1 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7 alternative always use QI mode, for -Os imov (=R,R) alternative
8 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12 Jeff Law <law@redhat.com>
14 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
15 (allocate_stack_space): New function, partially extracted from
17 (s390_emit_prologue): Track offset to most recent stack probe.
18 Code to allocate space moved into allocate_stack_space.
19 Dump actions when no stack is allocated.
20 (s390_prologue_plus_offset): New function.
21 (s390_emit_stack_probe): Likewise.
23 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
25 * common.opt (Wa, Wl, Wp, g, gz=): Add
27 (gno-column-info): Remove.
28 (gcolumn-info): Drop RejectNegative.
30 (gno-record-gcc-switches): Remove.
31 (grecord-gcc-switches): Drop RejectNegative.
32 (gno-split-dwarf): Remove.
33 (gsplit-dwarf): Drop RejectNegative.
34 (gno-strict-dwarf): Remove.
35 (gstrict-dwarf): Drop RejectNegative.
36 * config/darwin.opt (gfull, gused): Add RejectNegative.
37 * dwarf2out.c (gen_producer_string): Drop
38 gno-record-gcc-switches handler.
39 * optc-gen.awk: Add g to prefixes with negative forms.
40 * opts-common.c (remapping_prefix_p): New.
42 (generate_canonical_option): Test g prefix.
43 (option_map): Add -gno- mapping.
44 (add_misspelling_candidates): Check remapping_prefix_p.
46 2017-09-20 Jeff Law <law@redhat.com>
48 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
49 thinko in stack clash protection support.
51 * explow.c (compute_stack_clash_protection_loop_data): Use
52 CONST_INT_P instead of explicit test. Verify object is a
53 CONST_INT_P before looking at INTVAL.
54 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
55 instead of explicit test.
57 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
60 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
61 address instead of to r1 and r11.
63 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
65 * config.gcc: Support "knm".
66 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
67 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
69 * config/i386/i386.c (m_KNM): Define.
70 (processor_target_table): Add "knm".
72 (ix86_option_override_internal): Add "knm".
73 (ix86_issue_rate): Add PROCESSOR_KNM.
74 (ix86_adjust_cost): Ditto.
75 (ia32_multipass_dfa_lookahead): Ditto.
76 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
77 (fold_builtin_cpu): Add M_INTEL_KNM.
78 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
79 (processor_type): Add PROCESSOR_KNM.
80 * config/i386/x86-tune.def: Add m_KNM.
81 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
83 2017-09-20 Richard Biener <rguenther@suse.de>
85 PR tree-optimization/80213
86 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
87 are allowed in empty BBs as well.
88 (canonicalize_loop_closed_ssa): Also look for other complex
90 (scop_detection::get_sese): Include the loop-closed PHI block
92 (scop_detection::merge_sese): Remove code adding extra blocks.
93 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
94 (build_scops): Assert the final returned scop is invalid.
96 2017-09-20 Richard Biener <rguenther@suse.de>
98 PR tree-optimization/82264
99 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
101 (vn_phi_lookup): Likewise.
102 (vn_phi_insert): Likewise.
104 2017-09-20 Jakub Jelinek <jakub@redhat.com>
106 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
107 that fits into uhwi or shwi, add DW_AT_const_value regardless
108 of early_dwarf without going through RTL, using add_AT_unsigned
111 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
112 (DEBUG_LTO_ABBREV_SECTION): Likewise.
113 (DEBUG_LTO_MACINFO_SECTION): Likewise.
114 (DEBUG_MACRO_SECTION): Likewise.
115 (DEBUG_LTO_MACRO_SECTION): Likewise.
116 (DEBUG_STR_DWO_SECTION): Likewise.
117 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
118 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
119 (DEBUG_LTO_DWO_LINE_SECTION): Define.
120 (DEBUG_LTO_LINE_STR_SECTION): Define.
121 (init_sections_and_labels): Initialize debug_line_str_section
122 variable. Initialize debug_loc_section for -gdwarf-5 to
123 DEBUG_LOCLISTS_SECTION. Formatting fixes.
125 2017-09-20 Richard Biener <rguenther@suse.de>
127 * graphite-sese-to-poly.c (extract_affine): Properly handle
128 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
130 2017-09-20 Richard Biener <rguenther@suse.de>
132 PR tree-optimization/81373
133 * graphite-scop-detection.c (build_cross_bb_scalars_def):
134 Force SESE live-out defs to be handled even if they are
137 2017-09-19 Jeff Law <law@redhat.com>
139 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
140 nothing for stack adjustments with REG_STACK_CHECK.
141 * sched-deps.c (parse_add_or_inc): Reject insns with
142 REG_STACK_CHECK from dependency breaking.
143 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
144 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
145 * reg-notes.def (STACK_CHECK): New note.
147 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
148 (ix86_expand_prologue): Dump stack clash info as needed.
149 Call ix86_adjust_stack_and_probe_stack_clash as needed.
151 * function.c (dump_stack_clash_frame_info): New function.
152 * function.h (dump_stack_clash_frame_info): Prototype.
153 (enum stack_clash_probes): New enum.
155 * config/alpha/alpha.c (alpha_expand_prologue): Also check
156 flag_stack_clash_protection.
157 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
158 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
159 (arm_frame_pointer_required): Likewise.
160 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
161 (ia64_expand_prologue): Likewise.
162 * config/mips/mips.c (mips_expand_prologue): Likewise.
163 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
164 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
165 (sparc_flat_expand_prologue): Likewise.
166 * config/spu/spu.c (spu_expand_prologue): Likewise.
168 * explow.c: Include "params.h".
169 (anti_adjust_stack_and_probe_stack_clash): New function.
170 (get_stack_check_protect): Likewise.
171 (compute_stack_clash_protection_loop_data): Likewise.
172 (emit_stack_clash_protection_loop_start): Likewise.
173 (emit_stack_clash_protection_loop_end): Likewise.
174 (allocate_dynamic_stack_space): Use get_stack_check_protect.
175 Use anti_adjust_stack_and_probe_stack_clash.
176 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
177 (emit_stack_clash_protection_loop_start): Likewise.
178 (emit_stack_clash_protection_loop_end): Likewise.
179 * rtl.h (get_stack_check_protect): Prototype.
180 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
181 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
182 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
184 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
186 * doc/tm.texi: Rebuilt.
187 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
188 get_stack_check_protect.
189 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
190 * config/arm/arm.c (arm_expand_prologue): Likewise.
191 (arm_frame_pointer_required): Likewise.
192 * config/i386/i386.c (ix86_expand_prologue): Likewise.
193 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
194 * config/mips/mips.c (mips_expand_prologue): Likewise.
195 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
196 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
197 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
198 (sparc_flat_expand_prologue): Likewise.
200 * common.opt (-fstack-clash-protection): New option.
201 * flag-types.h (enum stack_check_type): Note difference between
202 -fstack-check= and -fstack-clash-protection.
203 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
204 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
205 * toplev.c (process_options): Issue warnings/errors for cases
206 not handled with -fstack-clash-protection.
207 * doc/invoke.texi (-fstack-clash-protection): Document new option.
208 (-fstack-check): Note additional problem with -fstack-check=generic.
209 Note that -fstack-check is primarily for Ada and refer users
210 to -fstack-clash-protection for stack-clash-protection.
211 Document new params for stack clash protection.
213 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
215 * config/i386/i386.c (ix86_split_long_move): Do not handle
216 address used for LEA in a special way.
218 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
220 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
222 2017-09-19 Martin Sebor <msebor@redhat.com>
225 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
226 of incompatible types.
228 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
230 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
231 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
232 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
233 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
235 2017-09-19 Richard Biener <rguenther@suse.de>
237 PR tree-optimization/82244
238 * tree-vrp.c (remove_range_assertions): Do not propagate
239 a constant to abnormals but replace the assert with a copy.
241 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
243 PR rtl-optimization/57878
244 PR rtl-optimization/68988
245 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
246 avoidance test involving non_reload_pseudos. Move frequency test
247 below the general fragmentation avoidance test.
249 2017-09-19 Richard Biener <rguenther@suse.de>
251 PR tree-optimization/69728
252 * graphite-sese-to-poly.c (schedule_error): New global.
253 (add_loop_schedule): Handle empty domain by failing the
255 (build_original_schedule): Handle schedule_error.
257 2017-09-19 Richard Biener <rguenther@suse.de>
259 * graphite-scop-detection.c (scop_detection::can_represent_loop):
260 Do not iterate to sibling loops but only to siblings of inner
263 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
266 * config/m68k/m68k.md (moveq feeding equality comparison): Check
267 that the registers are different.
269 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
271 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
272 to processor_model and "amdfam17h" to arch_names_table.
273 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
275 2017-09-18 Jakub Jelinek <jakub@redhat.com>
278 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
280 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
281 Alan Hayward <alan.hayward@arm.com>
282 David Sherwood <david.sherwood@arm.com>
284 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
286 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
288 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
289 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
290 vect_schedule_slp_instance.
291 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
292 Update call to vect_slp_analyze_node_operations. Simplify return
294 (vect_slp_analyze_bb_1): Update call accordingly.
295 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
296 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
297 (vect_schedule_slp): Update call accordingly.
299 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
300 Alan Hayward <alan.hayward@arm.com>
301 David Sherwood <david.sherwood@arm.com>
303 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
304 with types that aren't in fact scalar.
306 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
308 * tree-vect-slp.c (vect_record_max_nunits): New function,
310 (vect_build_slp_tree_1): ...here.
311 (vect_build_slp_tree_2): Call it for phis too.
313 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
315 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
316 to vect_get_vec_def_for_operand when getting the mask operand.
318 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
319 Alan Hayward <alan.hayward@arm.com>
320 David Sherwood <david.sherwood@arm.com>
322 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
325 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
326 Alan Hayward <alan.hayward@arm.com>
327 David Sherwood <david.sherwood@arm.com>
329 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
330 calculation for vector booleans.
332 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
333 Alan Hayward <alan.hayward@arm.com>
334 David Sherwood <david.sherwood@arm.com>
336 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
338 (vect_transform_stmt): ...here.
339 (vect_analyze_stmt): Use it instead of calling
340 vectorizable_live_operation directly.
342 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
344 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
345 non-SIMT targets in acc vector loops.
347 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
349 * configure.ac: Add arc and check if assembler supports gdwarf2.
350 * configure: Regenerate.
352 2017-09-18 Richard Biener <rguenther@suse.de>
354 PR tree-optimization/82220
355 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
356 epilogue niters from the min_profitable_iters compute.
358 2017-09-18 Jakub Jelinek <jakub@redhat.com>
361 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
362 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
363 (ix86_init_pic_reg): Revert 2017-09-01 changes.
365 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
368 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
369 switching to a new text section.
371 2017-09-18 Richard Biener <rguenther@suse.de>
373 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
375 (build_alias_set): Reject aliases with no access function.
377 2017-09-18 Richard Biener <rguenther@suse.de>
379 PR tree-optimization/79622
380 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
382 (build_cross_bb_scalars_use): Likewise.
384 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
386 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
388 2017-09-18 Alan Modra <amodra@gmail.com>
391 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
392 stack_pointer_rtx for count 0. Update comments. Break up
393 large rtl expression.
395 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
397 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
398 Increase to 20 bytes.
399 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
400 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
401 or avx version of the stub.
403 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
406 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
407 compute the minimum stack alignment. Also update preferred stack
408 boundary for leaf functions.
410 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
412 PR tree-optimization/82228
413 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
416 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
418 * common/config/nds32/nds32-common.c
419 (nds32_option_optimization_table): Refine formatting.
420 (nds32_option_optimization_table): Use -fsched-pressure and
421 -fomit-frame-pointer for specific optimization level.
423 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
425 * config/nds32/nds32.c: Refine formatting and comments.
426 * config/nds32/nds32.h: Likewise.
427 * config/nds32/nds32.md: Likewise.
428 * config/nds32/nds32-cost.c: Likewise.
429 * config/nds32/nds32-isr.c: Likewise.
430 * config/nds32/nds32-md-auxiliary.c: Likewise.
431 * config/nds32/nds32-multiple.md: Likewise.
432 * config/nds32/nds32-predicates.c: Likewise.
434 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
435 Jakub Jelinek <jakub@redhat.com>
437 Add support for -std=c++2a.
438 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
440 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
442 2017-09-15 Steve Ellcey <sellcey@cavium.com>
445 * doc/extend.texi (Common Function Attributes): Add
446 references to ARM, AArch64, and S/390 specific attributes.
447 (Function Specific Option Pragmas): Add AArch64 and S/390
448 to list of back ends that support the target pragma.
450 2017-09-15 Nathan Sidwell <nathan@acm.org>
452 * doc/standards.texi: Fix C++17 description. Update URLs for
455 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
457 * common.opt (Wcast-align=strict): New warning option.
458 * doc/invoke.texi: Document -Wcast-align=strict.
460 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
462 * cgraph.h (cgraph_thunk_info): Add comments.
463 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
464 assert for VIRTUAL_* arguments stricter.
466 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
468 PR tree-optimization/71026
469 * match.pd: Move RDIV patterns from fold-const.c
470 * fold-const.c (distribute_real_division): Removed.
471 (fold_binary_loc): Remove calls to distribute_real_divison.
473 2017-09-15 Jakub Jelinek <jakub@redhat.com>
475 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
476 c++1z and gnu++1z as deprecated. Change other references to
477 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
478 Change -Wc++1z-compat to -Wc++17-compat.
479 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
480 * dwarf2out.c (highest_c_language): Handle C++17.
481 (gen_compile_unit_die): Likewise.
483 2017-09-15 Jakub Jelinek <jakub@redhat.com>
485 PR rtl-optimization/82192
486 * combine.c (make_extraction): Don't look through non-paradoxical
487 SUBREGs or TRUNCATE if pos + len is or might be bigger than
490 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
491 Alan Hayward <alan.hayward@arm.com>
492 David Sherwood <david.sherwood@arm.com>
494 * target.def (function_arg_offset): New hook.
495 * targhooks.h (default_function_arg_offset): Declare.
496 * targhooks.c (default_function_arg_offset): New function.
497 * function.c (locate_and_pad_parm): Use
498 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
499 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
500 (TARGET_FUNCTION_ARG_OFFSET): ...this.
501 * doc/tm.texi: Regenerate.
502 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
503 * config/spu/spu.c (spu_function_arg_offset): New function.
504 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
505 * system.h (FUNCTION_ARG_OFFSET): Poison.
507 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
508 Alan Hayard <alan.hayward@arm.com>
509 David Sherwood <david.sherwood@arm.com>
511 * target.def (truly_noop_truncation): New hook.
512 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
513 than TRULY_NOOP_TRUNCATION.
514 * hooks.h (hook_bool_uint_uint_true): Declare.
515 * hooks.c (hook_bool_uint_uint_true): New function.
516 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
517 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
518 * doc/tm.texi: Regenerate.
519 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
520 rather than TRULY_NOOP_TRUNCATION in comments.
521 (simplify_comparison): Likewise.
522 (record_truncated_value): Likewise.
523 * expmed.c (extract_bit_field_1): Likewise.
524 (extract_split_bit_field): Likewise.
525 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
526 instead of TRULY_NOOP_TRUNCATION.
527 * function.c (assign_parm_setup_block): Likewise.
528 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
529 * rtlhooks.c: Include target.h.
530 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
531 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
532 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
533 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
534 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
535 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
536 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
537 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
538 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
539 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
540 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
541 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
542 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
543 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
544 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
545 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
546 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
547 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
548 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
549 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
550 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
551 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
552 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
553 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
554 * config/mips/mips.c (mips_truly_noop_truncation): New function.
555 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
556 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
557 rather than TRULY_NOOP_TRUNCATION in comments.
558 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
559 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
560 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
561 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
562 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
563 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
564 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
565 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
566 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
567 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
568 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
569 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
570 rather than TRULY_NOOP_TRUNCATION in comments.
571 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
572 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
573 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
574 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
575 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
576 TRULY_NOOP_TRUNCATION condition.
577 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
578 (TRULY_NOOP_TRUNCATION): Delete.
579 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
580 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
581 * config/spu/spu.c (spu_truly_noop_truncation): New function.
582 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
583 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
584 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
585 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
586 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
587 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
588 rather than TRULY_NOOP_TRUNCATION in comments.
589 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
590 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
591 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
592 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
593 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
594 * system.h (TRULY_NOOP_TRUNCATION): Poison.
596 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
599 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
600 (*cmp_ior): Likewise.
601 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
602 (*ior_scc_scc_cmp): Likewise.
603 (*and_scc_scc): Likewise.
604 (*and_scc_scc_cmp): Likewise.
606 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
607 Alan Hayard <alan.hayward@arm.com>
608 David Sherwood <david.sherwood@arm.com>
610 * target.def (can_change_mode_class): New hook.
611 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
612 (hard_regno_nregs): Likewise.
613 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
614 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
615 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
616 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
617 (LOAD_EXTEND_OP): Update accordingly.
618 * doc/tm.texi: Regenerate.
619 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
620 CANNOT_CHANGE_MODE_CLASS.
621 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
622 (REG_CAN_CHANGE_MODE_P): ...this new macro.
623 * combine.c (simplify_set): Update accordingly.
624 * emit-rtl.c (validate_subreg): Likewise.
625 * recog.c (general_operand): Likewise.
626 * regcprop.c (mode_change_ok): Likewise.
627 * reload1.c (choose_reload_regs): Likewise.
628 (inherit_piecemeal_p): Likewise.
629 * rtlanal.c (simplify_subreg_regno): Likewise.
630 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
631 instead of CANNOT_CHANGE_MODE_CLASS.
632 (reload_cse_simplify_operands): Likewise.
633 * reload.c (push_reload): Use targetm.can_change_mode_class
634 instead of CANNOT_CHANGE_MODE_CLASS.
635 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
636 REG_CANNOT_CHANGE_MODE_P.
637 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
638 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
639 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
640 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
641 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
642 (arm_can_change_mode_class): New function.
643 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
644 than CANNOT_CHANGE_MODE_CLASS in comments.
645 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
646 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
647 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
648 (ix86_can_change_mode_class): ...this new function, inverting the
649 sense of the return value.
650 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
651 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
652 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
653 (ia64_can_change_mode_class): New function.
654 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
655 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
656 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
657 (m32c_can_change_mode_class): ...this new function, inverting the
658 sense of the return value.
659 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
660 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
661 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
662 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
663 (mips_can_change_mode_class): ...this new function, inverting the
664 sense of the return value.
