther@suse.de>
16277 PR tree-optimization/79256
16278 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16279 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16281 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16283 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16286 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16287 ("*r<noxa>sbg_<mode>_sll_bitmask")
16288 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16289 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16290 Use contiguous_bitmask_nowrap_operand.
16292 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16295 * config/rs6000/altivec.h (vec_xl): Revise #define.
16296 (vec_xst): Likewise.
16298 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16300 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16302 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16304 PR rtl-optimization/79194
16305 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16306 traps before call to bypass_conditional_jumps.
16308 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16310 PR tree-optimization/71374
16311 * lra-constraints.c (check_conflict_input_operands): New.
16312 (match_reload): Use it.
16314 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16317 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16318 account to calculate conflict_set.
16320 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16322 PR rtl-optimization/78559
16323 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16324 other_insn in combine.
16326 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16328 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16329 uint16_type_node for BT_UINT16.
16331 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16333 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16334 "RTL Tests" to menu.
16335 (GIMPLE Tests): New node.
16336 (RTL Tests): New node.
16338 2017-01-27 Richard Biener <rguenther@suse.de>
16340 PR tree-optimization/79245
16341 * tree-loop-distribution.c (distribute_loop): Apply cost
16342 modeling also to detected patterns.
16344 2017-01-27 Richard Biener <rguenther@suse.de>
16346 PR tree-optimization/71433
16347 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16348 (compare_assert_loc): New function.
16349 (process_assert_insertions): Sort and optimize assert locations
16350 to remove duplicates and push down identical assertions on
16351 edges to their destination block.
16353 2017-01-27 Richard Biener <rguenther@suse.de>
16355 PR tree-optimization/79244
16356 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16357 out SSA names even if abnormal.
16359 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16361 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16362 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16363 instead of MPFR_RNDN.
16365 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16368 * arm.c (arm_option_override): Don't call build_target_option_node
16369 until after doing all option overrides.
16370 (arm_valid_target_attribute_tree): Likewise.
16372 2017-01-27 Martin Liska <mliska@suse.cz>
16374 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16375 for {cd}tors and C++ {cd}tors.
16377 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16379 * config/s390/s390.md ("*setmem_long_and")
16380 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16382 2017-01-26 Martin Sebor <msebor@redhat.com>
16384 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16387 2017-01-26 Martin Sebor <msebor@redhat.com>
16389 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16390 HAVE_DFmode before using XFmode or DFmode.
16391 (parse_directive): Avoid using the z length modifier to avoid
16392 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16394 PR middle-end/78703
16395 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16396 to accept adjustment as an array.
16397 (get_int_range): New function.
16398 (struct directive): Make width and prec arrays.
16399 (directive::set_width, directive::set_precision): Call get_int_range.
16400 (format_integer, format_floating): Handle width and precision ranges.
16401 (format_string, parse_directive): Same.
16403 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16406 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16407 comdat_type_p set, just clone them, but keep the children in the
16411 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16412 which have direct callers with -fvar-tracking-assignments enabled
16414 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16415 inside of type units.
16417 2017-01-26 Martin Sebor <msebor@redhat.com>
16419 PR middle-end/78703
16420 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16422 (struct format_result): Replace number_chars, number_chars_min,
16423 and number_chars_max with a single member of struct result_range.
16425 (format_result::operator+=): Adjust.
16426 (struct fmtresult): Remove bounded. Handle likely and unlikely
16428 (fmtresult::adjust_for_width_or_precision): New function.
16429 (fmtresult:type_max_digits): New function.
16430 (bytes_remaining): Handle likely and unlikely counters.
16431 (min_bytes_remaining): Remove.
16432 (format_percent): Simplify.
16433 (format_integer, format_floating): Set likely and unlikely counters.
16434 (get_string_length, format_character, format_string): Same.
16435 (format_plain, should_warn_p): New function.
16436 (maybe_warn): Call should_warn_p. Update diagnostic messages
16437 and handle those for all directives, including plain strings.
16438 (format_directive): Handle likely and unlikely counters.
16439 Remove unnecessary quoting from diagnostics. Add an informational
16441 (add_bytes): Remove.
16442 (pass_sprintf_length::compute_format_length): Simplify.
16443 (try_substitute_return_value): Handle likely and unlikely counters.
16445 2017-01-26 Carl Love <cel@us.ibm.com>
16447 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16448 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16450 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16453 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16454 endianess for subregs into account.
16455 * lra-constraints.c (lra_constraints): Do risky transformations
16456 always on the first iteration.
16457 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16458 last_call_used_reg_set.
16459 (process_bb_lives): Define and use last_call_used_reg_set.
16460 * lra.c (lra): Always continue after lra_constraints on the first
16463 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16465 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16467 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16469 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16471 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16472 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16473 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16474 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16475 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16476 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16477 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16478 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16479 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16481 2017-01-26 Marek Polacek <polacek@redhat.com>
16484 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16485 for the third operand.
16487 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16489 PR middle-end/79236
16490 * omp-low.c (struct omp_context): Add simt_stmt field.
16491 (scan_omp_for): Return omp_context *.
16492 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16493 context to the _simt_ SIMD stmt.
16494 (lower_omp_for): For combined SIMD with sibling _simt_
16495 SIMD, make sure to use the same decls in _looptemp_
16496 clauses as in the sibling.
16498 2017-01-26 David Sherwood <david.sherwood@arm.com>
16500 PR middle-end/79212
16501 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16504 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16507 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16508 emit fld b; fld a; if possible.
16510 * brig-builtins.def: Update copyright years.
16511 * config/arm/arm_acle_builtins.def: Update copyright years.
16513 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16516 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16517 constraint instead of o for the stxsd instruction.
16519 2017-01-25 Carl Love <cel@us.ibm.com>
16521 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16522 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16524 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16526 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16528 2017-01-25 Richard Biener <rguenther@suse.de>
16530 PR tree-optimization/69264
16531 * target.def (vector_alignment_reachable): Improve documentation.
16532 * doc/tm.texi: Regenerate.
16533 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16535 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16536 earlier changes with respect to TYPE_USER_ALIGN.
16537 (vector_alignment_reachable_p): Likewise. Improve dumping.
16539 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16542 * config/arm/arm.md (xordi3): Force constant operand into a register
16545 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16547 * doc/invoke.texi (-fstore-merging): Correct default optimization
16548 levels at which it is enabled.
16549 (-O): Move -fstore-merging from list to...
16552 2017-01-25 Richard Biener <rguenther@suse.de>
16555 * omp-expand.c: Include debug.h.
16556 (expand_omp_taskreg): Make sure to generate early debug before
16557 outlining anything from a function.
16558 (expand_omp_target): Likewise.
16559 (grid_expand_target_grid_body): Likewise.
16561 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16564 * asan.c (get_translation_unit_decl): New function.
16565 (asan_add_global): Extract modules file name from globals
16566 TRANSLATION_UNIT_DECL name.
16568 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16571 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16572 for long calls with APCS frame and VFP.
16574 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16576 * cfg.c (original_copy_tables_initialized_p): New function.
16577 * cfg.h (original_copy_tables_initialized_p): New decl.
16578 * cfgrtl.c (relink_block_chain): Guard the call to
16579 free_original_copy_tables with a call to
16580 original_copy_tables_initialized_p.
16581 * cgraph.h (symtab_node::native_rtl_p): New decl.
16582 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16583 (symtab_node::needed_p): Don't assert for early assembly output
16584 for __RTL functions.
16585 (cgraph_node::finalize_function): Set "force_output" for __RTL
16587 (cgraph_node::analyze): Bail out early for __RTL functions.
16588 (analyze_functions): Update assertion to support __RTL functions.
16589 (cgraph_node::expand): Bail out early for __RTL functions.
16590 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16592 * function.h (struct function): Update comment for field
16594 * gimple-expr.c: Include "tree-pass.h".
16595 (gimple_has_body_p): Return false for __RTL functions.
16596 * Makefile.in (OBJS): Add run-rtl-passes.o.
16597 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16599 (gcc::pass_manager::get_clean_slate): New accessor.
16600 * passes.c: Include "insn-addr.h".
16601 (should_skip_pass_p): Add logging. Update logic for running
16602 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16603 property-provider override so it is only done for gimple passes.
16605 (skip_pass): New function.
16606 (execute_one_pass): Call skip_pass when skipping passes.
16607 * read-md.c (md_reader::read_char): Support filtering
16608 the input to a subset of line numbers.
16609 (md_reader::md_reader): Initialize fields
16610 m_first_line and m_last_line.
16611 (md_reader::read_file_fragment): New function.
16612 * read-md.h (md_reader::read_file_fragment): New decl.
16613 (md_reader::m_first_line): New field.
16614 (md_reader::m_last_line): New field.
16615 * read-rtl-function.c (function_reader::create_function): Only
16616 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16617 curr_properties. Set DECL_INITIAL to a dummy block.
16618 (read_rtl_function_body_from_file_range): New function.
16619 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16621 * run-rtl-passes.c: New file.
16622 * run-rtl-passes.h: New file.
16624 2017-01-24 Jeff Law <law@redhat.com>
16626 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16629 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16631 PR tree-optimization/79159
16632 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16633 (record_nonwrapping_iv): Improve boundary using above function if no
16634 value range information.
16636 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16637 Martin Jambor <mjambor@suse.cz>
16639 * brig-builtins.def: New file.
16640 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16641 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16642 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16643 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16644 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16645 * builtin-types.def (BT_INT8): New.
16646 (BT_INT16): Likewise.
16647 (BT_UINT8): Likewise.
16648 (BT_UINT16): Likewise.
16649 (BT_FN_ULONG): Likewise.
16650 (BT_FN_UINT_INT): Likewise.
16651 (BT_FN_UINT_ULONG): Likewise.
16652 (BT_FN_UINT_LONG): Likewise.
16653 (BT_FN_UINT_PTR): Likewise.
16654 (BT_FN_ULONG_PTR): Likewise.
16655 (BT_FN_INT8_FLOAT): Likewise.
16656 (BT_FN_INT16_FLOAT): Likewise.
16657 (BT_FN_UINT32_FLOAT): Likewise.
16658 (BT_FN_UINT16_FLOAT): Likewise.
16659 (BT_FN_UINT8_FLOAT): Likewise.
16660 (BT_FN_UINT64_FLOAT): Likewise.
16661 (BT_FN_UINT16_UINT32): Likewise.
16662 (BT_FN_UINT32_UINT16): Likewise.
16663 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16664 (BT_FN_INT_PTR_INT): Likewise.