665 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
666 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
667 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
668 (msp430_can_change_mode_class): New function.
669 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
670 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
671 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
672 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
673 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
674 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
675 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
676 (pa_cannot_change_mode_class): Replace with...
677 (pa_can_change_mode_class): ...this new function, inverting the
678 sense of the return value.
679 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
680 than CANNOT_CHANGE_MODE_CLASS in comments.
681 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
682 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
683 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
684 (pdp11_cannot_change_mode_class): Replace with...
685 (pdp11_can_change_mode_class): ...this new function, inverting the
686 sense of the return value.
687 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
688 * config/powerpcspe/powerpcspe-protos.h
689 (rs6000_cannot_change_mode_class_ptr): Delete.
690 * config/powerpcspe/powerpcspe.c
691 (rs6000_cannot_change_mode_class_ptr): Delete.
692 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
693 (rs6000_option_override_internal): Assign to
694 targetm.can_change_mode_class instead of
695 rs6000_cannot_change_mode_class_ptr.
696 (rs6000_cannot_change_mode_class): Replace with...
697 (rs6000_can_change_mode_class): ...this new function, inverting the
698 sense of the return value.
699 (rs6000_debug_cannot_change_mode_class): Replace with...
700 (rs6000_debug_can_change_mode_class): ...this new function.
701 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
702 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
703 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
704 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
705 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
707 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
708 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
709 (rs6000_option_override_internal): Assign to
710 targetm.can_change_mode_class instead of
711 rs6000_cannot_change_mode_class_ptr.
712 (rs6000_cannot_change_mode_class): Replace with...
713 (rs6000_can_change_mode_class): ...this new function, inverting the
714 sense of the return value.
715 (rs6000_debug_cannot_change_mode_class): Replace with...
716 (rs6000_debug_can_change_mode_class): ...this new function.
717 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
718 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
719 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
720 (s390_can_change_mode_class): ...this new function, inverting the
721 sense of the return value.
722 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
723 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
724 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
725 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
726 (sh_cannot_change_mode_class): Replace with...
727 (sh_can_change_mode_class): ...this new function, inverting the
728 sense of the return value.
729 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
730 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
731 (sparc_can_change_mode_class): New function.
732 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
733 * config/spu/spu.c (spu_can_change_mode_class): New function.
734 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
735 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
736 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
737 (visium_can_change_mode_class): New function.
738 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
740 2017-09-15 Richard Biener <rguenther@suse.de>
742 PR tree-optimization/82217
743 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
744 but not undefined case.
746 2017-09-15 Jakub Jelinek <jakub@redhat.com>
749 * postreload.c (reload_cse_simplify_operands): Skip
750 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
752 2017-09-15 Richard Biener <rguenther@suse.de>
754 PR tree-optimization/68823
755 * graphite-scop-detection.c (build_alias_set): If we have a
756 possible dependence check whether we can handle them by just
757 looking at the DRs DR_ACCESS_FNs.
758 (build_scops): If build_alias_set fails, fail the SCOP.
760 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
762 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
763 to support float128 built-in functions that require the ISA 3.0
765 (BU_FLOAT128_3_HW): Likewise.
766 (SQRTF128): Add support for the IEEE 128-bit square root and fma
770 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
771 support for built-in functions that need the ISA 3.0 IEEE 128-bit
772 floating point instructions.
773 (rs6000_invalid_builtin): Likewise.
774 (rs6000_builtin_mask_names): Likewise.
775 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
776 (RS6000_BTM_FLOAT128_HW): Likewise.
777 (RS6000_BTM_COMMON): Likewise.
778 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
780 * doc/extend.texi (RS/6000 built-in functions): Document the
781 IEEE 128-bit floating point square root and fused multiply-add
784 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
786 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
787 reg (r2) isn't in the set of registers defined in the prologue.
789 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
790 Alan Hayward <alan.hayward@arm.com>
791 David Sherwood <david.sherwood@arm.com>
793 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
794 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
795 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
796 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
797 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
799 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
800 max_vectorization_factor.
801 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
803 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
804 Alan Hayward <alan.hayward@arm.com>
805 David Sherwood <david.sherwood@arm.com>
807 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
808 (vect_worthwhile_without_simd_p): Declare.
809 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
810 (vectorizable_reduction): Use it.
811 * tree-vect-stmts.c (vectorizable_shift): Likewise.
812 (vectorizable_operation): Likewise.
814 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
815 Alan Hayward <alan.hayward@arm.com>
816 David Sherwood <david.sherwood@arm.com>
818 * tree-vectorizer.h (vect_get_num_copies): New function.
819 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
820 * tree-vect-loop.c (vectorizable_reduction): Likewise.
821 (vectorizable_induction): Likewise.
822 (vectorizable_live_operation): Likewise.
823 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
824 (vectorizable_bswap): Likewise.
825 (vectorizable_call): Likewise.
826 (vectorizable_conversion): Likewise.
827 (vectorizable_assignment): Likewise.
828 (vectorizable_shift): Likewise.
829 (vectorizable_operation): Likewise.
830 (vectorizable_store): Likewise.
831 (vectorizable_load): Likewise.
832 (vectorizable_condition): Likewise.
833 (vectorizable_comparison): Likewise.
834 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
836 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
837 Alan Hayward <alan.hayward@arm.com>
838 David Sherwood <david.sherwood@arm.com>
840 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
843 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
844 Alan Hayward <alan.hayward@arm.com>
845 David Sherwood <david.sherwood@arm.com>
847 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
848 an inline wrapper that provides a location.
849 (gimple_build_vector): Likewise.
850 * gimple-fold.c (gimple_build_vector_from_val): New function.
851 (gimple_build_vector): Likewise.
852 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
853 functions to build the initial value. Always return a gimple value.
854 (get_initial_defs_for_reduction): Likewise. Only compute
856 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
857 vect_init_vector on the results from get_initial_def(s)_for_reduction.
858 (vectorizable_induction): Use gimple_build_vector rather than
861 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
862 Alan Hayward <alan.hayward@arm.com>
863 David Sherwood <david.sherwood@arm.com>
865 * target.h (vec_perm_indices): New typedef.
866 (auto_vec_perm_indices): Likewise.
867 * optabs-query.h: Include target.h
868 (can_vec_perm_p): Take a vec_perm_indices *.
869 * optabs-query.c (can_vec_perm_p): Likewise.
870 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
871 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
872 * tree-vect-generic.c (lower_vec_perm): Likewise.
873 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
874 (vect_grouped_load_supported): Likewise.
875 (vect_shift_permute_load_chain): Likewise.
876 (vect_permute_store_chain): Use auto_vec_perm_indices.
877 (vect_permute_load_chain): Likewise.
878 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
879 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
880 Update uses of can_vec_perm_p.
881 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
882 mode with a number of elements. Take a vec_perm_indices *.
883 (vect_create_epilog_for_reduction): Update accordingly.
884 Use auto_vec_perm_indices.
885 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
886 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
887 (vect_transform_slp_perm_load): Likewise.
888 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
889 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
890 (vect_gen_perm_mask_checked): Likewise.
891 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
892 (vect_gen_perm_mask_checked): Likewise.
893 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
894 (vectorizable_store): Likewise.
895 (vectorizable_load): Likewise.
896 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
897 (vectorizable_bswap): Likewise.
899 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
900 Alan Hayward <alan.hayward@arm.com>
901 David Sherwood <david.sherwood@arm.com>
903 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
904 * tree.c (build_vector): Likewise.
905 (build_vector_from_ctor): Update accordingly.
906 (build_vector_from_val): Likewise.
907 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
908 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
909 * tree-vect-generic.c (add_rshift): Likewise.
910 (expand_vector_divmod): Likewise.
911 (optimize_vector_constructor): Likewise.
912 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
913 (vect_transform_slp_perm_load): Likewise.
914 (vect_schedule_slp_instance): Likewise.
915 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
916 (vectorizable_call): Likewise.
917 (vect_gen_perm_mask_any): Likewise. Add elements in order.
918 * expmed.c (make_tree): Likewise.
919 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
920 a vector passed to build_vector.
921 (fold_convert_const): Likewise.
922 (exact_inverse): Likewise.
923 (fold_ternary_loc): Likewise.
924 (fold_relational_const): Likewise.
925 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
926 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
927 (const_unop): Likewise. Store the reduction accumulator in a
928 variable rather than an array.
929 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
930 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
931 the new vector, rather than constructing it after the input arrays.
932 (native_interpret_vector): Use auto_vec<tree> when building
933 a vector passed to build_vector. Add elements in order.
934 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
935 auto_vec<tree> when building a vector passed to build_vector.
936 (vect_create_epilog_for_reduction): Likewise.
937 (vectorizable_induction): Likewise.
938 (get_initial_def_for_reduction): Likewise. Fix indentation of
940 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
942 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
943 passed to build_vector.
945 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
946 Alan Hayward <alan.hayward@arm.com>
947 David Sherwood <david.sherwood@arm.com>
949 * tree-core.h (tree_base::u): Add an "nelts" field.
950 (tree_vector): Use VECTOR_CST_NELTS as the length.
951 * tree.c (tree_size): Likewise.
952 (make_vector): Initialize VECTOR_CST_NELTS.
953 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
954 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
955 TYPE_VECTOR_SUBPARTS.
956 * expr.c (const_vector_mask_from_tree): Consistently use "units"
957 as the number of units, setting it from VECTOR_CST_NELTS.
958 (const_vector_from_tree): Likewise.
959 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
960 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
961 (fold_negate_expr_1): Likewise.
962 (fold_convert_const): Likewise.
963 (const_binop): Likewise. Differentiate the number of output and
965 (const_unop): Likewise.
966 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
967 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
968 in cases that did the opposite.
970 2017-09-14 Richard Biener <rguenther@suse.de>
972 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
975 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
977 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
979 2017-09-14 Jakub Jelinek <jakub@redhat.com>
982 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
983 if and where to split a bb, except for splitting before debug insn
984 sequences followed by non-label real insn. Delete debug insns
985 in between basic blocks.
987 * combine.c (make_compound_operation_int): Formatting fixes.
989 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
990 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
991 * config/netbsd.h (LINK_EH_SPEC): Likewise.
992 * config/sol2.h (LINK_EH_SPEC): Likewise.
993 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
994 * config/s390/linux.h (LINK_SPEC): Likewise.
995 * config/freebsd.h (LINK_EH_SPEC): Likewise.
996 * config/openbsd.h (LINK_EH_SPEC): Likewise.
997 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
998 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
999 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1000 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1001 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1002 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1004 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1006 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1008 (ENDFILE_LINUX_SPEC): Likewise.
1009 (LINK_EH_SPEC): Likewise.
1010 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1011 (LINK_OS_LINUX_SPEC32): Likewise.
1012 (LINK_OS_LINUX_SPEC64): Likewise.
1013 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1014 (LINK_OS_LINUX_SPEC): Likewise.
1016 2017-09-13 Martin Liska <mliska@suse.cz>
1019 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1020 CASE_HIGH is NULL_TREE.
1022 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1023 Alan Hayward <alan.hayward@arm.com>
1024 David Sherwood <david.sherwood@arm.com>
1026 * target.def (secondary_memory_needed): New hook.
1027 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1028 instead of SECONDARY_MEMORY_NEEDED.
1029 (secondary_memory_needed_mode): Likewise.
1030 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1031 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1032 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1033 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1034 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1035 * doc/tm.texi: Regenerate.
1036 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1037 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1038 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1039 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1040 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1041 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1042 mode argument first and change the reg_class arguments to reg_class_t.
1043 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1044 Make static. Update the call to inline_secondary_memory_needed.
1045 (ix86_register_move_cost): Update the call to
1046 inline_secondary_memory_needed.
1047 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1048 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1050 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1051 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1053 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1054 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1055 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1056 and match hook interface. Add comment from mips.h.
1057 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1058 * config/mmix/mmix.md (truncdfsf2): Refer to
1059 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1061 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1062 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1063 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1064 (pa_secondary_memory_needed): New function.
1065 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1066 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1067 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1068 (pdp11_secondary_memory_needed): Make static and match hook interface.
1069 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1070 * config/powerpcspe/powerpcspe-protos.h
1071 (rs6000_secondary_memory_needed_ptr): Delete.
1072 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1074 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1075 (rs6000_option_override_internal): Assign to
1076 targetm.secondary_memory_needed rather than
1077 rs6000_secondary_memory_needed_ptr.
1078 (rs6000_secondary_memory_needed): Match hook interface.
1079 (rs6000_debug_secondary_memory_needed): Likewise.
1080 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1081 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
1082 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
1083 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1084 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
1085 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
1087 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
1088 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1089 (rs6000_option_override_internal): Assign to
1090 targetm.secondary_memory_needed rather than
1091 rs6000_secondary_memory_needed_ptr.
1092 (rs6000_secondary_memory_needed): Match hook interface.
1093 (rs6000_debug_secondary_memory_needed): Likewise.
1094 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
1095 * config/s390/s390.c (s390_secondary_memory_needed): New function.
1096 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1097 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
1098 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1099 (sparc_secondary_memory_needed): New function.
1100 * lra-constraints.c (check_and_process_move): Refer to
1101 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1103 (curr_insn_transform): Likewise.
1104 (process_alt_operands): Use targetm.secondary_memory_needed
1105 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1106 (check_secondary_memory_needed_p): Likewise.
1107 (choose_split_class): Likewise.
1108 * reload.c: Unconditionally include code that was previously
1109 conditional on SECONDARY_MEMORY_NEEDED.
1110 (push_secondary_reload): Use targetm.secondary_memory_needed
1111 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1112 (push_reload): Likewise.
1113 * reload1.c: Unconditionally include code that was previously
1114 conditional on SECONDARY_MEMORY_NEEDED.
1115 (choose_reload_regs): Use targetm.secondary_memory_needed
1116 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1117 (gen_reload): Likewise.
1118 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
1120 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1121 Alan Hayward <alan.hayward@arm.com>
1122 David Sherwood <david.sherwood@arm.com>
1124 * target.def (secondary_memory_needed_mode): New hook:
1125 * targhooks.c (default_secondary_memory_needed_mode): Declare.
1126 * targhooks.h (default_secondary_memory_needed_mode): New function.
1127 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
1128 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
1129 * doc/tm.texi: Regenerate.
1130 * lra-constraints.c (check_and_process_move): Use
1131 targetm.secondary_memory_needed_mode instead of
1132 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1133 (curr_insn_transform): Likewise.
1134 * reload.c (get_secondary_mem): Likewise.
1135 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1136 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1138 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1139 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1140 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1141 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1142 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1144 * config/powerpcspe/powerpcspe-protos.h
1145 (rs6000_secondary_memory_needed_mode): Delete.
1146 * config/powerpcspe/powerpcspe.c
1147 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1148 (rs6000_secondary_memory_needed_mode): Make static.
1149 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1150 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1152 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1154 (rs6000_secondary_memory_needed_mode): Make static.
1155 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1156 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1157 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1158 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1159 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1161 (sparc_secondary_memory_needed_mode): New function.
1162 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1164 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1166 * config/aarch64/constraints.md (Umq): New constraint.
1167 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1169 (mov<mode>): Update condition.
1171 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1173 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1174 when bitposition is the same.
1176 2017-09-13 Richard Biener <rguenther@suse.de>
1178 * dwarf2out.c (output_die_symbol): Remove.
1179 (output_die): Do not output a DIEs symbol.
1181 2017-09-13 Richard Biener <rguenther@suse.de>
1184 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1185 default-def to avoid breaking iterator update with the weird
1186 interaction with cgraph_update_edges_for_call_stmt_node.
1188 2017-09-13 Richard Biener <rguenther@suse.de>
1190 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1191 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1192 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1193 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1195 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1197 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1198 Disable pc relative literal load irrespective of
1199 TARGET_FIX_ERR_A53_84341 for default.
1201 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1203 * config/sparc/sparc.c (output_return): Output the source location of
1204 the insn in the delay slot, if any.
1205 (output_sibcall): Likewise.
1207 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1210 * common.opt (-static-pie): New alias.
1211 (shared): Negate static-pie.
1212 (-no-pie): Update help text.
1214 (static-pie): New option.
1215 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1216 -static-pie support.
1217 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1218 (LINK_EH_SPEC): Likewise.
1219 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1220 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1221 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1222 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1223 (init_gcc_specs): Likewise.
1224 (init_spec): Likewise.
1225 (display_help): Update help message for -pie.
1226 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1229 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1231 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1232 (movdi_aarch64): Likewise.
1233 (movti_aarch64): Likewise.
1235 2017-09-12 Simon Wright <simon@pushface.org>
1238 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1239 calculation of the minor version, always output as 0.
1241 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1244 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1245 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1246 on it early, rather than manual conversion late. For
1247 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1248 instead of performing manual conversion.
1250 2017-09-12 Carl Love <cel@us.ibm.com>
1252 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1253 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1255 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1256 VMULOSW): Add definitions.
1257 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1258 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1259 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1260 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1261 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1263 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1265 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1267 (movti_aarch64): Likewise.
1268 (movdf_aarch64): Likewise.
1269 (movtf_aarch64): Likewise.
1270 (load_pairdi): Likewise.
1271 (store_pairdi): Likewise.
1272 (load_pairdf): Likewise.
1273 (store_pairdf): Likewise.
1274 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1275 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1276 (ldr_got_small_<mode>): Likewise.
1277 (ldr_got_small_28k_<mode>): Likewise.
1278 (ldr_got_tiny): Likewise.
1279 * config/aarch64/iterators.md (ldst_sz): New.
1280 (ldpstp_sz): Likewise.
1281 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1283 (thunderx_load): Split load_8 to load_16.
1284 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1286 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1287 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1288 (xgene1_store_pair): Split store_8 to store_16.
1289 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1290 (falkor_st_0_st_sd): Split store_8 to store_16.
1292 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1294 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1295 and store1/2/3/4 to store_4/8/12/16.
1296 * config/aarch64/aarch64.md: Update for rename.
1297 * config/arm/arm.md: Likewise.: Likewise.
1298 * config/arm/arm.c: Likewise.
1299 * config/arm/thumb1.md: Likewise.
1300 * config/arm/thumb2.md: Likewise.
1301 * config/arm/vfp.md: Likewise.
1302 * config/arm/arm-generic.md: Likewise.
1303 * config/arm/arm1020e.md: Likewise.
1304 * config/arm/arm1026ejs.md: Likewise.
1305 * config/arm/arm1136jfs.md: Likewise.
1306 * config/arm/arm926ejs.md: Likewise.
1307 * config/arm/cortex-a15.md: Likewise.
1308 * config/arm/cortex-a17.md: Likewise.
1309 * config/arm/cortex-a5.md: Likewise.
1310 * config/arm/cortex-a53.md: Likewise.
1311 * config/arm/cortex-a57.md: Likewise.
1312 * config/arm/cortex-a7.md: Likewise.
1313 * config/arm/cortex-a8.md: Likewise.
1314 * config/arm/cortex-a9.md: Likewise.
1315 * config/arm/cortex-m4.md: Likewise.
1316 * config/arm/cortex-m7.md: Likewise.
1317 * config/arm/cortex-r4.md: Likewise.
1318 * config/arm/exynos-m1.md: Likewise.
1319 * config/arm/fa526.md: Likewise.
1320 * config/arm/fa606te.md: Likewise.
1321 * config/arm/fa626te.md: Likewise.
1322 * config/arm/fa726te.md: Likewise.
1323 * config/arm/fmp626.md: Likewise.
1324 * config/arm/iwmmxt.md: Likewise.
1325 * config/arm/ldmstm.md: Likewise.
1326 * config/arm/marvell-pj4.md: Likewise.
1327 * config/arm/xgene1.md: Likewise.
1328 * config/aarch64/thunderx.md: Likewise.
1329 * config/aarch64/thunderx2t99.md: Likewise.
1330 * config/aarch64/falkor.md: Likewise.
1332 2017-09-12 Martin Liska <mliska@suse.cz>
1334 * attribs.c (private_lookup_attribute): New function.
1335 * attribs.h (private_lookup_attribute): Declared here.
1336 (lookup_attribute): Called from this place.
1338 2017-09-12 Richard Biener <rguenther@suse.de>
1340 PR tree-optimization/82157
1341 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1342 stmts with side-effects.
1344 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1345 Alan Hayward <alan.hayward@arm.com>
1346 David Sherwood <david.sherwood@arm.com>
1348 * target.def (hard_regno_nregs): New hook.
1349 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1350 * targhooks.h (default_hard_regno_nregs): Declare.
1351 * targhooks.c (default_hard_regno_nregs): New function.
1352 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1353 (TARGET_HARD_REGNO_NREGS): ...this hook.
1354 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1355 (CLASS_MAX_NREGS): Likewise.
1356 * doc/tm.texi: Regenerate.
1357 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1358 instead of HARD_REGNO_NREGS.
1359 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1360 HARD_REGNO_NREGS in the comment.
1361 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1362 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1363 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1364 Return an unsigned int.
1365 (TARGET_HARD_REGNO_NREGS): Redefine.
1366 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1367 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1368 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1369 (arc_hard_regno_nregs): New function.
1370 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1371 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1372 (arm_hard_regno_nregs): New function.
1373 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1374 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1375 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1376 (TARGET_HARD_REGNO_NREGS): Redefine.
1377 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1378 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1379 (HARD_REGNO_NREGS): Delete.
1380 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1381 (cr16_hard_regno_nregs): New function.
1382 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1383 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1384 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1385 (cris_hard_regno_nregs): New function.
1386 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1387 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1388 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1389 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1390 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1391 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1392 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1393 (frv_hard_regno_nregs): Make static. Take and return an
1395 (frv_class_max_nregs): Remove outdated copy of documentation.
1396 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1397 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1398 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1399 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1400 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1401 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1402 (TARGET_HARD_REGNO_NREGS): Redefine.
1403 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1404 (CLASS_MAX_NREGS): Update comment.
1405 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1406 (ia64_hard_regno_nregs): New function.
1407 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1408 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1409 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1410 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1411 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1413 (m32c_hard_regno_nregs): Likewise. Make static.
1414 (TARGET_HARD_REGNO_NREGS): Redefine.
1415 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1416 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1417 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1418 (m68k_hard_regno_nregs): New function.
1419 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1420 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1421 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1422 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1423 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1424 Take and return an unsigned int.
1425 (TARGET_HARD_REGNO_NREGS): Redefine.
1426 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1427 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1428 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1429 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1430 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1431 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1432 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1433 (msp430_hard_regno_nregs): Make static. Take and return an
1435 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1436 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1437 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1438 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1439 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1440 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1441 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1442 (TARGET_HARD_REGNO_NREGS): Redefine.
1443 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1444 (PA_HARD_REGNO_NREGS): ...this.
1445 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1446 (PA_HARD_REGNO_NREGS): ...this.
1447 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1448 (pa_hard_regno_nregs): New function.
1449 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1450 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1451 (pdp11_hard_regno_nregs): New function.
1452 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1453 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1454 (rs6000_hard_regno_nregs_hook): New function.
1455 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1456 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1457 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1458 Take and return an unsigned int. Move earlier in file.
1459 (TARGET_HARD_REGNO_NREGS): Redefine.
1460 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1461 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1462 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1463 (rl78_hard_regno_nregs): Make static. Take and return an
1465 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1466 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1467 (rs6000_hard_regno_nregs_hook): New function.
1468 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1469 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1470 (TARGET_HARD_REGNO_NREGS): Redefine.
1471 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1472 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1473 instead of HARD_REGNO_NREGS.
1474 (s390_hard_regno_nregs): New function.
1475 (s390_hard_regno_mode_ok): Add comment from s390.h.
1476 (TARGET_HARD_REGNO_NREGS): Redefine.
1477 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1478 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1479 (sh_hard_regno_nregs): New function.
1480 (sh_pass_in_reg_p): Use it.
1481 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1482 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1483 (sparc_hard_regno_nregs): New function.
1484 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1485 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1486 (spu_function_arg_advance): Use it, supplying a valid register number.
1487 (TARGET_HARD_REGNO_NREGS): Redefine.
1488 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1489 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1490 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1491 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1492 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1493 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1494 (CLASS_MAX_NREGS): Remove copy of old documentation.
1495 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1496 (visium_hard_regno_nregs): New function.
1497 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1498 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1499 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1500 xtensa_hard_regno_nregs): New function.
1501 * system.h (HARD_REGNO_NREGS): Poison.
1503 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1505 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1506 hard_regno_nregs instead of HARD_REGNO_NREGS.
1507 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1508 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1509 (c6x_expand_epilogue): Likewise.
1510 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1511 (frv_read_iacc_argument): Likewise.
1512 * config/sh/sh.c: Include regs.h.
1513 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1514 (regs_used): Likewise.
1515 (output_stack_adjust): Likewise.
1516 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1517 * expmed.c: Include regs.h.
1518 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1519 * ree.c: Include regs.h.
1520 (combine_reaching_defs): Use hard_regno_nregs instead of
1522 (add_removable_extension): Likewise.
1524 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1526 * regs.h (hard_regno_nregs): Turn into a function.
1527 (end_hard_regno): Update accordingly.
1528 * caller-save.c (setup_save_areas): Likewise.
1529 (save_call_clobbered_regs): Likewise.
1530 (replace_reg_with_saved_mem): Likewise.
1531 (insert_restore): Likewise.
1532 (insert_save): Likewise.
1533 * combine.c (can_change_dest_mode): Likewise.
1534 (move_deaths): Likewise.
1535 (distribute_notes): Likewise.
1536 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1537 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1538 (rs6000_split_multireg_move): Likewise.
1539 (rs6000_register_move_cost): Likewise.
1540 (rs6000_memory_move_cost): Likewise.
1541 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1542 (rs6000_split_multireg_move): Likewise.
1543 (rs6000_register_move_cost): Likewise.
1544 (rs6000_memory_move_cost): Likewise.
1545 * cselib.c (cselib_reset_table): Likewise.
1546 (cselib_lookup_1): Likewise.
1547 * emit-rtl.c (set_mode_and_regno): Likewise.
1548 * function.c (aggregate_value_p): Likewise.
1549 * ira-color.c (setup_profitable_hard_regs): Likewise.
1550 (check_hard_reg_p): Likewise.
1551 (calculate_saved_nregs): Likewise.
1552 (assign_hard_reg): Likewise.
1553 (improve_allocation): Likewise.
1554 (calculate_spill_cost): Likewise.
1555 * ira-emit.c (modify_move_list): Likewise.
1556 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1557 (ira_hard_reg_in_set_p): Likewise.
1558 * ira.c (setup_reg_mode_hard_regset): Likewise.
1559 (clarify_prohibited_class_mode_regs): Likewise.
1560 (check_allocation): Likewise.
1561 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1562 (lra_setup_reg_renumber): Likewise.
1563 (setup_try_hard_regno_pseudos): Likewise.
1564 (spill_for): Likewise.
1565 (assign_hard_regno): Likewise.
1566 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1567 * lra-constraints.c (in_class_p): Likewise.
1568 (lra_constraint_offset): Likewise.
1569 (simplify_operand_subreg): Likewise.
1570 (lra_constraints): Likewise.
1571 (split_reg): Likewise.
1572 (split_if_necessary): Likewise.
1573 (invariant_p): Likewise.
1574 (inherit_in_ebb): Likewise.
1575 * lra-lives.c (process_bb_lives): Likewise.
1576 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1577 (get_hard_regs): Likewise.
1578 (do_remat): Likewise.
1579 * lra-spills.c (assign_spill_hard_regs): Likewise.
1580 * mode-switching.c (create_pre_exit): Likewise.
1581 * postreload.c (reload_combine_recognize_pattern): Likewise.
1582 * recog.c (peep2_find_free_register): Likewise.
1583 * regcprop.c (kill_value_regno): Likewise.
1584 (set_value_regno): Likewise.
1585 (copy_value): Likewise.
1586 (maybe_mode_change): Likewise.
1587 (find_oldest_value_reg): Likewise.
1588 (copyprop_hardreg_forward_1): Likewise.
1589 * regrename.c (check_new_reg_p): Likewise.
1590 (regrename_do_replace): Likewise.
1591 * reload.c (push_reload): Likewise.
1592 (combine_reloads): Likewise.
1593 (find_dummy_reload): Likewise.
1594 (operands_match_p): Likewise.
1595 (find_reloads): Likewise.
1596 (find_equiv_reg): Likewise.
1597 (reload_adjust_reg_for_mode): Likewise.
1598 * reload1.c (count_pseudo): Likewise.
1599 (count_spilled_pseudo): Likewise.
1600 (find_reg): Likewise.
1601 (clear_reload_reg_in_use): Likewise.
1602 (free_for_value_p): Likewise.
1603 (allocate_reload_reg): Likewise.
1604 (choose_reload_regs): Likewise.
1605 (reload_adjust_reg_for_temp): Likewise.
1606 (emit_reload_insns): Likewise.
1607 (delete_output_reload): Likewise.
1608 * rtlanal.c (subreg_get_info): Likewise.
1609 * sched-deps.c (sched_analyze_reg): Likewise.
1610 * sel-sched.c (init_regs_for_mode): Likewise.
1611 (mark_unavailable_hard_regs): Likewise.
1612 (choose_best_reg_1): Likewise.
1613 (verify_target_availability): Likewise.
1614 * valtrack.c (dead_debug_insert_temp): Likewise.
1615 * var-tracking.c (track_loc_p): Likewise.
1616 (emit_note_insn_var_location): Likewise.
1617 * varasm.c (make_decl_rtl): Likewise.
1618 * reginfo.c (choose_hard_reg_mode): Likewise.
1619 (init_reg_modes_target): Refer directly to
1620 this_target_regs->x_hard_regno_nregs.
1622 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1624 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1625 instead of hard_regno_nregs.
1627 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1629 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1630 end_hard_regno instead of hard_regno_nregs.
1631 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1632 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1633 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1634 * ira-color.c (improve_allocation): Likewise.
1635 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1636 * lra-lives.c (mark_regno_live): Likewise.
1637 (mark_regno_dead): Likewise.
1638 * lra-remat.c (operand_to_remat): Likewise.
1639 * lra.c (collect_non_operand_hard_regs): Likewise.
1640 * postreload.c (reload_combine_note_store): Likewise.
1641 (move2add_valid_value_p): Likewise.
1642 * reload.c (regno_clobbered_p): Likewise.
1644 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1646 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1648 * config/v850/v850.c (v850_reorg): Likewise.
1649 * reload.c (refers_to_regno_for_reload_p): Likewise.
1650 (find_equiv_reg): Likewise.
1651 * reload1.c (reload_reg_reaches_end_p): Likewise.
1653 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1655 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1657 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1658 * config/arm/arm.c (output_move_neon): Likewise.
1659 (arm_attr_length_move_neon): Likewise.
1660 (neon_split_vcombine): Likewise.
1661 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1662 (c6x_mark_reg_written): Likewise.
1663 (c6x_dwarf_register_span): Likewise.
1664 * config/i386/i386.c (ix86_save_reg): Likewise.
1665 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1666 (rws_access_reg): Likewise.
1667 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1668 * mode-switching.c (create_pre_exit): Likewise.
1669 * ree.c (combine_reaching_defs): Likewise.
1670 (add_removable_extension): Likewise.
1671 * regcprop.c (find_oldest_value_reg): Likewise.
1672 (copyprop_hardreg_forward_1): Likewise.
1673 * reload.c (reload_inner_reg_of_subreg): Likewise.
1674 (push_reload): Likewise.
1675 (combine_reloads): Likewise.
1676 (find_dummy_reload): Likewise.
1677 (reload_adjust_reg_for_mode): Likewise.
1678 * reload1.c (find_reload_regs): Likewise.
1679 (forget_old_reloads_1): Likewise.
1680 (reload_reg_free_for_value_p): Likewise.
1681 (reload_adjust_reg_for_temp): Likewise.
1682 (emit_reload_insns): Likewise.
1683 (delete_output_reload): Likewise.
1684 * sel-sched.c (choose_best_reg_1): Likewise.
1685 (choose_best_pseudo_reg): Likewise.
1687 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1688 Alan Hayward <alan.hayward@arm.com>
1689 David Sherwood <david.sherwood@arm.com>
1691 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1692 * target.def (slow_unaligned_access): New hook.
1693 * targhooks.h (default_slow_unaligned_access): Declare.
1694 * targhooks.c (default_slow_unaligned_access): New function.
1695 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1696 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1697 * doc/tm.texi: Regenerate.
1698 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1699 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1700 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1702 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1703 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1705 (rs6000_slow_unaligned_access): New function.
1706 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1707 (expand_block_compare): Likewise.
1708 (expand_strn_compare): Likewise.
1709 (rs6000_rtx_costs): Likewise.
1710 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1711 (riscv_slow_unaligned_access): Likewise.
1712 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1713 (riscv_slow_unaligned_access_p): ...this and make static.
1714 (riscv_option_override): Update accordingly.
1715 (riscv_slow_unaligned_access): New function.
1716 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1717 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1718 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1719 (rs6000_slow_unaligned_access): New function.
1720 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1721 (rs6000_rtx_costs): Likewise.
1722 * config/rs6000/rs6000-string.c (expand_block_compare)
1723 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1724 of SLOW_UNALIGNED_ACCESS.
1725 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1726 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1727 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1728 of SLOW_UNALIGNED_ACCESS.
1729 * expmed.c (simple_mem_bitfield_p): Likewise.
1730 * expr.c (alignment_for_piecewise_move): Likewise.
1731 (emit_group_load_1): Likewise.
1732 (emit_group_store): Likewise.
1733 (copy_blkmode_from_reg): Likewise.
1734 (emit_push_insn): Likewise.
1735 (expand_assignment): Likewise.
1736 (store_field): Likewise.
1737 (expand_expr_real_1): Likewise.
1738 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1739 * lra-constraints.c (simplify_operand_subreg): Likewise.
1740 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1741 * gimple-ssa-store-merging.c: Likewise in block comment at start
1743 * tree-ssa-strlen.c: Include target.h.
1744 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1745 of SLOW_UNALIGNED_ACCESS.
1746 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1748 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1750 PR rtl-optimization/82185
1751 * expmed.c (emit_store_flag_int): Only test tem if it has been
1754 2017-09-12 Richard Biener <rguenther@suse.de>
1757 * match.pd ((FTYPE) N CMP CST): Fix typo.
1759 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1761 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1763 (mips_near_type_p): Add 'short_call' attribute as a synonym
1765 * doc/extend.texi (short_call): Document new function attribute.
1767 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1770 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1771 assertion check that in the condition.
1772 (get_atomic_generic_size): Likewise. Before testing if parameter
1773 has pointer type, if it has array type, call for C++
1774 default_conversion to perform array-to-pointer conversion.
1776 2017-09-12 Richard Biener <rguenther@suse.de>
1778 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1779 for operations we cannot scalarize.
1781 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1783 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1784 vectors heap vectors. Clean up comments.
1785 Make visited_bbs a reference.
1786 (profitable_jump_thread_path): Make GC
1787 vectors heap vectors. Clean up comments.
1789 (convert_and_register_jump_thread_path): Make GC vectors heap
1791 (check_subpath_and_update_thread_path): Same. Clean up comments.
1792 Make visited_bbs a reference.
1793 (handle_phi): Abstract common code to to
1794 register_jump_thread_path_if_profitable.
1795 Rename VAR_BB to DEF_BB.
1797 Make GC vectors heap vectors.
1798 Make visited_bbs a reference.
1799 (handle_assignment): Same.
1800 (register_jump_thread_path_if_profitable): New.
1801 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1803 Make GC vectors heap vectors. Clean up comments.
1804 Make visited_bbs a reference.
1805 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1806 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1809 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1812 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1813 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1815 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1817 Revert r251800 and r251799.
1819 2017-09-11 Martin Jambor <mjambor@suse.cz>
1822 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1823 arguments in advance.
1824 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1825 use it to find predecessor edges.
1826 (naive_outof_ssa): Collect vector of predecessors.
1828 2017-09-08 Jason Merrill <jason@redhat.com>
1830 PR c++/70029 - ICE with ref-qualifier and -flto
1831 * langhooks.h (struct lang_hooks_for_types): Add
1832 copy_lang_qualifiers.
1833 * attribs.c (build_type_attribute_qual_variant): Use it.
1834 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1836 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1837 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1839 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1842 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1843 (*mulsi3_sp64): New instruction.
1844 (mulsi3): New expander.
1846 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1848 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1850 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1852 * sancov.c: Include memmodel.h.
1854 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1857 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1860 2017-09-07 Carl Love <cel@us.ibm.com>
1862 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1863 the sldi instruction.
1865 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1867 * sancov.c: Include tm_p.h.
1869 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1872 * output.h (switch_to_other_text_partition): New declaration.
1873 * varasm.c (switch_to_other_text_partition): New function.