16665 (BT_FN_UINT_PTR_UINT): Likewise.
16666 (BT_FN_LONG_PTR_LONG): Likewise.
16667 (BT_FN_ULONG_PTR_ULONG): Likewise.
16668 (BT_FN_VOID_UINT64_UINT64): Likewise.
16669 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16670 (BT_FN_INT8_INT8_INT8): Likewise.
16671 (BT_FN_INT16_INT16_INT16): Likewise.
16672 (BT_FN_INT_INT_INT): Likewise.
16673 (BT_FN_UINT_FLOAT_UINT): Likewise.
16674 (BT_FN_FLOAT_UINT_UINT): Likewise.
16675 (BT_FN_ULONG_UINT_UINT): Likewise.
16676 (BT_FN_ULONG_UINT_PTR): Likewise.
16677 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16678 (BT_FN_UINT_UINT_UINT): Likewise.
16679 (BT_FN_VOID_UINT_PTR): Likewise.
16680 (BT_FN_UINT_UINT_PTR: Likewise.
16681 (BT_FN_UINT32_UINT64_PTR): Likewise.
16682 (BT_FN_INT_INT_UINT_UINT): Likewise.
16683 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16684 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16685 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16686 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16687 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16688 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16689 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16690 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16691 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16692 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16693 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16694 * doc/frontends.texi: List BRIG FE.
16695 * doc/install.texi (Testing): Add BRIG tesring requirements.
16696 * doc/invoke.texi (Overall Options): Mention BRIG.
16697 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16699 2017-01-24 Richard Biener <rguenther@suse.de>
16701 PR translation/79208
16702 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16704 2017-01-24 Martin Jambor <mjambor@suse.cz>
16707 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16708 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16709 and known_contexts.
16711 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16713 PR middle-end/79123
16714 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16715 casts from signed to unsigned really don't have a range.
16717 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16719 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16720 GMP_RNDx for compatiblity.
16722 2017-01-24 Martin Liska <mliska@suse.cz>
16725 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16726 that would prevent us to call alloca with -1 as argument.
16728 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16730 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16731 Avoid -Wformat-security warning.
16733 2017-01-23 Andrew Pinski <apinski@cavium.com>
16735 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16738 2017-01-23 Martin Sebor <msebor@redhat.com>
16740 PR middle-end/78703
16741 * gimple-ssa-sprintf.c (warn_level): New global.
16742 (format_integer): Use it here and throughout the rest of the file.
16743 Use the same switch to compute sign as base.
16744 (maybe_warn): New function.
16745 (format_directive): Factor out warnings into maybe_warn.
16746 Add debugging output. Use warn_level.
16747 (add_bytes): Use warn_level.
16748 (pass_sprintf_length::compute_format_length): Add debugging output.
16749 (try_substitute_return_value): Same.
16750 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
16752 PR middle-end/78703
16753 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
16754 (struct fmtresult, format_integer, format_floating): Adjust.
16755 (fmtresult::fmtresult): Set max correctly in two argument ctor.
16756 (get_string_length, format_string,format_directive): Same.
16757 (pass_sprintf_length::compute_format_length): Same.
16758 (try_substitute_return_value): Simplify slightly.
16760 PR middle-end/78703
16761 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
16762 (fmtresult::operator+=): Outlined.
16763 (struct fmtresult): Add ctors.
16764 (struct conversion_spec): Rename...
16765 (struct directive): ...to this. Add and remove data members.
16766 (directive::set_width, directive::set_precision): New functions.
16767 (format_percent): Use fmtresult ctor.
16768 (get_width_and_precision): Remove.
16769 (format_integer): Make naming changes. Avoid computing width and
16771 (format_floating): Same. Adjust indentation.
16772 (format_character, format_none): New functions.
16773 (format_string): Moved character handling to format_character.
16774 (format_directive): Remove arguments, change return type.
16775 (parse_directive): New function.
16776 (pass_sprintf_length::compute_format_length): Move directive
16777 parsing to parse_directive.
16779 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16781 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
16782 (assign_assembler_name_if_needed): ... this.
16783 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
16784 (assign_assembler_name_if_needed): ... this.
16785 (free_lang_data_in_cgraph): Adjust callers.
16786 * cgraphunit.c (cgraph_node::analyze): Likewise.
16787 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
16790 2017-01-23 Richard Biener <rguenther@suse.de>
16792 PR tree-optimization/79088
16793 PR tree-optimization/79188
16794 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
16795 resetting loop bounds after last path deletion. Reset loop
16796 bounds of the target loop, make code match the comments.
16797 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
16798 Make sure loops need no fixups.
16800 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16802 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
16803 exponent support with double type for first argument.
16804 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
16805 type returned by __builtin_vec_extract_sig,
16806 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
16807 functions from "vector int" to "vector unsigned int" or from
16808 "vector long long int" to "vector unsigned long long int".
16809 Changed type returned by __builtin_vec_extract_exp,
16810 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
16811 functions from "vector int" to "vector unsigned int" or from
16812 "vector long long int" to "vector unsigned long long int".
16813 Changed return type of __builtin_vec_test_data_class,
16814 __builtin_vec_test_data_class_sp, and
16815 __builtin_vec_test_data_class_dp from "vector int" to
16816 "vector bool int" or from "vector long long int" to "vector bool
16817 long long int" and changed second argument type from "unsigned
16818 int" to "int". Added new overloaded function forms "vector float
16819 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
16820 "vector float __builtin_vec_insert_exp_sp (vector float, vector
16821 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
16822 double, vector unsigned long long int)" and "vector double
16823 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
16824 long int)". Changed return type of
16825 __builtin_scalar_test_data_class and
16826 __builtin_scalar_test_data_class_sp and
16827 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
16828 int" and changed second argument from "unsigned int" to "int".
16829 Changed type returned by __builtin_scalar_test_neg,
16830 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
16831 from "int" to "bool int". Added new overloaded function form
16832 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
16833 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
16834 exponent double-precision with floating point first argument.
16835 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16836 documentation of scalar_test_data_class, scalar_test_neg,
16837 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
16838 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
16839 vec_test_data_class built-in functions to reflect refinements in
16840 their type signatures.
16842 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
16844 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
16846 (aarch64_elf_asm_destructor): Likewise.
16848 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
16850 PR rtl-optimization/78634
16851 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
16852 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
16853 * ifcvt.c (noce_try_cmove): Add missing cost check.
16855 PR rtl-optimization/71724
16856 * combine.c (if_then_else_cond): Look for situations where it is
16857 beneficial to undo the work of one of the recursive calls.
16859 2017-01-23 Bin Cheng <bin.cheng@arm.com>
16861 PR tree-optimization/70754
16862 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
16863 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
16864 combined stmt before it if not NULL.
16865 (combine_chains): Process refs reversely and compute dominance point
16868 2017-01-23 Martin Liska <mliska@suse.cz>
16870 PR tree-optimization/79196
16871 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
16872 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
16874 (strlen_optimize_stmt): Call the renamed function.
16876 2017-01-23 Michael Matz <matz@suse.de>
16878 PR tree-optimization/78384
16879 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
16881 2017-01-23 Richard Biener <rguenther@suse.de>
16883 PR tree-optimization/79186
16884 * tree-vrp.c (register_new_assert_for): Make sure we've seen
16885 both incoming edges before moving an assert.
16887 2017-01-23 Martin Jambor <mjambor@suse.cz>
16889 * ipa-prop.c (load_from_param_1): Removed.
16890 (load_from_unmodified_param): Bits from load_from_param_1 put back
16892 (load_from_param): Removed.
16893 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
16894 with stmt. Reverted back to use of load_from_unmodified_param.
16896 2017-01-23 Martin Jambor <mjambor@suse.cz>
16899 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
16900 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
16901 field a pointer to garbage collected vector, mark lattices and
16902 ipcp_orig_node with GTY((skip)).
16903 (ipa_get_param_count): Adjust to descriptors being a pointer.
16904 (ipa_get_param): Likewise.
16905 (ipa_get_type): Likewise.
16906 (ipa_get_param_move_cost): Likewise.
16907 (ipa_set_param_used): Likewise.
16908 (ipa_get_controlled_uses): Likewise.
16909 (ipa_set_controlled_uses): Likewise.
16910 (ipa_is_param_used): Likewise.
16911 (ipa_node_params_t): Move into garbage collector. New methods insert
16913 (ipa_node_params_sum): Annotate wth GTY(()).
16914 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
16916 (ipa_load_from_parm_agg): Adjust declaration.
16917 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
16918 * ipa-profile.c (ipa_profile): Likewise.
16919 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
16920 (ipa_populate_param_decls): Make descriptors parameter garbage
16922 (ipa_dump_param): Adjust to descriptors being a pointer.
16923 (ipa_alloc_node_params): Likewise.
16924 (ipa_initialize_node_params): Likewise.
16925 (load_from_param_1): Make descriptors parameter garbage collected.
16926 (load_from_unmodified_param): Likewise.
16927 (load_from_param): Likewise.
16928 (ipa_load_from_parm_agg): Likewise.
16929 (ipa_node_params::~ipa_node_params): Removed.
16930 (ipa_free_all_node_params): Remove call to delete operator.
16931 (ipa_node_params_t::insert): New.
16932 (ipa_node_params_t::remove): Likewise.
16933 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
16934 copy known_csts and known_contexts vectors.
16935 (ipa_read_node_info): Adjust to descriptors being a pointer.
16936 (ipcp_modif_dom_walker): Make m_descriptors field garbage
16938 (ipcp_transform_function): Make descriptors variable garbage
16941 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
16943 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
16944 * config/i386/avx512dqintrin.h: Ditto.
16945 * config/i386/avx512fintrin.h: Ditto.
16946 * config/i386/i386.c: Handle new builtins.
16947 * config/i386/i386-builtin.def: Add new builtins.
16948 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
16949 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
16951 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16952 Martin Liska <mliska@suse.cz>
16954 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
16955 * asan.c (asan_expand_poison_ifn): Support stores and use
16956 appropriate ASAN report function.
16957 * internal-fn.c (expand_ASAN_POISON_USE): New function.
16958 * internal-fn.def (ASAN_POISON_USE): Declare.
16959 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
16960 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
16961 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
16962 ASAN_POISON calls w/o LHS.
16963 * tree-ssa.c (execute_update_addresses_taken): Create clobber
16964 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
16965 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
16966 * gimplify.c (asan_poison_variables): Add attribute
16967 use_after_scope_memory to variables that really needs to live
16969 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
16970 having the attribute.