1874 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1875 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1876 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1877 to the other text partition before emitting LCL label and switch back
1878 after emitting the word after it.
1880 2017-09-07 Richard Biener <rguenther@suse.de>
1882 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1883 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1884 critical edges here, after loop init.
1885 (pass_data_pre): Remove PROP_no_crit_edges flags.
1886 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1887 for valueization of call args to avoid leaking VN_TOP.
1888 (visit_use): Assert we do not visit default defs.
1889 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1890 Use error_mark_node to more easily detect leaking VN_TOP.
1891 All default-defs are varying, not VN_TOP. Mark them visited.
1892 (run_scc_vn): Make code match comment.
1894 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1896 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1897 OPTION_MASK_FLOAT128_KEYWORD.
1898 (POWERPC_MASKS): Likewise.
1899 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1900 support for the -mfloat128-type option, and make -mfloat128
1901 default on PowerPC Linux systems. Define or undefine
1902 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1903 Define __float128 to be __ieee128 if IEEE 128-bit support is
1904 enabled, or undefine it.
1905 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1906 Delete defining __FLOAT128_TYPE__.
1907 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1908 -mfloat128-type option and make -mfloat128 default on PowerPC
1910 (TARGET_FLOAT128_TYPE): Likewise.
1911 (-mfloat128-type): Likewise.
1912 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1913 Delete the -mfloat128-type option and make -mfloat128 default on
1914 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1915 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1916 128-bit floating point is enabled. Change tests from using
1917 -mfloat128-type to -mfloat128.
1918 (rs6000_mangle_type): Use the correct mangling for the __float128
1919 type even if normal long double is restricted to 64-bits.
1920 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1922 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1923 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1924 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1926 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1927 documentation for -mfloat128.
1929 2017-09-06 Olivier Hainque <hainque@adacore.com>
1931 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1933 2017-09-06 Wish Wu <wishwu007@gmail.com>
1934 Jakub Jelinek <jakub@redhat.com>
1936 * asan.c (initialize_sanitizer_builtins): Add
1937 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1938 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1939 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1940 BT_FN_VOID_UINT64_PTR variables.
1941 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1942 (BT_FN_VOID_UINT16_UINT16): Likewise.
1943 (BT_FN_VOID_UINT32_UINT32): Likewise.
1944 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1945 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1946 (BT_FN_VOID_UINT64_PTR): Likewise.
1947 * common.opt (flag_sanitize_coverage): New variable.
1948 (fsanitize-coverage=trace-pc): Remove.
1949 (fsanitize-coverage=): Add.
1950 * flag-types.h (enum sanitize_coverage_code): New enum.
1951 * fold-const.c (fold_range_test): Disable non-short-circuit
1952 optimization if flag_sanitize_coverage.
1953 (fold_truth_andor): Likewise.
1954 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1955 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1956 (coverage_sanitizer_opts): New array.
1957 (get_closest_sanitizer_option): Add OPTS argument, handle also
1958 OPT_fsanitize_coverage_.
1959 (parse_sanitizer_options): Adjusted to also handle
1960 OPT_fsanitize_coverage_.
1961 (common_handle_option): Add OPT_fsanitize_coverage_.
1962 * sancov.c (instrument_comparison, instrument_switch): New function.
1963 (sancov_pass): Add trace-cmp support.
1964 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1965 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1966 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1967 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1968 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1969 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1970 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1971 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1972 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1973 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1975 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1977 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1978 error. Only quit immediately if parsing is complete.
1979 (BEGIN): Initialize fatal_err and parse_done.
1980 (begin fpu, end fpu): Check number of arguments.
1981 (begin arch, end arch): Likewise.
1982 (begin cpu, end cpu): Likewise.
1983 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1984 (optalias): Likewise.
1986 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1988 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1989 * config/arm/arm-isa.h: Delete. Move definitions to ...
1990 * arm-cpus.in: ... here. Use new feature and fgroup values.
1991 * config/arm/arm.c (arm_option_override): Use lower case for feature
1993 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1994 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1995 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1997 (print_isa_bits_for): New function.
1998 (gen_isa): New function.
1999 (gen_comm_data): Use print_isa_bits_for.
2000 (define feature): New keyword.
2001 (define fgroup): New keyword.
2002 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2003 (arm-isa.h): Add rule to generate file.
2004 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2005 case for feature bit names.
2007 2017-09-06 Richard Biener <rguenther@suse.de>
2009 * tree-ssa-pre.c (NECESSARY): Remove.
2010 (create_expression_by_pieces): Do not touch pass-local flags.
2011 (insert_into_preds_of_block): Likewise.
2012 (do_pre_regular_insertion): Likewise.
2013 (eliminate_insert): Likewise.
2014 (eliminate_dom_walker::before_dom_children): Likewise.
2015 (fini_eliminate): Do not look at inserted_exprs.
2016 (mark_operand_necessary): Remove.
2017 (remove_dead_inserted_code): Replace with simple work-list
2018 algorithm based on inserted_exprs and SSA uses.
2019 (pass_pre::execute): Re-order fini_eliminate and
2020 remove_dead_inserted_code.
2022 2017-09-06 Olivier Hainque <hainque@adacore.com>
2024 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2025 for VxWorks 7. Adjust surrounding comments.
2027 2017-09-06 Richard Biener <rguenther@suse.de>
2029 * gimple-ssa-strength-reduction.c
2030 (find_candidates_dom_walker::before_dom_children): Also allow
2033 2017-09-06 Richard Biener <rguenther@suse.de>
2035 PR tree-optimization/82108
2036 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2037 for gap in the non-permutation SLP case.
2039 2017-09-06 Martin Jambor <mjambor@suse.cz>
2041 PR tree-optimization/82078
2042 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2043 add_access_to_work_queue...
2044 (build_accesses_from_assign): ...here.
2045 (propagate_all_subaccesses): Make sure racc is the group
2046 representative, if there is one.
2048 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2051 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2054 2017-09-06 Richard Biener <rguenther@suse.de>
2056 * gimple-ssa-strength-reduction.c
2057 (find_candidates_dom_walker::before_doom_children): Use a
2058 type and not a mode check.
2060 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2063 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2065 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2067 2017-09-05 Jeff Law <law@redhat.com>
2069 PR tree-optimization/64910
2070 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2071 cases where we have 3 or more operands.
2073 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2076 * omp-low.c (lower_omp_for): Recompute tree invariant if
2077 gimple_omp_for_initial/final is ADDR_EXPR.
2080 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2081 into gimple val before gimplification fo the COND_EXPR.
2083 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
2085 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
2086 REGION_COPY argument.
2087 (thread_through_all_blocks): Remove unused argument to
2088 duplicate_thread_path.
2090 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2091 Alan Hayward <alan.hayward@arm.com>
2092 David Sherwood <david.sherwood@arm.com>
2094 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
2095 Take a scalar_mode rather than a machine_mode.
2096 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2097 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
2098 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2099 (aarch64_gen_adjusted_ldpstp): Likewise.
2100 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2102 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2103 Alan Hayward <alan.hayward@arm.com>
2104 David Sherwood <david.sherwood@arm.com>
2106 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2107 Take a scalar_int_mode instead of a machine_mode.
2108 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2109 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2110 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2111 (aarch64_simd_attr_length_rglist): Delete.
2112 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
2113 a scalar_int_mode instead of a machine_mode.
2114 (aarch64_add_offset): Likewise.
2115 (aarch64_internal_mov_immediate): Likewise
2116 (aarch64_add_constant_internal): Likewise.
2117 (aarch64_add_constant): Likewise.
2118 (aarch64_movw_imm): Likewise.
2119 (aarch64_rtx_arith_op_extract_p): Likewise.
2120 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2121 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2122 Remove assert that the mode isn't a vector.
2123 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2124 (aarch64_expand_mov_immediate): Update calls after above changes.
2125 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
2126 (aarch64_and_bitmask_imm): Check for scalar integer modes.
2127 (aarch64_move_imm): Likewise.
2128 (aarch64_can_const_movi_rtx_p): Likewise.
2129 (aarch64_strip_extend): Likewise.
2130 (aarch64_extr_rtx_p): Likewise.
2131 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2132 a CONST_INT when the mode parameter is VOIDmode.
2133 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2135 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2137 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2138 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2139 (bitwise_type_for_mode): Update accordingly.
2141 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2143 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2144 * stor-layout.c (mode_for_size_tree): Likewise.
2145 (mode_for_array): Update accordingly.
2146 (layout_decl): Likewise.
2147 (compute_record_mode): Likewise. Only set the mode once.
2149 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2151 * target.def (get_mask_mode): Change return type to opt_mode.
2153 * doc/tm.texi: Regenerate.
2154 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2155 * targhooks.c (default_get_mask_mode): Likewise.
2156 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2157 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2158 targetm.get_mask_mode.
2159 * tree.c (build_truth_vector_type): Likewise.
2161 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2163 * machmode.h (mode_for_vector): Return an opt_mode.
2164 * stor-layout.c (mode_for_vector): Likewise.
2165 (mode_for_int_vector): Update accordingly.
2166 (layout_type): Likewise.
2167 * config/i386/i386.c (emit_memmov): Likewise.
2168 (ix86_expand_set_or_movmem): Likewise.
2169 (ix86_expand_vector_init): Likewise.
2170 (ix86_get_mask_mode): Likewise.
2171 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2173 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2174 * expmed.c (extract_bit_field_1): Likewise.
2175 * expr.c (expand_expr_real_2): Likewise.
2176 * optabs-query.c (can_vec_perm_p): Likewise.
2177 (can_vec_mask_load_store_p): Likewise.
2178 * optabs.c (expand_vec_perm): Likewise.
2179 * targhooks.c (default_get_mask_mode): Likewise.
2180 * tree-vect-stmts.c (vectorizable_store): Likewise.
2181 (vectorizable_load): Likewise.
2182 (get_vectype_for_scalar_type_and_size): Likewise.
2184 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2186 * machmode.h (mode_for_int_vector): New function.
2187 * stor-layout.c (mode_for_int_vector): Likewise.
2188 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2189 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2190 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2191 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2192 (s390_expand_vcond): Likewise.
2194 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2196 * machmode.h (opt_machine_mode): New type.
2197 (opt_mode<T>): Allow construction from anything that can be
2199 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2200 (mode_for_size): Return an opt_machine_mode.
2201 * stor-layout.c (mode_for_size): Likewise.
2202 (mode_for_size_tree): Update call accordingly.
2203 (bitwise_mode_for_mode): Likewise.
2204 (make_fract_type): Likewise.
2205 (make_accum_type): Likewise.
2206 * caller-save.c (replace_reg_with_saved_mem): Update call
2208 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2209 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2210 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2211 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2212 * expmed.c (extract_bit_field_1): Likewise.
2213 * reload.c (get_secondary_mem): Likewise.
2214 * varasm.c (assemble_integer): Likewise.
2215 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2218 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2220 * machmode.h (decimal_float_mode_for_size): New function.
2221 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2222 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2223 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2224 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2225 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2226 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2228 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2230 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2231 (get_builtin_sync_mode): Likewise.
2232 (expand_ifn_atomic_compare_exchange): Likewise.
2233 (expand_builtin_atomic_clear): Likewise.
2234 (expand_builtin_atomic_test_and_set): Likewise.
2235 (fold_builtin_atomic_always_lock_free): Likewise.
2236 * calls.c (compute_argument_addresses): Likewise.
2237 (emit_library_call_value_1): Likewise.
2238 (store_one_arg): Likewise.
2239 * combine.c (combine_instructions): Likewise.
2240 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2241 * config/arm/arm.c (arm_function_value): Likewise.
2242 (aapcs_allocate_return_reg): Likewise.
2243 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2244 * config/i386/i386.c (construct_container): Likewise.
2245 (ix86_gimplify_va_arg): Likewise.
2246 (ix86_expand_sse_cmp): Likewise.
2247 (emit_memmov): Likewise.
2248 (emit_memset): Likewise.
2249 (expand_small_movmem_or_setmem): Likewise.
2250 (ix86_expand_pextr): Likewise.
2251 (ix86_expand_pinsr): Likewise.
2252 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2253 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2255 * config/mips/mips.c (mips_function_value_1) Likewise.
2256 (mips_block_move_straight): Likewise.
2257 (mips_expand_ins_as_unaligned_store): Likewise.
2258 * config/powerpcspe/powerpcspe.c
2259 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2260 (rs6000_darwin64_record_arg_flush): Likewise.
2261 * config/rs6000/rs6000.c
2262 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2263 (rs6000_darwin64_record_arg_flush): Likewise.
2264 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2265 (sparc_function_value_1): Likewise.
2266 * config/spu/spu.c (adjust_operand): Likewise.
2267 (spu_emit_branch_or_set): Likewise.
2268 (arith_immediate_p): Likewise.
2269 * emit-rtl.c (gen_lowpart_common): Likewise.
2270 * expr.c (expand_expr_real_1): Likewise.
2271 * function.c (assign_parm_setup_block): Likewise.
2272 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2273 * reload1.c (alter_reg): Likewise.
2274 * stor-layout.c (mode_for_vector): Likewise.
2275 (layout_type): Likewise.
2277 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2279 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2280 (spu_convert_move): Likewise.
2281 * lower-subreg.c (resolve_simple_move): Likewise.
2283 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2286 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2287 define_insn to a define_expand.
2288 (altivec_vsum2sws_direct): New define_insn.
2289 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2291 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2293 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2296 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2301 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2302 realign stack if stack alignment needed is less than incoming
2305 2017-09-05 Marek Polacek <polacek@redhat.com>
2308 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2311 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2313 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2315 2017-09-05 Richard Biener <rguenther@suse.de>
2317 PR tree-optimization/82084
2318 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2320 2017-09-05 Richard Biener <rguenther@suse.de>
2322 PR tree-optimization/82102
2323 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2325 2017-09-05 Martin Liska <mliska@suse.cz>
2327 PR tree-optimization/82032
2328 * tree-cfg.c (generate_range_test): New function.
2329 * tree-cfg.h (generate_range_test): Declared here.
2330 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2331 (cleanup_control_expr_graph): Use it.
2332 * tree-switch-conversion.c (try_switch_expansion): Remove
2334 (emit_case_nodes): Use generate_range_test.
2336 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2339 * config/i386/i386.md (*<btsc><mode>_mask): Add
2340 TARGET_USE_BT to insn constraint.
2341 (*btr<mode>_mask): Ditto.
2343 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2345 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2346 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2348 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2351 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2352 TARGET_NEON and TARGET_IWMMXT.
2353 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2354 TARGET_NEON and TARGET_IWMMXT.
2355 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2357 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2359 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2360 (ix86_rewrite_tls_address): Ditto.
2361 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2362 (ix86_rewrite_tls_address_1): Ditto.
2363 (ix86_rewrite_tls_address): Ditto.
2364 * config/i386/predicates.md (tls_address_pattern): New predicate.
2365 * config/i386/i386.md (TLS address splitter): New splitter.
2367 2017-09-04 Richard Biener <rguenther@suse.de>
2369 PR tree-optimization/82084
2370 * fold-const.h (can_native_encode_string_p): Declare.
2371 * fold-const.c (can_native_encode_string_p): Factor out from ...
2372 (native_encode_string): ... here.
2373 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2374 vectorizing stores from constants we later cannot handle.
2376 2017-09-04 Marek Polacek <polacek@redhat.com>
2379 * doc/invoke.texi: Update -Wtautological-compare documentation.
2381 2017-09-04 Jeff Law <law@redhat.com>
2383 PR tree-optimization/64910
2384 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2385 swap the first and last operand if the last is a constant.
2387 2017-09-04 Marek Polacek <polacek@redhat.com>
2390 * convert.c (do_narrow): When sanitizing signed integer overflows,
2391 bail out for signed types.
2392 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2394 2017-09-04 Richard Biener <rguenther@suse.de>
2396 PR tree-optimization/82060
2397 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2398 Move devirtualization after stmt folding and before EH/AB/noreturn
2399 cleanup to get the stmt refs canonicalized. Use a bool instead
2400 of gimple_modified_p since that doesn't work for NOPs. Schedule
2401 NOPs generated by folding for removal.
2403 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2404 Alan Hayward <alan.hayward@arm.com>
2405 David Sherwood <david.sherwood@arm.com>
2407 * coretypes.h (pad_direction): New enum.
2408 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2409 (FUNCTION_ARG_PADDING): Likewise.
2410 * target.def (function_arg_padding): New hook.
2411 * targhooks.h (default_function_arg_padding): Declare.
2412 * targhooks.c (default_function_arg_padding): New function.
2413 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2414 (TARGET_FUNCTION_ARG_PADDING): ...this.
2415 * doc/tm.texi: Regenerate.
2416 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2417 instead of direction.
2418 (compute_argument_addresses): Likewise.
2419 (load_register_parameters): Likewise.
2420 (emit_library_call_value_1): Likewise.
2421 (store_one_arg): Use targetm.calls.function_arg_padding instead
2422 of FUNCTION_ARG_PADDING.
2423 (must_pass_in_stack_var_size_or_pad): Likewise.
2424 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2425 (emit_group_store): Likewise.
2426 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2427 instead of FUNCTION_ARG_PADDING.
2428 (emit_push_insn): Likewise, and propagate enum change throughout
2430 * function.h (direction): Delete.
2431 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2433 * function.c (assign_parm_find_stack_rtl): Likewise.
2434 (assign_parm_setup_block_p): Likewise.
2435 (assign_parm_setup_block): Likewise.
2436 (gimplify_parameters): Likewise.
2437 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2438 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2440 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2441 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2442 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2443 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2444 (aarch64_function_arg_padding): ...this new function.
2445 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2446 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2447 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2448 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2449 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2450 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2451 (arm_pad_arg_upward): Replace with...
2452 (arm_function_arg_padding): ...this new function.
2453 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2455 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2456 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2457 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2458 (ia64_hpux_function_arg_padding): Replace with...
2459 (ia64_function_arg_padding): ...this new function. Use pad_direction
2460 instead of direction. Check for TARGET_HPUX.
2461 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2462 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2463 (iq2000_function_arg_padding): New function.
2464 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2465 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2466 (mips_function_arg_padding): ...this new function.
2467 (mips_pad_reg_upward): Update accordingly.
2468 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2469 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2470 targetm.calls.function_arg_padding.
2471 (FUNCTION_ARG_PADDING): Delete.
2472 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2473 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2474 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2475 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2476 (nios2_block_reg_padding): Return pad_direction instead of direction.
2477 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2478 instead of direction.
2479 (nios2_function_arg_padding): Likewise. Make static.
2480 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2481 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2482 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2483 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2484 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2485 (pa_function_arg_padding): Make static. Return pad_direction instead
2487 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2488 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2489 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2490 instead of direction. Use targetm.calls.function_arg_padding.
2491 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2492 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2493 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2494 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2495 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2497 (function_arg_padding): Rename to...
2498 (rs6000_function_arg_padding): ...this. Make static. Return
2499 pad_direction instead of direction.
2500 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2501 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2502 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2503 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2504 instead of direction. Use targetm.calls.function_arg_padding.
2505 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2506 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2507 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2508 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2509 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2510 (function_arg_padding): Rename to...
2511 (rs6000_function_arg_padding): ...this. Make static. Return
2512 pad_direction instead of direction.
2513 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2514 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2515 * config/s390/s390.c (s390_function_arg_padding): New function.
2516 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2517 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2518 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2519 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2520 (function_arg_padding): Rename to...
2521 (sparc_function_arg_padding): ...this. Make static. Return
2522 pad_direction instead of direction.
2523 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2524 * config/spu/spu.c (spu_function_arg_padding): New function.
2525 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2526 * system.h (FUNCTION_ARG_PADDING): Poison.
2528 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2529 Alan Hayward <alan.hayward@arm.com>
2530 David Sherwood <david.sherwood@arm.com>
2532 * target.def (modes_tieable_p): New hook.
2533 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2534 (TARGET_MODES_TIEABLE_P): ...this.
2535 * doc/tm.texi.in: Regenerate.
2536 * hooks.h (hook_bool_mode_mode_true): Declare.
2537 * hooks.c (hook_bool_mode_mode_true): New function.
2538 * combine.c (subst): Use targetm.modes_tieable_p instead of
2540 * dse.c (find_shift_sequence): Likewise.
2541 * expmed.c (extract_low_bits): Likewise.
2542 * lower-subreg.c: Include target.h.
2543 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2545 * rtlanal.c (rtx_cost): Likewise.
2546 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2547 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2548 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2549 (TARGET_MODES_TIEABLE_P): Redefine.
2550 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2551 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2552 (TARGET_MODES_TIEABLE_P): Redefine.
2553 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2554 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2555 (arc_modes_tieable_p): New function.
2556 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2557 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2558 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2559 (arm_modes_tieable_p): Make static.
2560 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2561 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2562 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2563 (TARGET_MODES_TIEABLE_P): Redefine.
2564 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2565 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2566 (TARGET_MODES_TIEABLE_P): Redefine.
2567 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2568 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2569 (cr16_modes_tieable_p): New function.
2570 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2571 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2572 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2573 (TRULY_NOOP_TRUNCATION): Update comment.
2574 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2575 (TRULY_NOOP_TRUNCATION): Update comment.
2576 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2577 (frv_modes_tieable_p): New function.
2578 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2579 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2580 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2581 (TARGET_MODES_TIEABLE_P): Redefine.
2582 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2583 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2584 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2585 (TARGET_MODES_TIEABLE_P): Redefine.
2586 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2587 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2588 (ia64_modes_tieable_p): New function.
2589 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2590 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2591 (iq2000_modes_tieable_p): New function.
2592 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2593 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2594 (lm32_modes_tieable_p): New function.
2595 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2596 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2597 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2598 (TARGET_MODES_TIEABLE_P): Redefine.
2599 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2600 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2601 (m32r_modes_tieable_p): New function.
2602 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2603 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2604 (m68k_modes_tieable_p): New function.
2605 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2606 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2607 (mcore_modes_tieable_p): New function.
2608 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2609 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2611 (TARGET_MODES_TIEABLE_P): Redefine.
2612 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2613 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2614 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2615 (TARGET_MODES_TIEABLE_P): Redefine.
2616 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2617 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2618 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2619 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2620 (mn10300_modes_tieable_p): ...this and make static.
2621 (TARGET_MODES_TIEABLE_P): Redefine.
2622 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2623 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2624 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2625 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2626 (msp430_modes_tieable_p): Make static.
2627 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2628 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2629 (TARGET_MODES_TIEABLE_P): Redefine.
2630 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2631 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2632 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2633 (TARGET_MODES_TIEABLE_P): Redefine.
2634 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2635 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2636 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2637 (TARGET_MODES_TIEABLE_P): Redefine.
2638 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2639 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2640 (pdp11_modes_tieable_p): New function.
2641 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2642 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2643 (rs6000_modes_tieable_p): New function.
2644 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2645 * config/powerpcspe/powerpcspe.md: Update comment.
2646 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2647 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2648 (TARGET_MODES_TIEABLE_P): Redefine.
2649 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2650 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2651 (rl78_modes_tieable_p): New function.
2652 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2653 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2654 (rs6000_modes_tieable_p): New function.
2655 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2656 * config/rs6000/rs6000.md: Update comment.
2657 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2658 * config/rx/rx.c (rx_modes_tieable_p): New function.
2659 (TARGET_MODES_TIEABLE_P): Redefine.
2660 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2661 * config/s390/s390.c (s390_modes_tieable_p): New function.
2662 (TARGET_MODES_TIEABLE_P): Redefine.
2663 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2664 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2665 (sh_modes_tieable_p): New function.
2666 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2667 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2668 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2669 (sparc_modes_tieable_p): Make static.
2670 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2671 * config/spu/spu.c (spu_modes_tieable_p): New function.
2672 (TARGET_MODES_TIEABLE_P): Redefine.
2673 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2674 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2675 (TARGET_MODES_TIEABLE_P): Redefine.
2676 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2677 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2678 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2679 * config/v850/v850.c (v850_modes_tieable_p): New function.
2680 (TARGET_MODES_TIEABLE_P): Redefine.
2681 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2682 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2683 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2684 (visium_modes_tieable_p): New function.
2685 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2686 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2687 (xtensa_modes_tieable_p): New function.
2688 * system.h (MODES_TIEABLE_P): Poison.
2690 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2691 Alan Hayward <alan.hayward@arm.com>
2692 David Sherwood <david.sherwood@arm.com>
2694 * target.def (hard_regno_mode_ok): New hook.
2695 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2696 (TARGET_HARD_REGNO_MODE_OK): ...this.
2697 * doc/tm.texi.in: Regenerate.
2698 * hooks.h (hook_bool_uint_mode_true): Declare.
2699 * hooks.c (hook_bool_uint_mode_true): New function.
2700 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2702 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2703 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2704 instead of HARD_REGNO_MODE_OK.
2705 * caller-save.c: Include target.h.
2706 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2708 * combine.c (can_combine_p): Likewise.
2709 (combinable_i3pat): Likewise.
2710 (can_change_dest_mode): Likewise.
2711 * expr.c (init_expr_target): Likewise.
2712 (convert_move): Likewise.
2713 (convert_modes): Likewise.
2714 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2715 (setup_prohibited_mode_move_regs): Likewise.
2716 * ira.h (target_ira): Likewise.
2717 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2718 * lra-constraints.c (process_alt_operands): Likewise.
2719 (split_reg): Likewise.
2720 * recog.c (peep2_find_free_register): Likewise.
2721 * ree.c (combine_reaching_defs): Likewise.
2722 * regcprop.c (maybe_mode_change): Likewise.
2723 * reginfo.c (init_reg_sets_1): Likewise.
2724 (choose_hard_reg_mode): Likewise.
2725 (simplifiable_subregs): Likewise.
2726 * regrename.c (check_new_reg_p): Likewise.
2727 * reload.c (find_valid_class): Likewise.
2728 (find_valid_class_1): Likewise.
2729 (reload_inner_reg_of_subreg): Likewise.
2730 (push_reload): Likewise.
2731 (combine_reloads): Likewise.
2732 (find_dummy_reload): Likewise.
2733 (find_reloads): Likewise.
2734 * reload1.c (find_reg): Likewise.
2735 (set_reload_reg): Likewise.
2736 (allocate_reload_reg): Likewise.
2737 (choose_reload_regs): Likewise.
2738 (reload_adjust_reg_for_temp): Likewise.
2739 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2740 (simplify_subreg_regno): Likewise.
2741 * sel-sched.c (init_regs_for_mode): Likewise.
2742 * varasm.c (make_decl_rtl): Likewise.
2743 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2744 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2746 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2747 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2748 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2749 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2750 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2751 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2752 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2753 (arc_mode_class): Delete.
2754 (HARD_REGNO_MODE_OK): Delete.
2755 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2756 (arc_hard_regno_mode_ok): Rename old array to...
2757 (arc_hard_regno_mode_ok_modes): ...this.
2758 (arc_conditional_register_usage): Update accordingly.
2759 (arc_mode_class): Make static.
2760 (arc_hard_regno_mode_ok): New function.
2761 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2762 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2763 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2764 (arm_hard_regno_mode_ok): Make static.
2765 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2767 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2768 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2769 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2771 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2772 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2773 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2774 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2775 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2776 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2777 * config/bfin/predicates.md (valid_reg_operand): Use
2778 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2779 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2780 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2781 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2782 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2783 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2784 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2785 (cr16_hard_regno_mode_ok): Make static and return a bool.
2786 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2787 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2788 (cris_hard_regno_mode_ok): New function.
2789 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2790 (epiphany_mode_class): Delete.
2791 (HARD_REGNO_MODE_OK): Delete.
2792 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2793 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2794 (hard_regno_mode_ok): Rename to...
2795 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2796 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2797 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2799 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2800 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2801 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2802 (frv_hard_regno_mode_ok): Make static and return a bool.
2803 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2805 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2806 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2807 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2808 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2810 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2811 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2812 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2813 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2815 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2816 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2817 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2818 (ia64_hard_regno_mode_ok): New function.
2819 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2820 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2821 (iq2000_hard_regno_mode_ok): New function.
2822 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2823 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2824 (lm32_hard_regno_mode_ok): New function.
2825 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2826 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2827 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2828 instead of HARD_REGNO_MODE_OK.
2829 (m32c_hard_regno_ok): Rename to...
2830 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2831 (m32c_cannot_change_mode_class): Update accordingly.
2832 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2833 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2834 (m32r_mode_class): Delete.
2835 (HARD_REGNO_MODE_OK): Delete.
2836 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2837 (m32r_hard_regno_mode_ok): Rename to...
2838 (m32r_hard_regno_modes): ...this.
2839 (m32r_mode_class): Make static.
2840 (m32r_hard_regno_mode_ok): New function.
2841 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2842 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2843 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2844 (m68k_hard_regno_mode_ok): Make static.
2845 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2846 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2847 (mcore_hard_regno_mode_ok): New function.
2848 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2849 (HARD_REGNO_MODE_OK): Delete.
2850 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2852 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2853 (microblaze_hard_regno_mode_ok): New function.
2854 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2855 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2856 (mips_hard_regno_mode_ok): Delete.
2857 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2858 (mips_hard_regno_mode_ok_p): ...this and make static.
2859 (mips_hard_regno_mode_ok_p): Rename to...
2860 (mips_hard_regno_mode_ok_uncached): ...this.
2861 (mips_hard_regno_mode_ok): New function.
2862 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2863 of HARD_REGNO_MODE_OK.
2864 (mips_option_override): Update after above name changes.
2865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2866 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2867 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2868 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2869 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2870 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2871 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2872 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2873 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2874 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2875 (msp430_hard_regno_mode_ok): Make static and return a bool.
2876 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2877 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2878 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2880 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2881 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2882 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2883 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2884 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2885 (PA_HARD_REGNO_MODE_OK): ...this
2886 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2887 (PA_HARD_REGNO_MODE_OK): ...this.
2888 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2889 (pa_hard_regno_mode_ok): New function.
2890 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2891 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2892 (pdp11_hard_regno_mode_ok): New function.
2893 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2894 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2896 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2898 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2899 (rs6000_hard_regno_mode_ok): Rename to...
2900 (rs6000_hard_regno_mode_ok_uncached): ...this.
2901 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2902 (rs6000_hard_regno_mode_ok): New function.
2903 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2904 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2905 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2906 (riscv_hard_regno_mode_ok): ...this and make static.
2907 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2908 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2909 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2910 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2911 (rl78_hard_regno_mode_ok): Make static and return bool.
2912 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2913 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2915 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2916 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2917 (rs6000_hard_regno_mode_ok): Rename to...
2918 (rs6000_hard_regno_mode_ok_uncached): ...this.
2919 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2920 (rs6000_hard_regno_mode_ok): New function.
2921 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2922 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2923 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2924 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2925 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2926 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2928 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2929 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2930 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2931 (sh_hard_regno_mode_ok): Make static.
2932 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2933 instead of HARD_REGNO_MODE_OK.
2934 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2935 (sparc_mode_class): Delete.
2936 (HARD_REGNO_MODE_OK): Delete.
2937 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2938 (hard_regno_mode_classes): Make static.
2939 (sparc_mode_class): Likewise.
2940 (sparc_hard_regno_mode_ok): New function.
2941 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2942 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2943 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2945 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2946 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2947 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2948 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2949 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2950 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2951 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2952 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2953 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2954 (visium_hard_regno_mode_ok): New function.
2955 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2956 instead of HARD_REGNO_MODE_OK.
2957 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2958 (HARD_REGNO_MODE_OK): Delete.
2959 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2960 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2961 (xtensa_option_override): Update accordingly.
2962 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2963 (xtensa_hard_regno_mode_ok): New function.
2964 * system.h (HARD_REGNO_MODE_OK): Poison.
2966 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2967 Alan Hayward <alan.hayward@arm.com>
2968 David Sherwood <david.sherwood@arm.com>
2970 * target.def (hard_regno_call_part_clobbered): New hook.
2971 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2972 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2973 * doc/tm.texi: Regenerate.
2974 * hooks.h (hook_bool_uint_mode_false): Declare.
2975 * hooks.c (hook_bool_uint_mode_false): New function.
2976 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2977 * cselib.c (cselib_process_insn): Use
2978 targetm.hard_regno_call_part_clobbered instead of
2979 HARD_REGNO_CALL_PART_CLOBBERED.
2980 * ira-conflicts.c (ira_build_conflicts): Likewise.
2981 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2982 * lra-constraints.c (need_for_call_save_p): Likewise.
2983 * lra-lives.c: Include target.h.
2984 (check_pseudos_live_through_calls): Use
2985 targetm.hard_regno_call_part_clobbered instead of
2986 HARD_REGNO_CALL_PART_CLOBBERED.
2987 * regcprop.c: Include target.h.
2988 (copyprop_hardreg_forward_1): Use
2989 targetm.hard_regno_call_part_clobbered instead of
2990 HARD_REGNO_CALL_PART_CLOBBERED.
2991 * reginfo.c (choose_hard_reg_mode): Likewise.
2992 * regrename.c (check_new_reg_p): Likewise.
2993 * reload.c (find_equiv_reg): Likewise.
2994 * reload1.c (emit_reload_insns): Likewise.
2995 * sched-deps.c (deps_analyze_insn): Likewise.
2996 * sel-sched.c (init_regs_for_mode): Likewise.
2997 (mark_unavailable_hard_regs): Likewise.
2998 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2999 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3000 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3002 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3003 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3004 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3006 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3008 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3009 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3010 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3012 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3013 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3014 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3016 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3017 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3019 * config/powerpcspe/powerpcspe.c
3020 (rs6000_hard_regno_call_part_clobbered): New function.
3021 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3022 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3023 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3025 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3026 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3027 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3029 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3030 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3031 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3033 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3034 Alan Hayward <alan.hayward@arm.com>
3035 David Sherwood <david.sherwood@arm.com>
3037 * rtl.h (subreg_memory_offset): Declare.
3038 * emit-rtl.c (subreg_memory_offset): New function.
3039 * expmed.c (store_bit_field_1): Use it.
3040 * expr.c (undefined_operand_subword_p): Likewise.
3041 * simplify-rtx.c (simplify_subreg): Likewise.
3043 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3045 PR rtl-optimization/57448
3048 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3049 using atomic_load pattern.
3050 (expand_atomic_store): Likewise.
3052 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3055 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3056 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3057 handling. Use replace_call_with_value with NULL instead of
3058 gsi_replace, unlink_stmt_vdef and release_defs.
3060 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3063 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3065 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3068 * rtl.h (emit_library_call_value_1): Declare.
3069 (emit_library_call): Replace declaration with a series of overloads.
3070 Remove the parameter count argument.
3071 (emit_library_call_value): Likewise.
3072 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3073 with an "rtx_mode_t *".
3074 (emit_library_call_value): Delete.
3075 (emit_library_call): Likewise.
3076 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3077 (asan_emit_allocas_unpoison): Likewise.
3078 * builtins.c (expand_builtin_powi): Likewise.
3079 (expand_asan_emit_allocas_unpoison): Likewise.
3080 * cfgexpand.c (expand_main_function): Likewise.
3081 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
3082 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
3083 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
3084 * config/arm/arm.c (arm_trampoline_init): Likewise.
3085 (arm_call_tls_get_addr): Likewise.
3086 (arm_expand_divmod_libfunc): Likewise.
3087 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
3088 (smulsi3_highpart): Likewise.
3089 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
3090 (c6x_expand_compare): Likewise.
3091 (c6x_expand_movmem): Likewise.
3092 * config/frv/frv.c (frv_trampoline_init): Likewise.
3093 * config/i386/i386.c (ix86_trampoline_init): Likewise.
3094 (ix86_expand_divmod_libfunc): Likewise.
3095 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
3096 (ia64_expand_compare): Likewise.
3097 (ia64_profile_hook): Likewise.
3098 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
3099 (nonlocal_goto): Likewise.
3100 (restore_stack_nonlocal): Likewise.
3101 * config/m32r/m32r.c (block_move_call): Likewise.
3102 (m32r_trampoline_init): Likewise.
3103 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3104 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3105 (m68k_call_m68k_read_tp): Likewise.
3106 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3107 (microblaze_expand_divide): Likewise.
3108 * config/mips/mips.h (mips_args): Likewise.
3109 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
3110 (MIPS_ICACHE_SYNC): Likewise.
3111 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
3112 (nios2_trampoline_init): Likewise.
3113 * config/pa/pa.c (hppa_tls_call): Likewise.
3114 (pa_trampoline_init): Likewise.
3115 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
3116 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
3117 (expand_strn_compare): Likewise.
3118 (rs6000_generate_compare): Likewise.
3119 (rs6000_expand_float128_convert): Likewise.
3120 (output_profile_hook): Likewise.
3121 (rs6000_trampoline_init): Likewise.