16972 2017-01-23 Martin Liska <mliska@suse.cz>
16974 * asan.c (create_asan_shadow_var): New function.
16975 (asan_expand_poison_ifn): Likewise.
16976 * asan.h (asan_expand_poison_ifn): New declaration.
16977 * internal-fn.c (expand_ASAN_POISON): Likewise.
16978 * internal-fn.def (ASAN_POISON): New builtin.
16979 * sanopt.c (pass_sanopt::execute): Expand
16980 asan_expand_poison_ifn.
16981 * tree-inline.c (copy_decl_for_dup_finish): Make function
16983 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
16984 * tree-ssa.c (is_asan_mark_p): New function.
16985 (execute_update_addresses_taken): Rewrite local variables
16986 (identified just by use-after-scope as addressable) into SSA.
16988 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16990 * doc/install.texi (Specific): opensource.apple.com uses https
16991 now. Remove trailing slash.
16993 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16995 * README.Portability: Remove note on an Irix compatibility issue.
16997 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
16999 * gcov.c (INCLUDE_ALGORITHM): Define.
17000 (INCLUDE_VECTOR): Define.
17001 No longer include <vector> and <algorithm> directly.
17003 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17005 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17007 * doc/invoke.texi (Code Gen Options): Ditto.
17009 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17012 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17014 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17016 rtl-optimization/79125
17017 * cprop.c (local_cprop_pass): Handle cases where we make an
17018 unconditional trap.
17020 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17024 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17025 read from, for big endian.
17027 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17029 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17030 register pauth builtins for LP64 only.
17032 2017-01-20 Marek Polacek <polacek@redhat.com>
17035 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17038 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17040 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17042 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17043 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17044 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17046 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17049 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17050 in big-endian mode when they are not a single duplicated value.
17052 2017-01-20 Richard Biener <rguenther@suse.de>
17054 * BASE-VER: Bump to 7.0.1.
17056 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17058 * omp-low.c (omplow_simd_context): New struct. Use it...
17059 (lower_rec_simd_input_clauses): ...here and...
17060 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17061 references to idx, lane, max_vf, is_simt.
17063 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17065 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17068 2017-01-20 Martin Jambor <mjambor@suse.cz>
17070 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17071 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17073 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17074 (GTFILES): Rename hsa.c to hsa-common.c.
17075 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17076 * hsa-dump.c: Likewise.
17077 * hsa-gen.c: Likewise.
17078 * hsa-regalloc.c: Likewise.
17079 * ipa-hsa.c: Likewise.
17080 * omp-expand.c: Likewise.
17081 * omp-low.c: Likewise.
17082 * toplev.c: Likewise.
17084 2017-01-20 Marek Polacek <polacek@redhat.com>
17087 * doc/invoke.texi: Document -Wduplicated-branches.
17088 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17089 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17090 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17091 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17092 return 0 only when not OEP_LEXICOGRAPHIC.
17093 (fold_build_cleanup_point_expr): Use the expression
17094 location when building CLEANUP_POINT_EXPR.
17095 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17096 * tree.c (add_expr): Handle error_mark_node.
17098 2017-01-20 Martin Liska <mliska@suse.cz>
17101 * tree-profile.c (init_ic_make_global_vars): Do not call
17103 (gimple_init_gcov_profiler): Likewise.
17105 2017-01-20 Martin Liska <mliska@suse.cz>
17108 * cgraph.h (maybe_create_reference): Remove argument and
17110 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17112 * ipa-cp.c (create_specialized_node): Likewise.
17113 * symtab.c (symtab_node::maybe_create_reference): Handle
17114 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17116 2017-01-20 Martin Liska <mliska@suse.cz>
17118 * read-rtl-function.c (function_reader::create_function): Use
17119 build_decl instread of build_decl_stat.
17121 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17123 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17124 * config/i386/avx512dqintrin.h: Ditto.
17125 * config/i386/avx512fintrin.h: Ditto.
17126 * config/i386/i386-builtin-types.def: Add new types.
17127 * config/i386/i386.c: Handle new types.
17128 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17129 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17130 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17131 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17132 (__builtin_ia32_kshiftridi): New.
17133 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17135 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17139 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17140 define to rs6000_init_stack_protect_guard.
17141 (rs6000_init_stack_protect_guard): New function.
17143 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17144 Yunqiang Su <yunqiang.su@imgtec.com>
17146 * config.gcc (supported_defaults): Add madd4.
17147 (with_madd4): Add validation.
17148 (all_defaults): Add madd4.
17149 * config/mips/mips.opt (mmadd4): New option.
17150 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17152 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17154 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17155 (ISA_HAS_FUSED_MADD4): Likewise.
17156 * doc/invoke.texi (-mmadd4): Document the new option.
17157 * doc/install.texi (--with-madd4): Document the new option.
17159 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17161 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17162 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17163 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17164 (aarch64_init_pauth_hint_builtins): New.
17165 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17166 (aarch64_expand_builtin): Expand new builtins.
17168 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17170 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17171 * combine-stack-adj.c (no_unhandled_cfa): Handle
17172 REG_CFA_TOGGLE_RA_MANGLE.
17173 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17174 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17175 info for return address signing.
17176 (aarch64_expand_epilogue): Likewise.
17178 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17180 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17181 * config/aarch64/aarch64-protos.h
17182 (aarch64_return_address_signing_enabled): New declaration.
17183 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17185 (aarch64_expand_prologue): Sign return address before it's pushed onto
17187 (aarch64_expand_epilogue): Authenticate return address fetched from
17189 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17190 (aarch64_attributes): New function attributes for "sign-return-address".
17191 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17192 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17193 ("*do_return"): Generate combined instructions according to key index.
17194 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17195 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17197 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17198 * config/aarch64/aarch64.opt (msign-return-address=): New.
17199 * doc/extend.texi (AArch64 Function Attributes): Documents
17200 "sign-return-address=".
17201 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17203 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17205 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17206 overall option summary.
17208 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17210 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17211 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17212 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17213 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17215 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17217 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17218 -mpower9-minmax by default for -mcpu=power9.
17219 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17220 128-bit floating point.
17222 2017-01-20 Alan Modra <amodra@gmail.com>
17224 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17225 optimizing for size.
17227 2017-01-20 Alan Modra <amodra@gmail.com>
17230 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17231 for strcmp and strncmp from corresponding builtin decl.
17233 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17235 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17236 instead of i386/rtems-64.h.
17237 * config/i386/rtems-64.h: Remove.
17239 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17243 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17245 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17247 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17249 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17250 Change int to HOST_WIDE_INT.
17251 * config/aarch64/aarch64-protos.h
17252 (aarch64_simd_gen_const_vector_dup): Likewise.
17253 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17255 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17257 * langhooks-def.h (lhd_type_for_size): New decl.
17258 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17259 * langhooks.c (lhd_type_for_size): New function, taken from
17262 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17264 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17265 define_bypass for CR latency.
17266 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17267 (power9-alu2): Add define_bypass for CR latency.
17269 (power9-mul): Update insn latency.
17270 (power9-mul-compare): Update insn latency, bypass latency and remove
17273 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17275 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17277 * config/aarch64/aarch64.md
17278 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17279 aarch64_nopcrelative_literal_loads.
17280 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17282 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17284 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17285 TARGET_LOONGSON_3A.
17286 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17288 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17291 * config.gcc (supported_defaults): Add lxc1-sxc1.
17292 (with_lxc1_sxc1): Add validation.
17293 (all_defaults): Add lxc1-sxc1.
17294 * config/mips/mips.opt (mlxc1-sxc1): New option.
17295 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17297 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17298 __mips_no_lxc1_sxc1.
17299 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17300 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17301 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17303 2017-01-19 Richard Biener <rguenther@suse.de>
17305 PR tree-optimization/72488
17306 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17307 sure to restore SSA info.
17308 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17310 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17312 PR rtl-optimization/79121
17313 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17314 of the inner type when shifting an extended value.
17316 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17319 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17320 interposable aliases.
17322 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17325 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17326 Use the evmergelohi instruction.
17327 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17328 (mov_sitf_e500_subreg8_2_be): Likewise.
17329 (mov_sitf_e500_subreg12_2_le): Likewise.
17330 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17331 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17332 (mov_sitf_e500_subreg8_2_le): Likewise.
17333 (mov_sitf_e500_subreg12_2_be): Likewise.
17335 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17337 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17338 attribute from vecsimple to vecperm.
17339 (altivec_vbpermq2): Likewise.
17341 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17344 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17346 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17347 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17348 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17349 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17350 case where N arg is SIZE_MAX.
17351 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17352 (cmpstrsi): Add pattern.
17354 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17356 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17357 __builtin_vec_revb builtins.
17358 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17359 built-in functions to support generation of the ISA 3.0 XXBR<x>
17360 vector byte reverse instructions.
17361 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17362 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17363 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17364 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17365 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17366 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17367 (P9V_BUILTIN_VEC_REVB): Likewise.
17368 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17369 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17370 (p9_xxbrq_v16qi): Likewise.
17371 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17372 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17373 (p9_xxbrh_v8hi): Likewise.
17374 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17375 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17376 vec_revb built-in functions.
17378 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17380 PR rtl-optimization/78952
17381 * config/i386/i386.md (any_extract): New code iterator.
17382 (*insvqi_2): Use any_extract for source operand.
17383 (*insvqi_3): Use any_shiftrt for source operand.
17385 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17387 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17389 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17391 2017-01-18 Matthias Klose <doko@ubuntu.com>
17393 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17395 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17397 * config/rs6000/altivec.h (vec_bperm): Change #define.
17398 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17399 (altivec_vbpermq2): New define_insn.
17400 (altivec_vbpermd): Likewise.
17401 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17402 function interface.
17403 (VBPERMD): Likewise.
17404 (VBPERM): New polymorphic function interface.
17405 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17406 Add entries for P9V_BUILTIN_VEC_VBPERM.
17407 * doc/extend.texi: Add interfaces for vec_bperm.
17409 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17411 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17412 first letter of error messages.
17413 (s390_resolve_overloaded_builtin): Likewise.
17414 * config/s390/s390.c (s390_expand_builtin): Likewise.
17415 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17416 (s390_valid_target_attribute_inner_p): Likewise.
17417 * config/s390/s390.md ("tabort"): Likewise.
17419 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17421 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17422 (ISA_AVOID_DIV_HILO): New macro.
17423 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17424 (ISA_HAS_DDIV): Likewise.