3122 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
3123 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
3124 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
3125 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
3126 (rs6000_generate_compare): Likewise.
3127 (rs6000_expand_float128_convert): Likewise.
3128 (output_profile_hook): Likewise.
3129 (rs6000_trampoline_init): Likewise.
3130 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3131 * config/sh/sh.c (sh_trampoline_init): Likewise.
3132 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3133 (sparc_emit_float_lib_cmp): Likewise.
3134 (sparc32_initialize_trampoline): Likewise.
3135 (sparc64_initialize_trampoline): Likewise.
3136 (sparc_profile_hook): Likewise.
3137 * config/spu/spu.c (ea_load_store): Likewise.
3138 * config/spu/spu.md (floatunssidf2): Likewise.
3139 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3140 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3141 * config/visium/visium.c (expand_block_move_4): Likewise.
3142 (expand_block_move_2): Likewise.
3143 (expand_block_move_1): Likewise.
3144 (expand_block_set_4): Likewise.
3145 (expand_block_set_2): Likewise.
3146 (expand_block_set_1): Likewise.
3147 (visium_trampoline_init): Likewise.
3148 (visium_profile_hook): Likewise.
3149 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3150 (xtensa_setup_frame_addresses): Likewise.
3151 (xtensa_trampoline_init): Likewise.
3152 * except.c (sjlj_emit_function_enter): Likewise.
3153 (sjlj_emit_function_exit): Likewise.
3154 * explow.c (allocate_dynamic_stack_space): Likewise.
3155 (probe_stack_range): Likewise.
3156 * expr.c (convert_mode_scalar): Likewise.
3157 * optabs.c (expand_binop): Likewise.
3158 (expand_twoval_binop_libfunc): Likewise.
3159 (expand_unop): Likewise.
3160 (prepare_cmp_insn): Likewise.
3161 (prepare_float_lib_cmp): Likewise.
3162 (expand_float): Likewise.
3163 (expand_fix): Likewise.
3164 (expand_fixed_convert): Likewise.
3165 (maybe_emit_sync_lock_test_and_set): Likewise.
3166 (expand_atomic_compare_and_swap): Likewise.
3167 (expand_mem_thread_fence): Likewise.
3168 (expand_atomic_fetch_op): Likewise.
3170 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3172 * doc/generic.texi (OpenACC): Adjust URL.
3173 * doc/invoke.texi (C Dialect Options): Ditto.
3175 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3177 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3178 predicate for operand 1. Add (m,<S>) constraint.
3179 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3180 Prevent memory operand 1 with register operand 2.
3182 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3184 PR rtl-optimization/82024
3185 * combine.c (try_combine): If the combination result is a PARALLEL,
3186 and we only need to retain the SET in there that would be placed
3187 at I2, check that we can place that at I3 instead, before doing so.
3189 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3192 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3194 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3195 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3198 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3199 Jeff Law <law@redhat.com>
3201 * varasm.c (bss_initializer_p): Do not put constants into .bss
3202 (categorize_decl_for_section): Handle bss_initializer_p returning
3203 false when DECL_INITIAL is NULL.
3205 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3208 * config/s390/s390.c (s390_can_inline_p): New function.
3210 2017-09-01 Jeff Law <law@redhat.com>
3212 PR tree-optimization/82052
3213 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3214 Always initialize the returned slot after a hash table miss
3215 when INSERT is true.
3217 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3219 * config/s390/s390.md (mem_signal_fence): Remove.
3220 * doc/md.texi (mem_signal_fence): Remove.
3221 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3223 * target-insns.def (mem_signal_fence): Remove.
3225 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3228 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3231 * asan.c (create_odr_indicator): Strip name encoding from assembler
3232 name before appending it after __odr_asan_.
3234 2017-09-01 Martin Liska <mliska@suse.cz>
3236 PR tree-optimization/82059
3237 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3238 frequency only when an edge is redirected.
3240 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3242 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3243 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3244 (arc_conditional_register_usage): Remove ARC600 lp_count
3246 (arc_file_start): Emit Tag_ARC_CPU_variation.
3247 (arc_can_use_doloop_p): New conditions to use ZOLs.
3248 (hwloop_fail): New function.
3249 (hwloop_optimize): Likewise.
3250 (hwloop_pattern_reg): Likewise.
3251 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3252 (arc_reorg_loops): New function, calls reorg_loops.
3253 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3254 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3256 (arc_loop_hazard): Remove function, functionality moved into
3258 (arc_hazard): Remove arc_loop_hazard call.
3259 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3260 into hwloop_optimize.
3261 (arc_label_align): Remove ZOL handling.
3262 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3263 * config/arc/arc.md (doloop_begin): Remove pattern.
3264 (doloop_begin_i): Likewise.
3265 (doloop_end_i): Likewise.
3266 (doloop_fallback): Likewise.
3267 (doloop_fallback_m): Likewise.
3268 (doloop_end): Reimplement expand.
3269 (arc_lp): New pattern for LP instruction.
3270 (loop_end): New pattern.
3271 (loop_fail): Likewise.
3272 (decrement_and_branch_until_zero): Likewise.
3273 * config/arc/arc.opt (mlpc-width): New option.
3274 * doc/invoke.texi (mlpc-width): Document option.
3276 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3278 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3279 (arc_ccfsm_advance): Fix checking for delay slots.
3280 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3282 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3284 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3286 (movhi_insn): Update store instruction constraint which are saving
3287 6-bit short immediates.
3288 (movsi_insn): Consider also short scaled load operations.
3289 (zero_extendhisi2_i): Use Usd constraint instead of T.
3290 (extendhisi2_i): Add q constraint.
3291 (arc_clzsi2): Add type and length attributes.
3292 (arc_ctzsi2): Likewise.
3293 * config/arc/constraints.md (Usc): Update constraint, the
3294 assembler can parse two relocations for a single instruction.
3296 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3298 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3299 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3301 2017-08-31 Olivier Hainque <hainque@adacore.com>
3303 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3304 match as powerpc-wrs-vxworks*.
3306 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3308 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3309 register constraint for by-element operand.
3310 (aarch64_mls_elt_merge<mode>): Likewise.
3312 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3314 * config/arc/arc.c (arc_can_follow_jump): Check for short
3317 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3319 * config.gcc: Use g.opt for arc.
3320 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3321 functionality moved to ...
3322 (legitimate_scaled_address_p): New function, ...here.
3323 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3324 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3325 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3327 (arc_override_options): Handle G option.
3328 (arc_output_pic_addr_const): Correct function definition.
3329 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3330 (arc_decl_anon_ns_mem_p): Delete.
3331 (arc_in_small_data_p): Overhaul this function to take into
3332 consideration the value given via G option.
3333 (arc_rewrite_small_data_1): Renamed and corrected old
3334 arc_rewrite_small_data function.
3335 (arc_rewrite_small_data): New function.
3336 (small_data_pattern): Don't use pic_offset_table_rtx.
3337 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3338 * config/arc/simdext.md (movmisalignv2hi): Use
3339 prepare_move_operands function.
3341 (movmisalign*): Likewise.
3342 * doc/invoke.texi (ARC options): Document -G option.
3344 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3346 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3348 * config/arc/arc.c (arc_print_operand): Output scalled address for
3349 sdata whenever is possible.
3350 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3351 load/stores are available.
3352 (compact_sda_memory_operand): Check for the alignment required by
3353 code density instructions.
3354 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3356 * config/arc/constraints.md (Usd): Update constraint.
3357 (Us0): New constraint.
3358 (Usc): Update constraint.
3360 2017-08-31 Richard Biener <rguenther@suse.de>
3363 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3366 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3368 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3369 Resize type_signature.
3371 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3372 Alan Hayward <alan.hayward@arm.com>
3373 David Sherwood <david.sherwood@arm.com>
3375 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3376 subregs whose inner modes can be stored in GPRs.
3377 (aarch64_classify_index): Likewise.
3379 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3380 Alan Hayward <alan.hayward@arm.com>
3381 David Sherwood <david.sherwood@arm.com>
3383 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3384 (V_INT_EQUIV): ...this.
3385 (v_cmp_result): Rename to...
3386 (v_int_equiv): ...this.
3387 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3388 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3389 (copysign<mode>3): Likewise.
3390 (aarch64_simd_bsl<mode>_internal): Likewise.
3391 (aarch64_simd_bsl<mode>): Likewise.
3392 (vec_cmp<mode><mode>): Likewise.
3393 (vcond<mode><mode>): Likewise.
3394 (vcond<v_cmp_mixed><mode>): Likewise.
3395 (vcondu<mode><v_cmp_mixed>): Likewise.
3396 (aarch64_cm<optab><mode>): Likewise.
3397 (aarch64_cmtst<mode>): Likewise.
3398 (aarch64_fac<optab><mode>): Likewise.
3399 (vec_perm_const<mode>): Likewise.
3400 (vcond_mask_<mode><v_cmp_result>): Rename to...
3401 (vcond_mask_<mode><v_int_equiv>): ...this.
3402 (vec_cmp<mode><v_cmp_result>): Rename to...
3403 (vec_cmp<mode><v_int_equiv>): ...this.
3405 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3406 Alan Hayward <alan.hayward@arm.com>
3407 David Sherwood <david.sherwood@arm.com>
3409 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3411 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3412 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3413 (UNSPEC_LD4_DREG): New unspecs.
3414 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3415 (aarch64_ld2<mode>_dreg_be): Replace with...
3416 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3418 (aarch64_ld3<mode>_dreg_le)
3419 (aarch64_ld3<mode>_dreg_be): Replace with...
3420 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3422 (aarch64_ld4<mode>_dreg_le)
3423 (aarch64_ld4<mode>_dreg_be): Replace with...
3424 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3427 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3429 PR tree-optimization/81987
3430 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3431 insert an initializer in a location not dominated by the stride
3434 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3436 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3437 on the entire header of the finally block in the fallthru case.
3439 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3441 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3443 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3445 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3446 rs6000_emit_move_from_cr and call renamed function.
3447 (rs6000_emit_prologue): Call renamed functions.
3448 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3449 movesi_from_cr, remove volatile CRs.
3451 2017-08-30 Jon Beniston <jon@beniston.com>
3452 Richard Biener <rguenther@suse.de>
3454 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3455 of VECTOR_MODE_P check.
3456 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3457 element vector types.
3459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3461 * df.h (df_read_modify_subreg_p): Remove in favor of...
3462 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3463 const_rtx instead of an rtx.
3464 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3465 * df-problems.c (df_word_lr_mark_ref): Likewise.
3466 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3467 (mark_pseudo_reg_dead): Likewise.
3468 (mark_ref_dead): Likewise.
3469 * reginfo.c (init_subregs_of_mode): Likewise.
3470 * sched-deps.c (sched_analyze_1): Likewise.
3471 * df-scan.c (df_def_record_1): Likewise.
3472 (df_uses_record): Likewise.
3473 (df_read_modify_subreg_p): Remove in favor of...
3474 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3475 const_rtx instead of an rtx.
3477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3478 Alan Hayward <alan.hayward@arm.com>
3479 David Sherwood <david.sherwood@arm.com>
3481 * rtl.h (partial_subreg_p): New function.
3482 * caller-save.c (save_call_clobbered_regs): Use it.
3483 * calls.c (expand_call): Likewise.
3484 * combine.c (combinable_i3pat): Likewise.
3485 (simplify_set): Likewise.
3486 (make_extraction): Likewise.
3487 (make_compound_operation_int): Likewise.
3488 (gen_lowpart_or_truncate): Likewise.
3489 (force_to_mode): Likewise.
3490 (make_field_assignment): Likewise.
3491 (reg_truncated_to_mode): Likewise.
3492 (record_truncated_value): Likewise.
3493 (move_deaths): Likewise.
3494 * cse.c (record_jump_cond): Likewise.
3495 (cse_insn): Likewise.
3496 * cselib.c (cselib_lookup_1): Likewise.
3497 * expmed.c (extract_bit_field_using_extv): Likewise.
3498 * function.c (assign_parm_setup_reg): Likewise.
3499 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3500 * ira-build.c (create_insn_allocnos): Likewise.
3501 * lra-coalesce.c (merge_pseudos): Likewise.
3502 * lra-constraints.c (match_reload): Likewise.
3503 (simplify_operand_subreg): Likewise.
3504 (curr_insn_transform): Likewise.
3505 * lra-lives.c (process_bb_lives): Likewise.
3506 * lra.c (new_insn_reg): Likewise.
3507 (lra_substitute_pseudo): Likewise.
3508 * regcprop.c (mode_change_ok): Likewise.
3509 (maybe_mode_change): Likewise.
3510 (copyprop_hardreg_forward_1): Likewise.
3511 * reload.c (push_reload): Likewise.
3512 (find_reloads): Likewise.
3513 (find_reloads_subreg_address): Likewise.
3514 * reload1.c (alter_reg): Likewise.
3515 (eliminate_regs_1): Likewise.
3516 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3518 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3520 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3521 back to if statements, including unpack.
3523 2017-08-30 Martin Liska <mliska@suse.cz>
3526 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3528 (func_checker::compare_asm_inputs_outputs): ... this function.
3529 (func_checker::compare_gimple_asm): Use the function to compare
3530 also ASM constrains.
3531 * ipa-icf-gimple.h: Rename the function.
3533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3534 Alan Hayward <alan.hayward@arm.com>
3535 David Sherwood <david.sherwood@arm.com>
3537 * coretypes.h (complex_mode): New type.
3538 * gdbhooks.py (build_pretty_printer): Handle it.
3539 * machmode.h (complex_mode): New class.
3540 (complex_mode::includes_p): New function.
3541 (is_complex_int_mode): Likewise.
3542 (is_complex_float_mode): Likewise.
3543 * genmodes.c (get_mode_class): Handle complex mode classes.
3544 * function.c (expand_function_end): Use is_complex_int_mode.
3546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3547 Alan Hayward <alan.hayward@arm.com>
3548 David Sherwood <david.sherwood@arm.com>
3550 * coretypes.h (scalar_mode_pod): New typedef.
3551 * gdbhooks.py (build_pretty_printer): Handle it.
3552 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3553 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3554 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3555 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3559 Alan Hayward <alan.hayward@arm.com>
3560 David Sherwood <david.sherwood@arm.com>
3562 * machmode.h (mode_for_vector): Take a scalar_mode instead
3564 * stor-layout.c (mode_for_vector): Likewise.
3565 * explow.c (promote_mode): Use as_a <scalar_mode>.
3566 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3569 Alan Hayward <alan.hayward@arm.com>
3570 David Sherwood <david.sherwood@arm.com>
3572 * target.def (preferred_simd_mode): Take a scalar_mode
3573 instead of a machine_mode.
3574 * targhooks.h (default_preferred_simd_mode): Likewise.
3575 * targhooks.c (default_preferred_simd_mode): Likewise.
3576 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3577 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3578 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3579 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3580 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3581 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3582 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3583 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3585 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3586 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3587 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3588 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3589 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3590 * doc/tm.texi: Regenerate.
3591 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3595 Alan Hayward <alan.hayward@arm.com>
3596 David Sherwood <david.sherwood@arm.com>
3598 * target.def (scalar_mode_supported_p): Take a scalar_mode
3599 instead of a machine_mode.
3600 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3601 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3602 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3603 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3604 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3605 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3606 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3607 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3608 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3609 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3610 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3611 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3612 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3613 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3615 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3616 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3617 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3618 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3619 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3621 * doc/tm.texi: Regenerate.
3623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3624 Alan Hayward <alan.hayward@arm.com>
3625 David Sherwood <david.sherwood@arm.com>
3627 * coretypes.h (opt_scalar_mode): New typedef.
3628 * gdbhooks.py (build_pretty_printers): Handle it.
3629 * machmode.h (mode_iterator::get_2xwider): Add overload for
3631 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3633 * expr.c (convert_mode_scalar): Likewise.
3634 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3635 * optabs.c (expand_float): Likewise.
3636 (expand_fix): Likewise.
3637 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3640 Alan Hayward <alan.hayward@arm.com>
3641 David Sherwood <david.sherwood@arm.com>
3643 * optabs.c (expand_float): Explicitly check for scalars before
3644 using a branching expansion.
3645 (expand_fix): Likewise.
3647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3648 Alan Hayward <alan.hayward@arm.com>
3649 David Sherwood <david.sherwood@arm.com>
3651 * expr.c (convert_mode): Split scalar handling out into...
3652 (convert_mode_scalar): ...this new function. Treat the modes
3655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3656 Alan Hayward <alan.hayward@arm.com>
3657 David Sherwood <david.sherwood@arm.com>
3659 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3661 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3664 Alan Hayward <alan.hayward@arm.com>
3665 David Sherwood <david.sherwood@arm.com>
3667 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3668 rather than a machine_mode.
3669 (fixed_from_string): Likewise.
3670 (fixed_convert): Likewise.
3671 (fixed_convert_from_int): Likewise.
3672 (fixed_convert_from_real): Likewise.
3673 (real_convert_from_fixed): Likewise.
3674 * fixed-value.c (fixed_from_double_int): Likewise.
3675 (fixed_from_string): Likewise.
3676 (fixed_convert): Likewise.
3677 (fixed_convert_from_int): Likewise.
3678 (fixed_convert_from_real): Likewise.
3679 (real_convert_from_fixed): Likewise.
3680 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3683 Alan Hayward <alan.hayward@arm.com>
3684 David Sherwood <david.sherwood@arm.com>
3686 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3687 of separate mode class checks. Do not allow vector modes here.
3688 (immed_wide_int_const): Use as_a <scalar_mode>.
3689 * explow.c (trunc_int_for_mode): Likewise.
3690 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3691 (wi::shwi): Likewise.
3692 (wi::min_value): Likewise.
3693 (wi::max_value): Likewise.
3694 * dwarf2out.c (loc_descriptor): Likewise.
3695 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3699 Alan Hayward <alan.hayward@arm.com>
3700 David Sherwood <david.sherwood@arm.com>
3702 * tree.h (SCALAR_TYPE_MODE): New macro.
3703 * expr.c (expand_expr_addr_expr_1): Use it.
3704 (expand_expr_real_2): Likewise.
3705 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3706 (fold_convert_const_fixed_from_int): Likewise.
3707 (fold_convert_const_fixed_from_real): Likewise.
3708 (native_encode_fixed): Likewise
3709 (native_encode_complex): Likewise
3710 (native_encode_vector): Likewise.
3711 (native_interpret_fixed): Likewise.
3712 (native_interpret_real): Likewise.
3713 (native_interpret_complex): Likewise.
3714 (native_interpret_vector): Likewise.
3715 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3716 (simd_clone_adjust_argument_types): Likewise.
3717 (simd_clone_init_simd_arrays): Likewise.
3718 (simd_clone_adjust): Likewise.
3719 * stor-layout.c (layout_type): Likewise.
3720 * tree.c (build_minus_one_cst): Likewise.
3721 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3722 * tree-inline.c (estimate_move_cost): Likewise.
3723 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3724 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3725 (vectorizable_reduction): Likewise.
3726 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3727 (vect_recog_mixed_size_cond_pattern): Likewise.
3728 (check_bool_pattern): Likewise.
3729 (adjust_bool_pattern): Likewise.
3730 (search_type_for_mask_1): Likewise.
3731 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3732 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3733 (vectorizable_load): Likewise.
3734 (vectorizable_store): Likewise.
3735 * ubsan.c (ubsan_encode_value): Likewise.
3736 * varasm.c (output_constant): Likewise.
3738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3739 Alan Hayward <alan.hayward@arm.com>
3740 David Sherwood <david.sherwood@arm.com>
3742 * coretypes.h (scalar_mode): New class.
3743 * machmode.h (scalar_mode): Likewise.
3744 (scalar_mode::includes_p): New function.
3745 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3746 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3747 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3748 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3749 * expmed.c (store_bit_field_1): Likewise.
3750 (extract_bit_field_1): Likewise.
3751 * expr.c (write_complex_part): Likewise.
3752 (read_complex_part): Likewise.
3753 (emit_move_complex_push): Likewise.
3754 (expand_expr_real_2): Likewise.
3755 * function.c (assign_parm_setup_reg): Likewise.
3756 (assign_parms_unsplit_complex): Likewise.
3757 * optabs.c (expand_binop): Likewise.
3758 * rtlanal.c (subreg_get_info): Likewise.
3759 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3760 * varasm.c (output_constant_pool_2): Likewise.
3762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3763 Alan Hayward <alan.hayward@arm.com>
3764 David Sherwood <david.sherwood@arm.com>
3766 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3768 (expmed_mult_highpart_optab): Likewise.
3769 (expmed_mult_highpart): Likewise.
3771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3772 Alan Hayward <alan.hayward@arm.com>
3773 David Sherwood <david.sherwood@arm.com>
3775 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3776 instead of a machine_mode.
3777 (builtin_memset_read_str): Likewise.
3778 * builtins.c (c_readstr): Likewise.
3779 (builtin_memcpy_read_str): Likewise.
3780 (builtin_strncpy_read_str): Likewise.
3781 (builtin_memset_read_str): Likewise.
3782 (builtin_memset_gen_str): Likewise.
3783 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3784 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3785 instead of a machine_mode.
3786 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3788 (make_extraction): Likewise.
3789 (try_widen_shift_mode): Take and return scalar_int_modes instead
3791 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3792 a scalar_int_mode instead of a machine_mode.
3793 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3794 (avr_addr_space_pointer_mode): Likewise.
3795 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3796 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3797 (msp430_unwind_word_mode): Likewise.
3798 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3799 (spu_addr_space_pointer_mode): Likewise.
3800 (spu_addr_space_address_mode): Likewise.
3801 (spu_libgcc_cmp_return_mode): Likewise.
3802 (spu_libgcc_shift_count_mode): Likewise.
3803 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3804 (rl78_addr_space_pointer_mode): Likewise.
3805 (fl78_unwind_word_mode): Likewise.
3806 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3808 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3809 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3810 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3811 (mips_valid_pointer_mode): Likewise.
3812 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3813 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3814 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3816 (ft32_addr_space_address_mode): Likewise.
3817 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3818 scalar_int_mode instead of a machine_mode.
3819 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3821 (m32c_addr_space_address_mode): Likewise.
3822 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3823 (rs6000_eh_return_filter_mode): Likewise.
3824 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3825 (rs6000_eh_return_filter_mode): Likewise.
3826 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3827 (s390_libgcc_shift_count_mode): Likewise.
3828 (s390_unwind_word_mode): Likewise.
3829 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3831 * target.def (mode_rep_extended): Likewise.
3832 (valid_pointer_mode): Likewise.
3833 (addr_space.valid_pointer_mode): Likewise.
3834 (eh_return_filter_mode): Return a scalar_int_mode rather than
3836 (libgcc_cmp_return_mode): Likewise.
3837 (libgcc_shift_count_mode): Likewise.
3838 (unwind_word_mode): Likewise.
3839 (addr_space.pointer_mode): Likewise.
3840 (addr_space.address_mode): Likewise.
3841 * doc/tm.texi: Regenerate.
3842 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3844 (do_jump): Use scalar_int_mode for local variables.
3845 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3846 rather than a machine_mode.
3847 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3848 (scompare_loc_descriptor_wide): Likewise.
3849 (scompare_loc_descriptor_narrow): Likewise.
3850 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3852 * except.c (sjlj_emit_dispatch_table): Likewise.
3853 (expand_builtin_eh_copy_values): Likewise.
3854 * explow.c (convert_memory_address_addr_space_1): Likewise.
3855 Take a scalar_int_mode rather than a machine_mode.
3856 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3857 than a machine_mode.
3858 (memory_address_addr_space): Use scalar_int_mode for local variables.
3859 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3860 rather than a machine_mode.
3861 * expmed.c (mask_rtx): Likewise.
3862 (init_expmed_one_conv): Likewise.
3863 (expand_mult_highpart_adjust): Likewise.
3864 (extract_high_half): Likewise.
3865 (expmed_mult_highpart_optab): Likewise.
3866 (expmed_mult_highpart): Likewise.
3867 (expand_smod_pow2): Likewise.
3868 (expand_sdiv_pow2): Likewise.
3869 (emit_store_flag_int): Likewise.
3870 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3872 (extract_low_bits): Likewise.
3873 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3875 * expr.c (pieces_addr::adjust): Likewise.
3876 (can_store_by_pieces): Likewise.
3877 (store_by_pieces): Likewise.
3878 (clear_by_pieces_1): Likewise.
3879 (expand_expr_addr_expr_1): Likewise.
3880 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3881 (expand_expr_real_1): Likewise.
3882 (try_casesi): Likewise.
3883 * final.c (shorten_branches): Likewise.
3884 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3885 type of "mode" to machine_mode.
3886 * internal-fn.c (expand_arith_overflow_result_store): Take a
3887 scalar_int_mode rather than a machine_mode.
3888 (expand_mul_overflow): Use scalar_int_mode for local variables.
3889 * loop-doloop.c (doloop_modify): Likewise.
3890 (doloop_optimize): Likewise.
3891 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3892 than a machine_mode.
3893 (expand_doubleword_shift_condmove): Likewise.
3894 (expand_doubleword_shift): Likewise.
3895 (expand_doubleword_clz): Likewise.
3896 (expand_doubleword_popcount): Likewise.
3897 (expand_doubleword_parity): Likewise.
3898 (expand_absneg_bit): Use scalar_int_mode for local variables.
3899 (prepare_float_lib_cmp): Likewise.
3900 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3901 rather than a machine_mode.
3902 (convert_memory_address_addr_space): Likewise.
3903 (get_mode_bounds): Likewise.
3904 (get_address_mode): Return a scalar_int_mode rather than a
3906 * rtlanal.c (get_address_mode): Likewise.
3907 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3908 than a machine_mode.
3909 * targhooks.c (default_mode_rep_extended): Likewise.
3910 (default_valid_pointer_mode): Likewise.
3911 (default_addr_space_valid_pointer_mode): Likewise.
3912 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3913 than a machine_mode.
3914 (default_libgcc_cmp_return_mode): Likewise.
3915 (default_libgcc_shift_count_mode): Likewise.
3916 (default_unwind_word_mode): Likewise.
3917 (default_addr_space_pointer_mode): Likewise.
3918 (default_addr_space_address_mode): Likewise.
3919 * targhooks.h (default_eh_return_filter_mode): Likewise.
3920 (default_libgcc_cmp_return_mode): Likewise.
3921 (default_libgcc_shift_count_mode): Likewise.
3922 (default_unwind_word_mode): Likewise.
3923 (default_addr_space_pointer_mode): Likewise.
3924 (default_addr_space_address_mode): Likewise.
3925 (default_mode_rep_extended): Take a scalar_int_mode rather than
3927 (default_valid_pointer_mode): Likewise.
3928 (default_addr_space_valid_pointer_mode): Likewise.
3929 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3931 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3932 rather than a machine_mode.
3933 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3934 for local variables.
3935 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3936 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3937 than a machine_mode.
3939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3940 Alan Hayward <alan.hayward@arm.com>
3941 David Sherwood <david.sherwood@arm.com>
3943 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3944 the mode argument to scalar_int_mode.
3945 (do_jump_by_parts_zero_rtx): Likewise.
3946 (do_jump_by_parts_equality_rtx): Likewise.
3947 (do_jump_by_parts_greater): Take a mode argument.
3948 (do_jump_by_parts_equality): Likewise.
3949 (do_jump_1): Update calls accordingly.
3951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3952 Alan Hayward <alan.hayward@arm.com>
3953 David Sherwood <david.sherwood@arm.com>
3955 * is-a.h (safe_dyn_cast): New function.
3956 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3957 (jump_table_for_label): Likewise.
3958 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3959 instead of an rtx_insn *.
3960 (shorten_branches): Use dyn_cast instead of LABEL_P and
3961 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3962 rtx_jump_table_data::get_data_mode.
3963 (final_scan_insn): Likewise.
3965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3966 Alan Hayward <alan.hayward@arm.com>
3967 David Sherwood <david.sherwood@arm.com>
3969 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3970 trying to combine a full-register integer set with a subreg
3973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3974 Alan Hayward <alan.hayward@arm.com>
3975 David Sherwood <david.sherwood@arm.com>
3977 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3978 that is always either address_mode or pointer_mode.
3980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3981 Alan Hayward <alan.hayward@arm.com>
3982 David Sherwood <david.sherwood@arm.com>
3984 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3985 when the two are known to be equal.
3987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3988 Alan Hayward <alan.hayward@arm.com>
3989 David Sherwood <david.sherwood@arm.com>
3991 * simplify-rtx.c (simplify_const_unary_operation): Use
3992 is_a <scalar_int_mode> instead of checking for a nonzero
3993 precision. Forcibly convert op_mode to a scalar_int_mode
3994 in that case. More clearly differentiate the operand and
3995 result modes and use the former when deciding what the value
3996 of a count-bits operation should be. Use is_int_mode instead
3997 of checking for a MODE_INT. Remove redundant check for whether
3998 this mode has a zero precision.
4000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4001 Alan Hayward <alan.hayward@arm.com>
4002 David Sherwood <david.sherwood@arm.com>
4004 * optabs.c (widen_leading): Change the type of the mode argument
4005 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4006 (widen_bswap): Likewise.
4007 (expand_parity): Likewise.
4008 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4009 (expand_ffs): Likewise.
4010 (epand_unop): Check for scalar integer modes before calling the
4013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4014 Alan Hayward <alan.hayward@arm.com>
4015 David Sherwood <david.sherwood@arm.com>
4017 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4019 (expand_expr_real_1): Update call accordingly.
4021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4022 Alan Hayward <alan.hayward@arm.com>
4023 David Sherwood <david.sherwood@arm.com>
4025 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4026 value_mode arguments. Use scalar_int_mode internally.
4027 (store_bit_field_1): Rename the new integer mode from imode
4028 to op0_mode and use it instead of GET_MODE (op0). Update calls
4029 to store_split_bit_field, store_bit_field_using_insv and
4030 store_fixed_bit_field.
4031 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4032 Use scalar_int_mode internally. Use a bit count rather than a mode
4033 when calculating the largest bit size for get_best_mode.
4034 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4035 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4036 Remove assertion that OP0 has a scalar integer mode.
4037 (store_split_bit_field): Add op0_mode and value_mode arguments.
4038 Update calls to extract_fixed_bit_field.
4039 (extract_bit_field_using_extv): Add an op0_mode argument.
4040 Use scalar_int_mode internally.
4041 (extract_bit_field_1): Rename the new integer mode from imode to
4042 op0_mode and use it instead of GET_MODE (op0). Update calls to
4043 extract_split_bit_field, extract_bit_field_using_extv and
4044 extract_fixed_bit_field.
4045 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4046 to extract_split_bit_field and extract_fixed_bit_field_1.
4047 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4048 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4050 (extract_split_bit_field): Add an op0_mode argument. Update call
4051 to extract_fixed_bit_field.
4053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4054 Alan Hayward <alan.hayward@arm.com>
4055 David Sherwood <david.sherwood@arm.com>
4057 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4058 * explow.c (hard_function_value): Likewise.
4059 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4060 convert_to_mode call outside the loop.
4061 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4062 for the mode iterator. Require the mode specified by max_pieces
4064 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4066 (copy_blkmode_to_reg): Likewise.
4067 (set_storage_via_setmem): Likewise.
4068 * optabs.c (prepare_cmp_insn): Likewise.
4069 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4070 * stor-layout.c (finish_bitfield_representative): Likewise.
4072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4073 Alan Hayward <alan.hayward@arm.com>
4074 David Sherwood <david.sherwood@arm.com>
4076 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4077 * expr.c (convert_move): Use them.
4078 (convert_modes): Likewise.
4079 (store_expr_with_bounds): Likewise.
4081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4082 Alan Hayward <alan.hayward@arm.com>
4083 David Sherwood <david.sherwood@arm.com>
4085 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
4086 parameter for the mode of "x". Remove the "known_x", "known_mode"
4087 and "known_ret" arguments. Change the type of the mode argument
4089 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
4090 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
4091 (reg_num_sign_bit_copies_for_combine): Likewise.
4092 * rtlanal.c (nonzero_bits1): Likewise.
4093 (num_sign_bit_copies1): Likewise.
4094 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
4095 (reg_num_sign_bit_copies_general): Likewise.
4096 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
4097 (reg_nonzero_bits_general): Likewise.
4099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4100 Alan Hayward <alan.hayward@arm.com>
4101 David Sherwood <david.sherwood@arm.com>
4103 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4104 than in subroutines. Return 1 for non-integer modes.
4105 (cached_num_sign_bit_copies): Change the type of the mode parameter
4107 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
4108 classes. Handle CONST_INT_P first and then check whether X also
4109 has a scalar integer mode. Check the same thing for inner registers
4110 of a SUBREG and for values that are being extended or truncated.
4112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4113 Alan Hayward <alan.hayward@arm.com>
4114 David Sherwood <david.sherwood@arm.com>
4116 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
4117 in subroutines. Return the mode mask for non-integer modes.
4118 (cached_nonzero_bits): Change the type of the mode parameter
4120 (nonzero_bits1): Likewise. Remove early exit for other mode
4121 classes. Handle CONST_INT_P first and then check whether X
4122 also has a scalar integer mode.
4124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4125 Alan Hayward <alan.hayward@arm.com>
4126 David Sherwood <david.sherwood@arm.com>
4128 * expr.c (widest_int_mode_for_size): Make the comment match the code.
4129 Return a scalar_int_mode and assert that the size is greater than
4131 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4132 (op_by_pieces_d::op_by_pieces_d): Likewise.
4133 (op_by_pieces_d::run): Likewise.
4134 (can_store_by_pieces): Likewise.
4136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4137 Alan Hayward <alan.hayward@arm.com>
4138 David Sherwood <david.sherwood@arm.com>
4140 * combine.c (extract_left_shift): Add a mode argument and update
4142 (make_compound_operation_int): Change the type of the mode parameter
4143 to scalar_int_mode and update the call to extract_left_shift.
4145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4146 Alan Hayward <alan.hayward@arm.com>
4147 David Sherwood <david.sherwood@arm.com>
4149 * combine.c (simplify_and_const_int): Change the type of the mode
4150 parameter to scalar_int_mode.
4151 (simplify_and_const_int_1): Likewise. Update recursive call.
4153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4154 Alan Hayward <alan.hayward@arm.com>
4155 David Sherwood <david.sherwood@arm.com>
4157 * combine.c (simplify_compare_const): Check that the mode is a
4158 scalar_int_mode (rather than VOIDmode) before testing its
4160 (simplify_comparison): Move COMPARISON_P handling out of the
4161 loop and restrict the latter part of the loop to scalar_int_modes.
4162 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4163 and when considering SUBREG_REGs. Use is_int_mode instead of
4164 checking GET_MODE_CLASS against MODE_INT.
4166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4167 Alan Hayward <alan.hayward@arm.com>
4168 David Sherwood <david.sherwood@arm.com>
4170 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4171 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4172 shift_unit_mode and for modes involved in scalar shifts.
4174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4175 Alan Hayward <alan.hayward@arm.com>
4176 David Sherwood <david.sherwood@arm.com>
4178 * combine.c (force_int_to_mode): New function, split out from...
4179 (force_to_mode): ...here. Keep xmode up-to-date and use it
4180 instead of GET_MODE (x).
4182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4183 Alan Hayward <alan.hayward@arm.com>
4184 David Sherwood <david.sherwood@arm.com>
4186 * optabs-query.h (extraction_insn::struct_mode): Change type to
4187 opt_scalar_int_mode and update comment.
4188 (extraction_insn::field_mode): Change type to scalar_int_mode.
4189 (extraction_insn::pos_mode): Likewise.
4190 * combine.c (make_extraction): Update accordingly.
4191 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4192 (get_optab_extraction_insn): Likewise.
4193 * recog.c (simplify_while_replacing): Likewise.
4194 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4195 parameter to opt_scalar_int_mode.
4197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4198 Alan Hayward <alan.hayward@arm.com>
4199 David Sherwood <david.sherwood@arm.com>
4201 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4202 to a scalar_int_mode instead of a machine_mode.
4203 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4204 (get_best_mode): Return a boolean and use a pointer argument to store
4205 the selected mode. Replace the limit mode parameter with a bit limit.
4206 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4207 for the values returned by bit_field_mode_iterator::next_mode.
4208 (store_bit_field): Update call to get_best_mode.
4209 (store_fixed_bit_field): Likewise.
4210 (extract_fixed_bit_field): Likewise.
4211 * expr.c (optimize_bitfield_assignment_op): Likewise.
4212 * fold-const.c (optimize_bit_field_compare): Likewise.
4213 (fold_truth_andor_1): Likewise.
4214 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4215 Update for new type of m_mode.
4216 (get_best_mode): As above.