17426 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17428 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17430 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17432 * doc/invoke.texi (fabi-version): Spelling fix.
17434 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17437 * doc/invoke.texi (fabi-version): Mention mangling fix for
17440 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17443 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17445 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17448 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17449 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17451 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17453 (stack_protect_test): Ditto.
17454 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17455 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17457 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17458 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17459 -mstack-protector-guard-offset=.
17460 (RS/6000 and PowerPC Options): Ditto.
17462 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17464 * config/i386/i386.h (MASK_CLASS_P): New define.
17465 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17466 there are no registers from different register sets also when
17467 mask registers are used. Update function comment.
17468 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17469 to (*k/*r) and (*k/*km) alternatives.
17471 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17473 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17474 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17475 (EH_RETURN_HANDLER_RTX): New define.
17476 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17477 Force frame pointer in EH return functions.
17478 (aarch64_expand_epilogue): Add barrier for eh_return.
17479 (aarch64_final_eh_return_addr): Remove.
17480 (aarch64_eh_return_handler_rtx): New function.
17481 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17483 (aarch64_eh_return_handler_rtx): New prototype.
17485 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17487 * config/rs6000/altivec.h (vec_rlmi): New #define.
17488 (vec_vrlnm): Likewise.
17489 (vec_rlnm): Likewise.
17490 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17491 (UNSPEC_VRLNM): Likewise.
17492 (VIlong): New mode iterator.
17493 (altivec_vrl<VI_char>mi): New define_insn.
17494 (altivec_vrl<VI_char>nm): Likewise.
17495 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17497 (VRLDNM): Likewise.
17498 (RLNM): New polymorphic function entry.
17499 (VRLWMI): New monomorphic function entry.
17500 (VRLDMI): Likewise.
17501 (RLMI): New polymorphic function entry.
17502 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17503 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17504 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17507 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17510 * dwarf2out.c (field_byte_offset): Restore the
17511 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17512 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17513 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17516 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17519 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17521 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17524 * dwarf2out.c (add_data_member_location_attribute): For constant
17525 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17526 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17527 DW_AT_byte_size attributes.
17529 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17531 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17532 after forcing to constant memory when the code model is medium.
17534 2017-01-17 Julia Koval <julia.koval@intel.com>
17537 * config/i386/avx512fintrin.h
17538 (_mm512_i32gather_ps): Change __addr type to void const*.
17539 (_mm512_mask_i32gather_ps): Ditto.
17540 (_mm512_i32gather_pd): Ditto.
17541 (_mm512_mask_i32gather_pd): Ditto.
17542 (_mm512_i64gather_ps): Ditto.
17543 (_mm512_mask_i64gather_ps): Ditto.
17544 (_mm512_i64gather_pd): Ditto.
17545 (_mm512_mask_i64gather_pd): Ditto.
17546 (_mm512_i32gather_epi32): Ditto.
17547 (_mm512_mask_i32gather_epi32): Ditto.
17548 (_mm512_i32gather_epi64): Ditto.
17549 (_mm512_mask_i32gather_epi64): Ditto.
17550 (_mm512_i64gather_epi32): Ditto.
17551 (_mm512_mask_i64gather_epi32): Ditto.
17552 (_mm512_i64gather_epi64): Ditto.
17553 (_mm512_mask_i64gather_epi64): Ditto.
17554 (_mm512_i32scatter_ps): Change __addr type to void*.
17555 (_mm512_mask_i32scatter_ps): Ditto.
17556 (_mm512_i32scatter_pd): Ditto.
17557 (_mm512_mask_i32scatter_pd): Ditto.
17558 (_mm512_i64scatter_ps): Ditto.
17559 (_mm512_mask_i64scatter_ps): Ditto.
17560 (_mm512_i64scatter_pd): Ditto.
17561 (_mm512_mask_i64scatter_pd): Ditto.
17562 (_mm512_i32scatter_epi32): Ditto.
17563 (_mm512_mask_i32scatter_epi32): Ditto.
17564 (_mm512_i32scatter_epi64): Ditto.
17565 (_mm512_mask_i32scatter_epi64): Ditto.
17566 (_mm512_i64scatter_epi32): Ditto.
17567 (_mm512_mask_i64scatter_epi32): Ditto.
17568 (_mm512_i64scatter_epi64): Ditto.
17569 (_mm512_mask_i64scatter_epi64): Ditto.
17570 * config/i386/avx512pfintrin.h
17571 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17572 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17573 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17574 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17575 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17576 (_mm512_prefetch_i32scatter_ps): Ditto.
17577 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17578 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17579 (_mm512_prefetch_i64scatter_pd): Ditto.
17580 (_mm512_prefetch_i64scatter_ps): Ditto.
17581 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17582 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17583 * config/i386/avx512vlintrin.h
17584 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17585 (_mm_mmask_i32gather_ps): Ditto.
17586 (_mm256_mmask_i32gather_pd): Ditto.
17587 (_mm_mmask_i32gather_pd): Ditto.
17588 (_mm256_mmask_i64gather_ps): Ditto.
17589 (_mm_mmask_i64gather_ps): Ditto.
17590 (_mm256_mmask_i64gather_pd): Ditto.
17591 (_mm_mmask_i64gather_pd): Ditto.
17592 (_mm256_mmask_i32gather_epi32): Ditto.
17593 (_mm_mmask_i32gather_epi32): Ditto.
17594 (_mm256_mmask_i32gather_epi64): Ditto.
17595 (_mm_mmask_i32gather_epi64): Ditto.
17596 (_mm256_mmask_i64gather_epi32): Ditto.
17597 (_mm_mmask_i64gather_epi32): Ditto.
17598 (_mm256_mmask_i64gather_epi64): Ditto.
17599 (_mm_mmask_i64gather_epi64): Ditto.
17600 (_mm256_i32scatter_ps): Change __addr type to void*.
17601 (_mm256_mask_i32scatter_ps): Ditto.
17602 (_mm_i32scatter_ps): Ditto.
17603 (_mm_mask_i32scatter_ps): Ditto.
17604 (_mm256_i32scatter_pd): Ditto.
17605 (_mm256_mask_i32scatter_pd): Ditto.
17606 (_mm_i32scatter_pd): Ditto.
17607 (_mm_mask_i32scatter_pd): Ditto.
17608 (_mm256_i64scatter_ps): Ditto.
17609 (_mm256_mask_i64scatter_ps): Ditto.
17610 (_mm_i64scatter_ps): Ditto.
17611 (_mm_mask_i64scatter_ps): Ditto.
17612 (_mm256_i64scatter_pd): Ditto.
17613 (_mm256_mask_i64scatter_pd): Ditto.
17614 (_mm_i64scatter_pd): Ditto.
17615 (_mm_mask_i64scatter_pd): Ditto.
17616 (_mm256_i32scatter_epi32): Ditto.
17617 (_mm256_mask_i32scatter_epi32): Ditto.
17618 (_mm_i32scatter_epi32): Ditto.
17619 (_mm_mask_i32scatter_epi32): Ditto.
17620 (_mm256_i32scatter_epi64): Ditto.
17621 (_mm256_mask_i32scatter_epi64): Ditto.
17622 (_mm_i32scatter_epi64): Ditto.
17623 (_mm_mask_i32scatter_epi64): Ditto.
17624 (_mm256_i64scatter_epi32): Ditto.
17625 (_mm256_mask_i64scatter_epi32): Ditto.
17626 (_mm_i64scatter_epi32): Ditto.
17627 (_mm_mask_i64scatter_epi32): Ditto.
17628 (_mm256_i64scatter_epi64): Ditto.
17629 (_mm256_mask_i64scatter_epi64): Ditto.
17630 (_mm_i64scatter_epi64): Ditto.
17631 (_mm_mask_i64scatter_epi64): Ditto.
17632 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17633 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17634 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17635 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17636 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17637 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17638 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17639 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17640 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17641 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17642 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17643 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17644 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17645 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17646 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17647 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17648 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17649 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17650 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17651 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17652 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17653 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17654 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17655 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17656 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17657 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17658 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17659 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17660 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17661 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17662 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17663 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17664 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17665 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17666 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17667 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17668 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17669 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17670 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17671 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17672 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17673 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17674 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17675 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17676 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17677 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17678 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17679 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17680 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17681 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17682 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17683 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17684 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17685 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17686 definitions accordingly.
17688 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17689 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17692 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17695 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17698 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17699 hard regs for inner regno.
17701 2017-01-17 Martin Liska <mliska@suse.cz>
17704 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17705 assumption and add comment.
17707 2017-01-17 Nathan Sidwell <nathan@acm.org>
17709 * ipa-visibility.c (localize_node): New function, broken out of ...
17710 (function_and_variable_visibility): ... here. Call it.
17712 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17714 PR middle-end/77445
17715 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17716 correctly set frequency of oudgoing edge.
17717 (duplicate_thread_path): Fix profile updating.
17719 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17722 * configure.ac: Add GCC_BASE_VER.
17723 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17724 version from BASE-VER file.
17725 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17726 (gcc.o): Depend on $(BASEVER).
17727 * common.opt (dumpfullversion): New option.
17728 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17729 * doc/invoke.texi: Document -dumpfullversion.
17730 * doc/install.texi: Document --with-gcc-major-version-only.
17731 * configure: Regenerated.
17733 2017-01-17 Richard Biener <rguenther@suse.de>
17735 PR tree-optimization/71433
17736 * tree-vrp.c (register_new_assert_for): Merge same asserts
17737 on all incoming edges.
17738 (process_assert_insertions_for): Handle insertions at the
17741 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17743 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17744 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17746 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
17749 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
17752 2017-01-17 Alan Modra <amodra@gmail.com>
17755 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
17756 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
17757 symbolic stack limit when pic.
17759 2017-01-16 Martin Sebor <msebor@redhat.com>
17761 PR tree-optimization/78608
17762 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
17764 2017-01-16 Jeff Law <law@redhat.com>
17767 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
17768 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
17769 for several include directories that may be relative to sysroot.
17770 * config/i386/x-mingw32 (gplus_includedir): Define.
17771 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
17772 (native_system_includedir): Likewise.
17773 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
17774 override if TARGET_SYSTEM_ROOT is defined.
17775 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
17777 PR tree-optimization/79090
17778 PR tree-optimization/33562
17779 PR tree-optimization/61912
17780 PR tree-optimization/77485
17781 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
17782 and computed trims into the dump file.
17784 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17786 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
17788 2017-01-16 Jakub Jelinek <jakub@redhat.com>
17791 * gimplify.c (gimplify_init_constructor): If want_value and
17792 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
17796 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
17797 sequence. Formatting fixes.
17798 (doloop_optimize): Formatting fixes.
17801 * gcc.c (debug_level_greater_than_spec_func): New function.
17802 (static_spec_functions): Add debug-level-gt spec function.
17803 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
17805 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17806 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17807 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
17808 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
17809 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
17810 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
17812 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
17814 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
17815 QImode fixups to general and mask registers only.
17817 2017-01-16 Carl Love <cel@us.ibm.com>
17819 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
17820 for built-in functions
17821 vector signed char vec_nabs (vector signed char)
17822 vector signed short vec_nabs (vector signed short)
17823 vector signed int vec_nabs (vector signed int)
17824 vector signed long long vec_nabs (vector signed long long)
17825 vector float vec_nabs (vector float)
17826 vector double vec_nabs (vector double)
17827 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
17829 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
17830 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
17831 * doc/extend.texi: Update the documentation file for the new built-in
17834 2017-01-16 Martin Sebor <msebor@redhat.com>
17836 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
17839 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17841 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
17842 UNSPEC_VSX__XXSPLTD to require special splat handling.
17844 2017-01-16 David Malcolm <dmalcolm@redhat.com>
17847 * system.h: Poison strndup.
17849 2017-01-16 Alan Modra <amodra@gmail.com>
17852 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
17855 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
17857 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
17859 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
17861 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
17862 call recog here. Assert that INSN_CODE (insn) is non-negative.
17864 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17867 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
17869 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
17870 in the currently scheduled RTL fragment.
17872 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17874 PR rtl-optimization/78751
17875 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
17878 2017-01-14 Jeff Law <law@redhat.com>
17880 PR tree-optimization/79090
17881 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
17882 variable length stores.
17883 (compute_trims): Delete dead assignment to *trim_tail.
17884 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
17887 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
17889 PR rtl-optimization/78626
17890 PR rtl-optimization/78727
17891 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
17892 of a block, and split such blocks after everything else is finished.
17894 2017-01-14 Alan Modra <amodra@gmail.com>
17897 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
17898 target legitimate_combined_insn.
17899 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
17900 (rs6000_legitimate_combined_insn): New function.
17901 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
17903 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
17904 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
17905 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
17907 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
17909 * doc/frontends.texi (G++ and GCC): Remove references to Java.
17911 2017-01-13 Jeff Law <law@redhat.com>
17913 PR tree-optimization/33562
17914 PR tree-optimization/61912
17915 PR tree-optimization/77485
17916 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
17918 (delete_dead_assignment): Likewise.
17919 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
17920 statement to delete_dead_call and delete_dead_assignment.
17922 2017-01-13 David Malcolm <dmalcolm@redhat.com>
17925 * substring-locations.c (format_warning_va): Strengthen case 1 so
17926 that both endpoints of the substring must be within the format
17927 range for just the substring to be printed.
17929 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
17931 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
17932 * config/i386/i386.c (ix86_target_string): Add missing options
17933 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
17934 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
17935 flags_other and ix86_target_other to flags2_other. Display unknown
17937 (ix86_valid_target_attribute_inner_p): Add missing options and
17938 reorder options by implied ISAs, as in ix86_target_string.
17940 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17942 * hash-table.h (hash_table::too_empty_p): New function.
17943 (hash_table::expand): Use it.
17944 (hash_table::traverse): Likewise.
17945 (hash_table::empty_slot): Use sizeof (value_type) instead of
17946 sizeof (PTR) to convert bytes to elements. Shrink the table
17947 if the current size is excessive for the current number of
17950 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17952 * ira-costs.c (record_reg_classes): Break from the inner loop
17953 early once alt_fail is known to be true. Update outer loop
17954 handling accordingly.
17956 2017-01-13 Jeff Law <law@redhat.com>
17958 * tree-ssa-dse.c (decrement_count): New function.
17959 (increment_start_addr, maybe_trim_memstar_call): Likewise.
17960 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
17961 when we know the partially dead statement is a mem* function.
17963 PR tree-optimization/61912
17964 PR tree-optimization/77485
17965 * tree-ssa-dse.c: Include expr.h.
17966 (maybe_trim_constructor_store): New function.
17967 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
17969 PR tree-optimization/33562
17970 PR tree-optimization/61912
17971 PR tree-optimization/77485
17972 * doc/invoke.texi: Document new dse-max-object-size param.
17973 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
17974 * tree-ssa-dse.c: Include params.h.
17975 (dse_store_status): New enum.
17976 (initialize_ao_ref_for_dse): New, partially extracted from
17978 (valid_ao_ref_for_dse, normalize_ref): New.
17979 (setup_live_bytes_from_ref, compute_trims): Likewise.
17980 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
17981 (maybe_trim_partially_dead_store): Likewise.
17982 (maybe_trim_complex_store): Likewise.
17983 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
17984 Track what bytes live from the original store. Return tri-state
17985 for dead, partially dead or live.
17986 (dse_dom_walker): Add constructor, destructor and new private members.
17987 (delete_dead_call, delete_dead_assignment): New extracted from
17989 (dse_optimize_stmt): Make a member of dse_dom_walker.
17990 Use initialize_ao_ref_for_dse.
17992 PR tree-optimization/33562
17993 PR tree-optimization/61912
17994 PR tree-optimization/77485
17995 * sbitmap.h (bitmap_count_bits): Prototype.
17996 (bitmap_clear_range, bitmap_set_range): Likewise.
17997 * sbitmap.c (bitmap_clear_range): New function.
17998 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18000 2017-01-13 Martin Liska <mliska@suse.cz>
18003 * function.c (set_cfun): Add new argument force.
18004 * function.h (set_cfun): Likewise.
18005 * ipa-inline-transform.c (inline_call): Use the function when
18006 strict alising from is dropped for function we inline to.
18008 2017-01-13 Richard Biener <rguenther@suse.de>
18010 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18011 for dumping GIMPLE INTEGER_CSTs.
18013 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18015 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18016 to 201112L since C++17.
18018 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18021 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18022 if -fsanitize=kernel-address is present.
18024 2017-01-13 Richard Biener <rguenther@suse.de>
18026 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18027 as _Literal ( type ) number in case usual suffixes do not
18028 preserve all information.
18030 2017-01-13 Richard Biener <rguenther@suse.de>
18032 PR tree-optimization/77283
18033 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18034 and ssa-iterators.h.
18035 (is_feasible_trace): Implement a cost model based on joiner
18038 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18041 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18042 char or short to __float128/_Float128 directly.
18044 2017-01-12 Martin Sebor <msebor@redhat.com>
18046 to -Wformat-overflow.
18047 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18048 (min_bytes_remaining): Same.
18049 (get_string_length): Same.
18050 (format_string): Same.
18051 (format_directive): Same.
18053 (pass_sprintf_length::handle_gimple_call): Same.
18055 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18057 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18058 info.nowrite calls with no lhs that can't throw. Return bool
18059 whether gsi_remove has been called or not.
18060 (pass_sprintf_length::handle_gimple_call): Return bool whether
18061 try_substitute_return_value called gsi_remove. Formatting fix.
18062 (pass_sprintf_length::execute): Don't use gsi_remove if
18063 handle_gimple_call returned true.
18066 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18067 be removed due to side-effects, don't remove following barrier nor
18068 turn the successor edge into fallthru edge.
18070 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18073 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18074 element-reversing loads and stores as not swappable.
18076 2017-01-12 Nathan Sidwell <nathan@acm.org>
18077 Nicolai Stange <nicstange@gmail.com>
18079 * combine.c (try_combine): Don't ignore result of overlap checking
18080 loop. Combine overlap & asm check into single loop.
18082 2017-01-12 Richard Biener <rguenther@suse.de>
18084 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18085 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18087 2017-01-12 Richard Biener <rguenther@suse.de>
18089 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18090 and TS_TARGET_OPTION directly derive from TS_BASE.
18091 * tree-core.h (tree_optimization_option): Derive from tree_base.
18092 (tree_target_option): Likewise.
18094 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18096 * config/i386/i386.c (memory_address_length): Increase len
18097 only when rip_relative_addr_p returns false.
18099 2017-01-11 Julia Koval <julia.koval@intel.com>
18101 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18102 (OPTION_MASK_ISA_SGX_SET): New.
18103 (ix86_handle_option): Handle OPT_msgx.
18104 * config.gcc: Added sgxintrin.h.
18105 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18106 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18107 * config/i386/i386.c (ix86_target_string): Add -msgx.
18109 (ix86_option_override_internal): Handle new options.
18110 (ix86_valid_target_attribute_inner_p): Add sgx.
18111 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18112 * config/i386/i386.opt: Add msgx.
18113 * config/i386/sgxintrin.h: New file.
18114 * config/i386/x86intrin.h: Add sgxintrin.h.
18116 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18119 * fold-const.c (maybe_nonzero_address): Return 1 for function
18121 (tree_single_nonzero_warnv_p): Don't handle function local objects
18125 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18128 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18131 * opts.c: Include "spellcheck.h"
18132 (struct string_fragment): New struct.
18133 (struct edit_distance_traits<const string_fragment &>): New
18135 (get_closest_sanitizer_option): New function.
18136 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18138 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18140 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18142 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18143 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18144 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18145 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18146 for initial die_offset if dwarf_split_debug_info.
18147 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18148 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18150 (output_skeleton_debug_sections): Formatting fix. Use
18151 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18152 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18154 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18156 * config/arm/cortex-a53.md: Add bypasses for
18157 cortex_a53_r2f_cvt.
18158 (cortex_a53_r2f): Only use for transfers.
18159 (cortex_a53_f2r): Likewise.
18160 (cortex_a53_r2f_cvt): Add reservation for conversions.
18161 (cortex_a53_f2r_cvt): Likewise.
18163 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18165 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18166 to all inlined functions, change static to extern.
18168 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18171 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18173 (arm_assemble_integer): Likewise.
18175 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18177 * config.gcc: Use new awk script to check CPU, FPU and architecture
18178 parameters for --with-... options.
18179 * config/arm/parsecpu.awk: New file
18180 * config/arm/arm-cpus.in: New file.
18181 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18183 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18185 * config/arm/t-arm: Update dependency rules.
18186 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18187 of processing .def files.
18188 * config/arm/genopt.sh: Deleted.
18189 * config/arm/gentune.sh: Deleted.
18190 * config/arm/arm-cores.def: Deleted.
18191 * config/arm/arm-arches.def: Deleted.
18192 * config/arm/arm-fpus.def: Deleted.
18193 * config/arm/arm-tune.md: Regenerated.
18194 * config/arm/arm-tables.opt: Regenerated.
18195 * config/arm/arm-cpu.h: New generated file.
18196 * config/arm/arm-cpu-data.h: New generated file.
18197 * config/arm/arm-cpu-cdata.h: New generated file.
18199 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18202 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18204 (input_varpool_node): Unpack dynamically_initialized bit.
18206 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18208 PR rtl-optimization/79032
18209 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18210 the alignment of the adjusted memory reference against that of MODE,
18211 instead of the alignment of the original memory reference.
18213 2017-01-11 Martin Jambor <mjambor@suse.cz>
18215 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18217 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18218 decorated functions.
18220 2017-01-11 Richard Biener <rguenther@suse.de>
18222 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18223 set range/nonnull info for PHI results. Do not set it on
18224 stmts marked for removal.
18226 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18228 * expr.c (store_field): In the bitfield case, fetch the return value
18229 from the registers before applying a single big-endian adjustment.
18230 Always do a final load for a BLKmode value not larger than a word.
18232 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18235 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18236 that we correctly handle column numbers greater than
18237 LINE_MAP_MAX_COLUMN_NUMBER.
18239 2017-01-10 Martin Sebor <msebor@redhat.com>
18241 PR middle-end/78245
18242 * gimple-ssa-sprintf.c (get_destination_size): Call
18243 {init,fini}object_sizes.
18244 * tree-object-size.c (addr_object_size): Adjust.
18245 (pass_through_call): Adjust.
18246 (pass_object_sizes::execute): Adjust.
18247 * tree-object-size.h (fini_object_sizes): Declare.
18249 2017-01-10 Martin Sebor <msebor@redhat.com>
18251 PR tree-optimization/78775
18252 * builtins.c (get_size_range): Move...
18253 * calls.c: ...to here.
18254 (alloc_max_size): Accept zero argument.
18255 (operand_signed_p): Remove.
18256 (maybe_warn_alloc_args_overflow): Call get_size_range.
18257 * calls.h (get_size_range): Declare.
18259 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18261 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18262 from TI's devices.csv file as of September 2016.
18263 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18265 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18267 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18268 * doc/invoke.texi: Likewise.
18269 * doc/md.texi: Likewise.
18270 * doc/objc.texi: Likewise.
18272 2017-01-10 Joshua Conner <joshconner@google.com>
18274 * config/arm/fuchsia-elf.h: New file.
18275 * config/fuchsia.h: New file.
18276 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18277 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18279 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18281 2016-01-10 Richard Biener <rguenther@suse.de>
18283 PR tree-optimization/79034
18284 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18285 Propagate out degenerate PHIs in the joiner.
18287 2017-01-10 Martin Liska <mliska@suse.cz>
18289 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18290 (sort_congruence_classes_by_decl_uid): Likewise.
18291 (sort_congruence_class_groups_by_decl_uid): Likewise.
18292 (sem_item_optimizer::merge_classes): Sort class, groups in these
18293 classes and members in the groups by DECL_UID of declarations.
18294 This would make merge operations stable.
18296 2017-01-10 Martin Liska <mliska@suse.cz>
18298 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18299 usage of m_classes_vec.
18300 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18301 (sem_item_optimizer::get_group_by_hash): Likewise.
18302 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18303 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18304 (sem_item_optimizer::verify_classes): Likewise.
18305 (sem_item_optimizer::process_cong_reduction): Likewise.
18306 (sem_item_optimizer::dump_cong_classes): Likewise.
18307 (sem_item_optimizer::merge_classes): Likewise.
18308 * ipa-icf.h (congruence_class_hash): Rename from
18309 congruence_class_group_hash. Remove declaration of m_classes_vec.
18311 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18313 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18314 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18315 * config.gcc: Add avx512vpopcntdqintrin.h.
18316 * config/i386/avx512vpopcntdqintrin.h: New.
18317 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18318 * config/i386/i386-builtin-types.def: Add new types.
18319 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18320 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18321 __builtin_ia32_vpopcountq_v8di_mask): New.
18322 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18323 __AVX512VPOPCNTDQ__.
18324 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18325 (PTA_AVX512VPOPCNTDQ): Define.
18326 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18327 TARGET_AVX512VPOPCNTDQ_P): Define.
18328 * config/i386/i386.opt: Add mavx512vpopcntdq.
18329 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18330 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18332 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18334 PR middle-end/77484
18335 * predict.def (PRED_CALL): Set to 67.
18337 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18339 * expr.c (store_field): In the bitfield case, if the value comes from
18340 a function call and is of an aggregate type returned in registers, do
18341 not modify the field mode; extract the value in all cases if the mode
18342 is BLKmode and the size is not larger than a word.
18344 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18347 * config/i386/cpuid.h: Fix undefined behavior.
18349 2017-01-04 Jeff Law <law@redhat.com>
18351 PR tree-optimization/79007
18352 PR tree-optimization/67955
18353 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18354 conservative for pt.null when flag_non_call_exceptions is on.
18356 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18358 PR translation/79019
18359 PR translation/79020
18360 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18361 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18362 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18364 * config/avr/avr.opt (maccumulate-args): Likewise.
18365 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18366 * common.opt (freport-bug): Likewise.
18367 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18368 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18369 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18370 translatable string.
18371 * config/i386/i386.c (function_value_32): Likewise.
18372 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18373 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18375 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18376 * common/config/msp430/msp430-common.c (msp430_handle_option):
18378 * symtab.c (symtab_node::verify_base): Likewise.
18379 * opts.c (set_debug_level): Likewise.
18380 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18381 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18382 missing whitespace to translatable strings.
18383 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18384 * config/sh/superh.h: Likewise.
18385 * config/i386/xopintrin.h: Likewise.
18386 * config/i386/znver1.md: Likewise.
18387 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18388 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18389 * double-int.h (struct double_int): Likewise.
18390 * double-int.c (div_and_round_double): Likewise.
18391 * wide-int.cc: Likewise.
18392 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18393 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18394 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18395 (crossjumps_occurred): ... this.
18396 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18399 PR tree-optimization/78899
18400 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18401 returning bool return struct loop *, NULL for failure and the new
18403 (versionable_outer_loop_p): Don't version outer loop if it has
18404 dont_vectorized bit set.
18405 (tree_if_conversion): When versioning outer loop, ensure
18406 tree_if_conversion is performed also on the inner loop of the
18407 non-vectorizable outer loop copy.
18408 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18409 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18410 prevent vectorization of it.
18411 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18412 the outer loop vectorization of the non-scalar version is attempted
18413 before vectorization of the inner loop in scalar version. If
18414 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18415 vectorization of its inner loop.
18416 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18417 has 2 inner loops, rename also on edges from bb whose single pred
18418 is outer_loop->header. Fix typo in function comment.
18420 2017-01-09 Martin Sebor <msebor@redhat.com>
18423 * asan.c (asan_emit_stack_protection): Increase local buffer size
18424 to avoid snprintf truncation warning.
18426 2017-01-09 Andrew Pinski <apinski@cavium.com>
18428 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18429 to reference thunderx2t99 for the tuning structure
18430 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18432 (thunderx2t99_extra_costs): This.
18433 * config/aarch64/aarch64-tune.md: Regenerate.
18434 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18435 (vulcan_addrcost_table): This.
18436 (vulcan_regmove_cost): Rename to ...
18437 (thunderx2t99_regmove_cost): This.
18438 (vulcan_vector_cost): Rename to ...
18439 (thunderx2t99_vector_cost): this.
18440 (vulcan_branch_cost): Rename to ...
18441 (thunderx2t99_branch_cost): This.
18442 (vulcan_tunings): Rename to ...
18443 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18444 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18446 2017-01-09 Martin Jambor <mjambor@suse.cz>
18450 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18451 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18452 (propagate_vr_accross_jump_function): Use the above function for all
18453 value range computations for pass-through jump functions and type
18454 converasion from explicit value range values.
18455 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18456 parameters from TYPE_ARG_TYPES.
18457 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18458 (ipa_write_node_info): Stream type of the actual argument.
18459 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18461 2017-01-09 Martin Liska <mliska@suse.cz>
18464 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18465 (lookup_compiler): Do not show error message with have_E.
18467 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18469 PR tree-optimization/78938
18470 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18471 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18472 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18473 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18476 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18478 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18481 2017-01-09 Richard Biener <rguenther@suse.de>
18483 PR tree-optimization/78997
18484 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18485 name condition properly.
18487 2017-01-09 Richard Biener <rguenther@suse.de>
18490 * dwarf2out.c (is_cxx): New overload with context.
18491 (is_naming_typedef_decl): Use it.
18493 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18495 * invoke.texi (Option Summary): Correct spacing in option lists
18496 and add line breaks to fix over-long lines.
18498 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18500 PR middle-end/17660
18502 * extend.texi (Common Variable Attributes): Add xref to GCC
18503 Internals manual to explain mode attribute keywords.
18505 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18508 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18509 and Preprocessor Options.
18510 (Options for Linking): Document -pthread here....
18511 (RS/6000 and PowerPC Options): ...not here.
18512 (Solaris 2 Options): ...or here.
18513 * doc/cppopts.texi: Document -pthread.
18515 2017-01-08 Martin Sebor <msebor@redhat.com>
18517 PR middle-end/77708
18518 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18519 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18520 New member functions.
18521 (format_directive): Used them.
18523 (pass_sprintf_length::handle_gimple_call): Same.
18524 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18525 to avoid truncation for any argument.
18526 (extract_affine_mul): Same.
18527 * tree.c (get_file_function_name): Same.
18529 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18531 PR middle-end/77484
18532 * predict.def (PRED_INDIR_CALL): Set to 86.
18534 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18536 PR preprocessor/54124
18537 * doc/cppopts.texi: Reformat -d subtable to list the full name
18538 of the options. Add cross-reference to the docs for the general
18539 compiler -d options.
18540 * doc/invoke.texi (Developer Options): Add cross-reference to the
18541 preprocessor-specific -d option documentation.
18543 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18545 PR preprocessor/13498
18546 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18547 redudant material, and reflect new command-line options.
18548 (System Headers): Likewise.
18550 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18552 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18553 -isystem, and -idirafter. Copy-edit.
18554 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18555 default for -ftrack-macro-expansion. Delete obsolete and
18556 badly-formatted implementation details about -fdebug-cpp output.
18557 * doc/cppwarnopts.texi: Copy-edit.
18559 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18562 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18563 that the transition from a max line width >= 1<<10 to narrower
18564 lines works correctly.
18566 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18568 * doc/options.texi (PerFunction): New.
18569 * opt-functions.awk (switch_flags): Map both Optimization and
18570 PerFunction to CL_OPTIMIZATION.
18571 * opth-gen.awk: Test for PerFunction flag along with
18573 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18574 it only when the latter is present. Skip those that don't in
18575 the hash function generator.
18576 * common.opt (fvar-tracking): Mark as PerFunction instead of
18578 (fvar-tracking-assignments): Likewise.
18579 (fvar-tracking-assignments-toggle): Likewise.
18580 (fvar-tracking-uninit): Likewise.
18582 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18584 PR translation/79018
18585 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18588 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18591 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18592 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18593 TARGET_LONG_JUMP_TABLE_OFFSETS.
18594 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18595 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18596 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18597 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18598 * config/m68k/m68k.md (tablejump expander): Likewise.
18599 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18600 TARGET_LONG_JUMP_TABLE_OFFSETS.
18601 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18602 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18604 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18605 David Holsgrove <david.holsgrove@xilinx.com>
18607 * common/config/microblaze/microblaze-common.c
18608 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18609 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18611 * config/microblaze/microblaze.c (microblaze_must_save_register)
18612 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18614 (microblaze_eh_return): New function.
18615 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18616 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18617 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18618 * config/microblaze/microblaze.md (eh_return): New pattern.
18620 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18622 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18623 GCC_DIAGNOSTIC_STRINGIFY): Define.
18625 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18627 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18629 * config/arm/arm.md (<mcrr>): New.
18631 * config/arm/arm.c (arm_arch5te): New.
18632 (arm_option_override): Set arm_arch5te.
18633 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18635 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18636 (arm_mcrr_qualifiers): ... this. New.
18637 (MRRC_QUALIFIERS): Define to...
18638 (arm_mrrc_qualifiers): ... this. New.
18639 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18641 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18642 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18643 (MRRCI, mrrc, MRRC): New.
18644 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18645 VUNSPEC_MRRC2): New.
18647 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18649 * config/arm/arm.md (<mcr>): New.
18651 * config/arm/arm.c (arm_coproc_builtin_available): Add
18652 support for mcr, mrc, mcr2 and mrc2.
18653 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18654 (arm_mcr_qualifiers): ... this. New.
18655 (MRC_QUALIFIERS): Define to ...
18656 (arm_mrc_qualifiers): ... this. New.
18657 (MCR_QUALIFIERS): Define to ...
18658 (arm_mcr_qualifiers): ... this. New.
18659 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18661 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18662 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18663 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18664 VUNSPEC_MRC2): New.
18666 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18668 * config/arm/arm.md (*ldc): New.
18672 * config/arm/arm.c (arm_coproc_builtin_available): Add
18673 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18674 (arm_coproc_ldc_stc_legitimate_address): New.
18675 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18676 'qualifier_const_pointer'.
18677 (LDC_QUALIFIERS): Define to...
18678 (arm_ldc_qualifiers): ... this. New.
18679 (STC_QUALIFIERS): Define to...
18680 (arm_stc_qualifiers): ... this. New.
18681 * config/arm/arm-protos.h
18682 (arm_coproc_ldc_stc_legitimate_address): New.
18683 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18684 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18685 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18686 stc2, stcl, stc2l): New.
18687 * config/arm/constraints.md (Uz): New.
18688 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18689 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18690 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18691 VUNSPEC_STC2L): New.
18693 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18695 * config/arm/arm.md (<cdp>): New.
18696 * config/arm/arm.c (neon_const_bounds): Rename this ...
18697 (arm_const_bounds): ... this.
18698 (arm_coproc_builtin_available): New.
18699 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18700 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18701 (CDP_QUALIFIERS): Define to...
18702 (arm_cdp_qualifiers): ... this. New.
18704 (arm_expand_builtin_args): Add case for 6 arguments.
18705 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18706 (arm_const_bounds): ... this.
18707 (arm_coproc_builtin_available): New.
18708 * config/arm/arm_acle.h (__arm_cdp): New.
18710 * config/arm/arm_acle_builtins.def (cdp): New.
18712 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18713 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18714 'arm_const_bounds'.
18715 * config/arm/types.md (coproc): New.
18716 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18717 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18718 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18719 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18721 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18723 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18724 (UBINOP_QUALIFIERS): New.
18726 (acle_builtin_data): New. Change comment.
18727 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18728 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18729 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18730 arm_acle_builtins.def.
18731 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18732 (arm_init_acle_builtins): New.
18733 (CRC32_BUILTIN): Remove.
18734 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18735 crc32cb, crc32ch and crc32cw.
18736 (arm_init_crc32_builtins): Remove.
18737 (arm_init_builtins): Use arm_init_acle_builtins rather
18738 than arm_init_crc32_builtins.
18739 (arm_expand_acle_builtin): New.
18740 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18741 * config/arm/arm_acle_builtins.def: New.
18743 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18745 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18746 (arm_builtin_datum): ... this.
18747 (arm_init_neon_builtin): Rename to ...
18748 (arm_init_builtin): ... this. Add a new parameters PREFIX
18749 and USE_SIG_IN_NAME.
18750 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
18751 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
18752 'arm_builtin_datum'.
18753 (arm_init_vfp_builtins): Likewise.
18754 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
18755 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
18756 (arm_expand_neon_args): Rename to ...
18757 (arm_expand_builtin_args): ... this. Rename builtin_arg
18758 enum values and differentiate between ARG_BUILTIN_MEMORY
18759 and ARG_BUILTIN_NEON_MEMORY.
18760 (arm_expand_neon_builtin_1): Rename to ...
18761 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
18762 values, arm_expand_builtin_args and add bool parameter NEON.
18763 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
18764 (arm_expand_vfp_builtin): Likewise.
18765 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
18767 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18769 PR middle-end/77484
18770 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
18771 * predict.c (tree_estimate_probability_bb): Reverse direction of
18772 polymorphic call predictor.
18774 2017-01-06 David Malcolm <dmalcolm@redhat.com>
18776 * passes.c (execute_one_pass): Split out pass-skipping logic into...
18777 (determine_pass_name_match): ...this new function and...
18778 (should_skip_pass_p): ...this new function.
18780 2017-01-06 Nathan Sidwell <nathan@acm.org>
18782 * ipa-visibility.c (function_and_variable_visibility): Reformat
18783 comments and long lines. Remove extrneous if.
18784 * symtab.c (symtab_node::make_decl_local): Fix code format.
18785 (symtab_node::set_section_for_node): Fix comment typo.
18787 2017-01-06 Martin Liska <mliska@suse.cz>
18790 * lra-constraints.c: Rename invariant to lra_invariant.
18791 * predict.c (set_even_probabilities): Initialize e to NULL.
18793 2017-01-05 Martin Sebor <msebor@redhat.com>
18795 PR tree-optimization/78910
18796 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
18797 (format_integer): Correct off-by-one error in the handling
18798 of precision with negative numbers in signed conversions..
18800 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
18802 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
18804 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18806 PR tree-optimization/71016
18807 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
18808 factor_out_conditional_conversion. Formatting fix.
18809 (factor_out_conditional_conversion): Add cond_stmt argument.
18810 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
18811 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
18814 2017-01-05 David Malcolm <dmalcolm@redhat.com>
18816 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
18817 read-rtl-function.o, and selftest-rtl.o.
18818 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
18819 (selftest::aarch64_test_loading_full_dump): New function.
18820 (selftest::aarch64_run_selftests): New function.
18821 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
18822 selftest::aarch64_run_selftests.
18823 * config/i386/i386.c
18824 (selftest::ix86_test_loading_dump_fragment_1): New function.
18825 (selftest::ix86_test_loading_call_insn): New function.
18826 (selftest::ix86_test_loading_full_dump): New function.
18827 (selftest::ix86_test_loading_unspec): New function.
18828 (selftest::ix86_run_selftests): Call the new functions.
18829 * emit-rtl.c (maybe_set_max_label_num): New function.
18830 * emit-rtl.h (maybe_set_max_label_num): New decl.
18831 * function.c (instantiate_decls): Guard call to
18832 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
18833 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
18835 * gensupport.c (gen_reader::gen_reader): Pass "false"
18836 for new "compact" param of rtx_reader.
18837 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
18838 rather than an empty string for NULL strings.
18839 * read-md.c: Potentially include config.h rather than bconfig.h.
18840 Wrap include of errors.h with #ifdef GENERATOR_FILE.
18841 (have_error): New global, copied from errors.c.
18842 (md_reader::read_name): Rename to...
18843 (md_reader::read_name_1): ...this, adding "out_loc" param,
18844 and converting "missing name or number" to returning false, rather
18846 (md_reader::read_name): Reimplement in terms of read_name_1.
18847 (md_reader::read_name_or_nil): New function.
18848 (md_reader::read_string): Handle "(nil)" by returning NULL.
18849 (md_reader::md_reader): Add new param "compact".
18850 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
18851 (md_reader::read_file): New method.
18852 * read-md.h (md_reader::md_reader): Add new param "compact".
18853 (md_reader::read_file): New method.
18854 (md_reader::is_compact): New accessor.
18855 (md_reader::read_name): Convert return type from void to file_location.
18856 (md_reader::read_name_or_nil): New decl.
18857 (md_reader::read_name_1): New decl.
18858 (md_reader::m_compact): New field.
18859 (noop_reader::noop_reader): Pass "false" for new "compact" param
18861 (rtx_reader::rtx_reader): Add new "compact" param.
18862 (rtx_reader::read_rtx_operand): Make virtual and convert return
18863 type from void to rtx.
18864 (rtx_reader::read_until): New decl.
18865 (rtx_reader::handle_any_trailing_information): New virtual function.
18866 (rtx_reader::postprocess): New virtual function.
18867 (rtx_reader::finalize_string): New virtual function.
18868 (rtx_reader::m_in_call_function_usage): New field.
18869 (rtx_reader::m_reuse_rtx_by_id): New field.
18870 * read-rtl-function.c: New file.
18871 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
18872 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
18873 (selftest::verify_three_block_rtl_cfg): New decl.
18874 * read-rtl-function.h: New file.
18875 * read-rtl.c: Potentially include config.h rather than bconfig.h.
18876 For host, include function.h, memmodel.h, and emit-rtl.h.
18877 (one_time_initialization): New function.
18878 (struct compact_insn_name): New struct.
18879 (compact_insn_names): New array.
18880 (find_code): Handle insn codes in compact dumps.
18881 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
18882 (bind_subst_iter_and_attr): Likewise.
18883 (add_condition_to_string): Likewise.
18884 (add_condition_to_rtx): Likewise.
18885 (apply_attribute_uses): Likewise.
18886 (add_current_iterators): Likewise.
18887 (apply_iterators): Likewise.
18888 (initialize_iterators): Guard usage of apply_subst_iterator with
18889 #ifdef GENERATOR_FILE.
18890 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
18891 (md_reader::read_mapping): Likewise.
18892 (add_define_attr_for_define_subst): Likewise.
18893 (add_define_subst_attr): Likewise.
18894 (read_subst_mapping): Likewise.
18895 (check_code_iterator): Likewise.
18896 (rtx_reader::read_rtx): Likewise. Move one-time initialization
18898 (one_time_initialization): New function.
18899 (rtx_reader::read_until): New method.
18900 (read_flags): New function.
18901 (parse_reg_note_name): New function.
18902 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
18903 Handle reuse_rtx ids.
18904 Wrap iterator lookup within #ifdef GENERATOR_FILE.
18905 Add parsing support for RTL dumps, mirroring the special-cases in
18906 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
18907 values, and calling handle_any_trailing_information.
18908 (rtx_reader::read_rtx_operand): Convert return type from void
18909 to rtx, returning return_rtx. Handle case 'e'. Call
18910 finalize_string on XSTR and XTMPL fields.
18911 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
18912 "(nil)" values were omitted. Call the postprocess vfunc on the
18914 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
18915 class ctor. Initialize m_in_call_function_usage. Call
18916 one_time_initialization.
18917 * rtl-tests.c (selftest::test_uncond_jump): Call
18918 set_new_first_and_last_insn.
18919 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
18920 * selftest-rtl.c: New file.
18921 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
18922 (selftest::get_insn_by_uid): New decl.
18923 * selftest-run-tests.c (selftest::run_tests): Call
18924 read_rtl_function_c_tests.
18925 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
18926 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
18929 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
18931 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
18932 operands in a special way. Assert that pos+len <= mode precision.
18934 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18936 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
18937 3 argument Alias with unlimited for the negative form.
18938 (fno-vect-cost-model): Removed.
18940 2017-01-05 Martin Liska <mliska@suse.cz>
18942 * hsa-gen.c (gen_hsa_divmod): New function.
18943 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
18945 2017-01-05 Martin Liska <mliska@suse.cz>
18948 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
18951 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18953 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
18954 small constant length operands.
18956 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18958 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
18959 between loop iterations.
18961 2017-01-05 Martin Liska <mliska@suse.cz>
18964 * gimplify.c (gimplify_decl_expr): Compare to
18965 asan_poisoned_variables instread of checking flags.
18966 (gimplify_target_expr): Likewise.
18967 (gimplify_expr): Likewise.
18968 (gimplify_function_tree): Conditionally initialize
18969 asan_poisoned_variables.
18971 2017-01-04 Jeff Law <law@redhat.com>
18973 PR tree-optimizatin/78812
18974 * rtl.h (contains_mem_rtx_p): Prototype.
18975 * ifcvt.c (containts_mem_rtx_p): Move from here to...
18976 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
18977 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
18978 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
18979 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
18981 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18983 * input.c (assert_char_at_range): Default-initialize actual_range.
18985 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18987 * df-scan.c (df_ref_create_structure): Make regno unsigned,
18988 to match the caller.
18990 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18992 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
18993 insns after final jump in test to emit dummy move.
18995 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18997 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
18998 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19000 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19002 * multiple_target.c (create_dispatcher_calls): Init e_next.
19003 * tree-ssa-loop-split.c (split_loop): Init border.
19004 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19007 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19012 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19013 (altivec_register_operand): Do not return true if the operand
19014 contains a SUBREG mixing SImode and SFmode.
19015 (vsx_register_operand): Likewise.
19016 (vsx_reg_sfsubreg_ok): New predicate.
19017 (vfloat_operand): Do not return true if the operand contains a
19018 SUBREG mixing SImode and SFmode.
19019 (vint_operand): Likewise.
19020 (vlogical_operand): Likewise.
19021 (gpc_reg_operand): Likewise.
19022 (int_reg_operand): Likewise.
19023 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19024 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19025 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19027 (rs6000_emit_move_si_sf_subreg): New helper function.
19028 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19029 fixup SUBREGs involving SImode and SFmode.
19030 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19031 numbers for the new peephole2 optimization.
19032 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19033 the GLIBC math library that do AND/IOR/XOR operations on single
19034 precision floating point.
19035 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19036 target macros to say whether we need to avoid SUBREGs mixing
19038 (TARGET_ALLOW_SF_SUBREG): Likewise.
19039 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19040 (UNSPEC_SI_FROM_SF): Likewise.
19041 (iorxor): Change spacing.
19042 (and_ior_xor): New iterator for AND, IOR, and XOR.
19043 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19044 (movdi_from_sf_zero_ext): Likewise.
19045 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19046 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19047 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19048 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19049 (fms<mode>4): Likewise.
19050 (fnma<mode>4): Likewise.
19051 (fnms<mode>4): Likewise.
19052 (nfma<mode>4): Likewise.
19053 (nfms<mode>4): Likewise.
19055 2017-01-04 Marek Polacek <polacek@redhat.com>
19058 * doc/invoke.texi: Document -Wpointer-compare.
19060 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19062 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19065 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19066 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19069 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19072 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19073 documentation of the powerpc_popcntb_ok attribute.
19074 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19075 code to issue warning messages if a requested CPU configuration is
19076 not supported by the binary (assembler and loader) toolchain.
19077 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19078 made to define a built-in function that has been disabled.
19079 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19080 made to define a built-in function that has been disabled.
19081 (altivec_init_builtins): Add comment explaining why definition
19082 of the DST built-in functions is not preceded by an assertion
19083 check. Add assertions to prevent ICE if attempts are made to
19084 define an altivec predicate or an abs* built-in function that has
19086 (htm_init_builtins): Add comment explaining why definition of the
19087 htm built-in functions is not preceded by an assertion check.
19089 2017-01-04 Jeff Law <law@redhat.com>
19091 PR tree-optimizatin/67955
19092 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19093 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19094 the points-to solution does not include pt_null. Use DECL_PT_UID
19097 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19099 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19100 Use gen_int_mode instead of gen_lopwart for const_int operands.
19102 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19104 PR tree-optimization/71563
19105 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19106 out of range value - has low bits known to be zero.
19108 2017-01-04 Alan Modra <amodra@gmail.com>
19110 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19111 * configure: Regenerate.
19112 * config.in: Regenerate.
19114 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19117 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19118 a substring of the message, but strcmp with the whole message. Ifdef
19119 ENABLE_NLS, translate the message first using dgettext.
19121 2017-01-03 Jeff Law <law@redhat.com>
19123 PR tree-optimizatin/78856
19124 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19125 (mark_threaded_blocks): Remove code to truncate thread paths that
19126 cross multiple loop headers. Instead invalidate the cached loop
19127 iteration information and handle case of a thread path walking
19128 into an irreducible region.
19130 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19133 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19134 assertions. Add support for doing the signbit if the IEEE 128-bit
19135 floating point value is in a GPR.
19136 * config/rs6000/rs6000.md (Fsignbit): Delete.
19137 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19138 Update the length attribute if the value is in a GPR.
19139 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19140 the sign or zero extension instruction, since the value is always 0/1.
19141 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19144 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19145 extracting SImode to a GPR register so that we can generate a
19146 store, limit the vector to be in a traditional Altivec register
19147 for the vextuwrx instruction.
19149 2017-01-03 Ian Lance Taylor <iant@google.com>
19151 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19153 2017-01-03 Martin Sebor <msebor@redhat.com>
19155 PR tree-optimization/78696
19156 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19157 precision. Use MPFR for %f for greater fidelity. Correct handling
19159 (pass_sprintf_length::compute_format_length): Set width and precision
19160 specified by asrerisk to void_node for vararg functions.
19161 (try_substitute_return_value): Adjust dump output.
19163 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19165 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19167 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19169 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19170 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19171 -mlra/-mno-lra was passed to the compiler.
19173 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19175 PR rtl-optimization/65618
19176 * emit-rtl.c (try_split): Move initialization of "before" and
19177 "after" to just before the call to emit_insn_after_setloc.
19179 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19181 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19183 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19185 * dwarf2out.c (gen_enumeration_type_die): When
19186 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19188 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19190 PR tree-optimization/78965
19191 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19192 Change first argument from const call_info & to call_info &. For %n
19193 set info.nowrite to false.
19195 PR middle-end/78901
19196 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19197 possibly throwing calls.
19199 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19200 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19201 and fns handling, rather than in a separate case SSA_NAME.
19203 2017-01-02 Jeff Law <law@redhat.com>
19205 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19206 fixes for first_period and second_period variables.
19208 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19211 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19212 (*insvqi_1): New insn pattern.
19213 (*insvqi_1_mem_rex64): Ditto.
19214 (*insvqi_2): Ditto.
19215 (*insvqi_3): Rename from *insvqi.
19217 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19219 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19221 * doc/cfg.texi (Edges): Remove reference to Java.
19222 (Maintaining the CFG): Ditto.
19224 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19226 PR middle-end/77674
19227 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19228 transparent aliases.
19230 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19232 PR middle-end/77484
19233 * predict.def (PRED_CALL): Update hitrate.
19234 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19235 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19236 into direct/indirect/polymorphic variants.
19238 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19240 Update copyright years.
19242 * gcc.c (process_command): Update copyright notice dates.
19243 * gcov-dump.c (print_version): Ditto.
19244 * gcov.c (print_version): Ditto.
19245 * gcov-tool.c (print_version): Ditto.
19246 * gengtype.c (create_file): Ditto.
19247 * doc/cpp.texi: Bump @copying's copyright year.
19248 * doc/cppinternals.texi: Ditto.
19249 * doc/gcc.texi: Ditto.
19250 * doc/gccint.texi: Ditto.
19251 * doc/gcov.texi: Ditto.
19252 * doc/install.texi: Ditto.
19253 * doc/invoke.texi: Ditto.
19255 Copyright (C) 2017 Free Software Foundation, Inc.
19257 Copying and distribution of this file, with or without modification,
19258 are permitted in any medium without royalty provided the copyright
19259 notice and this notice are preserved.