4218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4219 Alan Hayward <alan.hayward@arm.com>
4220 David Sherwood <david.sherwood@arm.com>
4222 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4223 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4224 (store_bit_field): Check is_a <scalar_int_mode> before calling
4225 strict_volatile_bitfield_p.
4226 (extract_bit_field): Likewise.
4228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4229 Alan Hayward <alan.hayward@arm.com>
4230 David Sherwood <david.sherwood@arm.com>
4232 * target.def (cstore_mode): Return a scalar_int_mode.
4233 * doc/tm.texi: Regenerate.
4234 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4235 * targhooks.h (default_cstore_mode): Likewise.
4236 * targhooks.c (default_cstore_mode): Likewise, using a forced
4238 * expmed.c (emit_cstore): Expect the target of the cstore to be
4241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4242 Alan Hayward <alan.hayward@arm.com>
4243 David Sherwood <david.sherwood@arm.com>
4245 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4247 (niter_desc): Likewise mode.
4248 (iv_analyze): Add a mode parameter.
4250 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4251 and change its type to scalar_int_mode.
4252 * loop-iv.c: Update commentary at head of file.
4253 (iv_constant): Pass the mode paraeter before the rtx it describes
4254 and change its type to scalar_int_mode. Remove VOIDmode handling.
4255 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4256 (iv_extend): Likewise.
4257 (shorten_into_mode): Likewise.
4258 (iv_add): Use scalar_int_mode.
4259 (iv_mult): Likewise.
4260 (iv_shift): Likewise.
4261 (canonicalize_iv_subregs): Likewise.
4262 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4263 it describes and change its mode to scalar_int_mode. Also change
4264 the type of the returned inner_mode to scalar_int_mode.
4265 (get_biv_step): Likewise, turning outer_mode from a pointer
4266 into a direct parameter. Update call to get_biv_step_1.
4267 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4268 iv_constant and get_biv_step.
4269 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4270 and change its type to scalar_int_mode. Don't initialise iv->mode
4271 to VOIDmode and remove later checks for its still being VOIDmode.
4272 Update calls to iv_analyze_op and iv_analyze_expr. Check
4273 is_a <scalar_int_mode> when changing the mode under consideration.
4274 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4275 Update call to iv_analyze_expr.
4276 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4277 inner register is not also a scalar_int_mode. Update call to
4279 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4280 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4281 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4282 separate mode class checks. Update calls to iv_analyze. Remove
4283 fix-up of VOIDmodes after iv_analyze_biv.
4284 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4285 don't have a scalar_int_mode. Update call to biv_p.
4287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4288 Alan Hayward <alan.hayward@arm.com>
4289 David Sherwood <david.sherwood@arm.com>
4291 * cfgexpand.c (convert_debug_memory_address): Use
4292 as_a <scalar_int_mode>.
4293 * combine.c (expand_compound_operation): Likewise.
4294 (make_extraction): Likewise.
4295 (change_zero_ext): Likewise.
4296 (simplify_comparison): Likewise.
4297 * cse.c (cse_insn): Likewise.
4298 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4299 (mem_loc_descriptor): Likewise.
4300 (loc_descriptor): Likewise.
4301 * expmed.c (init_expmed_one_mode): Likewise.
4302 (synth_mult): Likewise.
4303 (emit_store_flag_1): Likewise.
4304 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4305 of a comparison with size.
4306 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4307 (reduce_to_bit_field_precision): Likewise.
4308 * function.c (expand_function_end): Likewise.
4309 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4310 * loop-doloop.c (doloop_modify): Likewise.
4311 * optabs.c (expand_binop): Likewise.
4312 (expand_unop): Likewise.
4313 (expand_copysign_absneg): Likewise.
4314 (prepare_cmp_insn): Likewise.
4315 (maybe_legitimize_operand): Likewise.
4316 * recog.c (const_scalar_int_operand): Likewise.
4317 * rtlanal.c (get_address_mode): Likewise.
4318 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4319 (simplify_cond_clz_ctz): Likewise.
4320 * tree-nested.c (get_nl_goto_field): Likewise.
4321 * tree.c (build_vector_type_for_mode): Likewise.
4322 * var-tracking.c (use_narrower_mode): Likewise.
4324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4325 Alan Hayward <alan.hayward@arm.com>
4326 David Sherwood <david.sherwood@arm.com>
4328 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4329 * builtins.c (expand_builtin_signbit): Use it.
4330 * cfgexpand.c (expand_debug_expr): Likewise.
4331 * dojump.c (do_jump): Likewise.
4332 (do_compare_and_jump): Likewise.
4333 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4334 * expmed.c (make_tree): Likewise.
4335 * expr.c (expand_expr_real_2): Likewise.
4336 (expand_expr_real_1): Likewise.
4337 (try_casesi): Likewise.
4338 * fold-const-call.c (fold_const_call_ss): Likewise.
4339 * fold-const.c (unextend): Likewise.
4340 (extract_muldiv_1): Likewise.
4341 (fold_single_bit_test): Likewise.
4342 (native_encode_int): Likewise.
4343 (native_encode_string): Likewise.
4344 (native_interpret_int): Likewise.
4345 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4346 * internal-fn.c (expand_addsub_overflow): Likewise.
4347 (expand_neg_overflow): Likewise.
4348 (expand_mul_overflow): Likewise.
4349 (expand_arith_overflow): Likewise.
4350 * match.pd: Likewise.
4351 * stor-layout.c (layout_type): Likewise.
4352 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4353 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4354 * tree-ssanames.c (get_range_info): Likewise.
4355 * tree-switch-conversion.c (array_value_type) Likewise.
4356 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4357 (vect_recog_divmod_pattern): Likewise.
4358 (vect_recog_mixed_size_cond_pattern): Likewise.
4359 * tree-vrp.c (extract_range_basic): Likewise.
4360 (simplify_float_conversion_using_ranges): Likewise.
4361 * tree.c (int_fits_type_p): Likewise.
4362 * ubsan.c (instrument_bool_enum_load): Likewise.
4363 * varasm.c (mergeable_string_section): Likewise.
4364 (narrowing_initializer_constant_valid_p): Likewise.
4365 (output_constant): Likewise.
4367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4368 Alan Hayward <alan.hayward@arm.com>
4369 David Sherwood <david.sherwood@arm.com>
4371 * machmode.h (NARROWEST_INT_MODE): New macro.
4372 * expr.c (alignment_for_piecewise_move): Use it instead of
4373 GET_CLASS_NARROWEST_MODE (MODE_INT).
4374 (push_block): Likewise.
4375 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4377 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4380 Alan Hayward <alan.hayward@arm.com>
4381 David Sherwood <david.sherwood@arm.com>
4383 * postreload.c (move2add_valid_value_p): Change the type of the
4384 mode parameter to scalar_int_mode.
4385 (move2add_use_add2_insn): Add a mode parameter and use it instead
4387 (move2add_use_add3_insn): Likewise.
4388 (reload_cse_move2add): Update accordingly.
4390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4391 Alan Hayward <alan.hayward@arm.com>
4392 David Sherwood <david.sherwood@arm.com>
4394 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4396 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4397 * optabs.c (expand_unop): Likewise.
4399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4400 Alan Hayward <alan.hayward@arm.com>
4401 David Sherwood <david.sherwood@arm.com>
4403 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4404 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4405 (popcount_loc_descriptor): Likewise.
4406 (bswap_loc_descriptor): Likewise.
4407 (rotate_loc_descriptor): Likewise.
4408 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4409 calling the functions above.
4411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4412 Alan Hayward <alan.hayward@arm.com>
4413 David Sherwood <david.sherwood@arm.com>
4415 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4417 (try_combine): Likewise.
4418 (simplify_if_then_else): Likewise.
4419 * cse.c (cse_insn): Likewise.
4420 * dwarf2out.c (mem_loc_descriptor): Likewise.
4421 * emit-rtl.c (gen_lowpart_common): Likewise.
4422 * simplify-rtx.c (simplify_truncation): Likewise.
4423 (simplify_binary_operation_1): Likewise.
4424 (simplify_const_relational_operation): Likewise.
4425 (simplify_ternary_operation): Likewise.
4426 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4429 Alan Hayward <alan.hayward@arm.com>
4430 David Sherwood <david.sherwood@arm.com>
4432 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4433 * reload.c (push_reload): Likewise.
4434 (find_reloads): Likewise.
4436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4437 Alan Hayward <alan.hayward@arm.com>
4438 David Sherwood <david.sherwood@arm.com>
4440 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4441 (make_compound_operation_int): Likewise.
4442 (change_zero_ext): Likewise.
4443 * expr.c (convert_move): Likewise.
4444 (convert_modes): Likewise.
4445 * fwprop.c (forward_propagate_subreg): Likewise.
4446 * loop-iv.c (get_biv_step_1): Likewise.
4447 * optabs.c (widen_operand): Likewise.
4448 * postreload.c (move2add_valid_value_p): Likewise.
4449 * recog.c (simplify_while_replacing): Likewise.
4450 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4451 (simplify_binary_operation_1): Likewise. Remove redundant
4452 mode equality check.
4454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4455 Alan Hayward <alan.hayward@arm.com>
4456 David Sherwood <david.sherwood@arm.com>
4458 * combine.c (combine_simplify_rtx): Add checks for
4459 is_a <scalar_int_mode>.
4460 (simplify_if_then_else): Likewise.
4461 (make_field_assignment): Likewise.
4462 (simplify_comparison): Likewise.
4463 * ifcvt.c (noce_try_bitop): Likewise.
4464 * loop-invariant.c (canonicalize_address_mult): Likewise.
4465 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4468 Alan Hayward <alan.hayward@arm.com>
4469 David Sherwood <david.sherwood@arm.com>
4471 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4472 is_a <scalar_int_mode> instead of != BLKmode.
4474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4475 Alan Hayward <alan.hayward@arm.com>
4476 David Sherwood <david.sherwood@arm.com>
4478 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4479 instead of != VOIDmode.
4480 * combine.c (if_then_else_cond): Likewise.
4481 (change_zero_ext): Likewise.
4482 * dwarf2out.c (mem_loc_descriptor): Likewise.
4483 (loc_descriptor): Likewise.
4484 * rtlanal.c (canonicalize_condition): Likewise.
4485 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4488 Alan Hayward <alan.hayward@arm.com>
4489 David Sherwood <david.sherwood@arm.com>
4491 * simplify-rtx.c (simplify_binary_operation_1): Use
4492 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4495 Alan Hayward <alan.hayward@arm.com>
4496 David Sherwood <david.sherwood@arm.com>
4498 * wide-int.h (int_traits<unsigned char>) New class.
4499 (int_traits<unsigned short>) Likewise.
4500 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4501 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4503 * combine.c (set_nonzero_bits_and_sign_copies): Use
4504 is_a <scalar_int_mode>.
4505 (find_split_point): Likewise.
4506 (combine_simplify_rtx): Likewise.
4507 (simplify_logical): Likewise.
4508 (expand_compound_operation): Likewise.
4509 (expand_field_assignment): Likewise.
4510 (make_compound_operation): Likewise.
4511 (extended_count): Likewise.
4512 (change_zero_ext): Likewise.
4513 (simplify_comparison): Likewise.
4514 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4515 (ucompare_loc_descriptor): Likewise.
4516 (minmax_loc_descriptor): Likewise.
4517 (mem_loc_descriptor): Likewise.
4518 (loc_descriptor): Likewise.
4519 * expmed.c (init_expmed_one_mode): Likewise.
4520 * lra-constraints.c (lra_constraint_offset): Likewise.
4521 * optabs.c (prepare_libcall_arg): Likewise.
4522 * postreload.c (move2add_note_store): Likewise.
4523 * reload.c (operands_match_p): Likewise.
4524 * rtl.h (load_extend_op): Likewise.
4525 * rtlhooks.c (gen_lowpart_general): Likewise.
4526 * simplify-rtx.c (simplify_truncation): Likewise.
4527 (simplify_unary_operation_1): Likewise.
4528 (simplify_binary_operation_1): Likewise.
4529 (simplify_const_binary_operation): Likewise.
4530 (simplify_const_relational_operation): Likewise.
4531 (simplify_subreg): Likewise.
4532 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4533 * var-tracking.c (adjust_mems): Likewise.
4534 (prepare_call_arguments): Likewise.
4536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4537 Alan Hayward <alan.hayward@arm.com>
4538 David Sherwood <david.sherwood@arm.com>
4540 * machmode.h (is_int_mode): New fuction.
4541 * combine.c (find_split_point): Use it.
4542 (combine_simplify_rtx): Likewise.
4543 (simplify_if_then_else): Likewise.
4544 (simplify_set): Likewise.
4545 (simplify_shift_const_1): Likewise.
4546 (simplify_comparison): Likewise.
4547 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4548 * cse.c (notreg_cost): Likewise.
4549 (cse_insn): Likewise.
4550 * cselib.c (cselib_lookup_1): Likewise.
4551 * dojump.c (do_jump_1): Likewise.
4552 (do_compare_rtx_and_jump): Likewise.
4553 * dse.c (get_call_args): Likewise.
4554 * dwarf2out.c (rtl_for_decl_init): Likewise.
4555 (native_encode_initializer): Likewise.
4556 * expmed.c (emit_store_flag_1): Likewise.
4557 (emit_store_flag): Likewise.
4558 * expr.c (convert_modes): Likewise.
4559 (store_field): Likewise.
4560 (expand_expr_real_1): Likewise.
4561 * fold-const.c (fold_read_from_constant_string): Likewise.
4562 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4563 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4564 * optabs.c (expand_binop): Likewise.
4565 (expand_unop): Likewise.
4566 (expand_abs_nojump): Likewise.
4567 (expand_one_cmpl_abs_nojump): Likewise.
4568 * simplify-rtx.c (mode_signbit_p): Likewise.
4569 (val_signbit_p): Likewise.
4570 (val_signbit_known_set_p): Likewise.
4571 (val_signbit_known_clear_p): Likewise.
4572 (simplify_relational_operation_1): Likewise.
4573 * tree.c (vector_type_mode): Likewise.
4575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4576 Alan Hayward <alan.hayward@arm.com>
4577 David Sherwood <david.sherwood@arm.com>
4579 * machmode.h (smallest_mode_for_size): Fix formatting.
4580 (smallest_int_mode_for_size): New function.
4581 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4582 instead of smallest_mode_for_size.
4583 * combine.c (make_extraction): Likewise.
4584 * config/arc/arc.c (arc_expand_movmem): Likewise.
4585 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4586 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4587 * config/s390/s390.c (s390_expand_insv): Likewise.
4588 * config/sparc/sparc.c (assign_int_registers): Likewise.
4589 * config/spu/spu.c (spu_function_value): Likewise.
4590 (spu_function_arg): Likewise.
4591 * coverage.c (get_gcov_type): Likewise.
4592 (get_gcov_unsigned_t): Likewise.
4593 * dse.c (find_shift_sequence): Likewise.
4594 * expmed.c (store_bit_field_1): Likewise.
4595 * expr.c (convert_move): Likewise.
4596 (store_field): Likewise.
4597 * internal-fn.c (expand_arith_overflow): Likewise.
4598 * optabs-query.c (get_best_extraction_insn): Likewise.
4599 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4600 * stor-layout.c (layout_type): Likewise.
4601 (initialize_sizetypes): Likewise.
4602 * targhooks.c (default_get_mask_mode): Likewise.
4603 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4606 Alan Hayward <alan.hayward@arm.com>
4607 David Sherwood <david.sherwood@arm.com>
4609 * machmode.h (opt_mode::else_blk): New function.
4610 (int_mode_for_mode): Declare.
4611 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4612 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4614 * cfgexpand.c (expand_debug_expr): Likewise.
4615 * combine.c (gen_lowpart_or_truncate): Likewise.
4616 (gen_lowpart_for_combine): Likewise.
4617 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4618 * config/avr/avr.c (avr_to_int_mode): Likewise.
4619 (avr_out_plus_1): Likewise.
4620 (avr_out_plus): Likewise.
4621 (avr_out_round): Likewise.
4622 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4623 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4624 (s390_expand_vcond): Likewise.
4625 * config/spu/spu.c (spu_split_immediate): Likewise.
4626 (spu_expand_mov): Likewise.
4627 * dse.c (get_stored_val): Likewise.
4628 * expmed.c (store_bit_field_1): Likewise.
4629 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4631 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4632 (extract_low_bits): Likewise.
4633 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4634 handling rather than repeating the check.
4635 (emit_group_store): Likewise.
4636 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4637 * optabs.c (expand_absneg_bit): Likewise.
4638 (expand_copysign_absneg): Likewise.
4639 (expand_copysign_bit): Likewise.
4640 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4641 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4642 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4643 * var-tracking.c (prepare_call_arguments): Likewise.
4644 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4645 int_mode_for_mode instead of mode_for_size.
4646 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4649 Alan Hayward <alan.hayward@arm.com>
4650 David Sherwood <david.sherwood@arm.com>
4652 * machmode.h (int_mode_for_size): New function.
4653 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4654 instead of mode_for_size.
4655 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4657 * combine.c (expand_field_assignment): Use int_mode_for_size
4658 instead of mode_for_size.
4659 (make_extraction): Likewise.
4660 (simplify_shift_const_1): Likewise.
4661 (simplify_comparison): Likewise.
4662 * dojump.c (do_jump): Likewise.
4663 * dwarf2out.c (mem_loc_descriptor): Likewise.
4664 * emit-rtl.c (init_derived_machine_modes): Likewise.
4665 * expmed.c (flip_storage_order): Likewise.
4666 (convert_extracted_bit_field): Likewise.
4667 * expr.c (copy_blkmode_from_reg): Likewise.
4668 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4669 * internal-fn.c (expand_mul_overflow): Likewise.
4670 * lower-subreg.c (simple_move): Likewise.
4671 * optabs-libfuncs.c (init_optabs): Likewise.
4672 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4673 * tree.c (vector_type_mode): Likewise.
4674 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4675 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4676 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4677 * tree-vect-stmts.c (vectorizable_load): Likewise.
4678 (vectorizable_store): Likewise.
4680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4681 Alan Hayward <alan.hayward@arm.com>
4682 David Sherwood <david.sherwood@arm.com>
4684 * coretypes.h (pod_mode): New type.
4685 (scalar_int_mode_pod): New typedef.
4686 * machmode.h (pod_mode): New class.
4687 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4688 * genmodes.c (emit_mode_int_n): Update accordingly.
4689 * lower-subreg.h (target_lower_subreg): Change type to
4690 scalar_int_mode_pod.
4691 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4692 scalar_int_mode_pod.
4694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4695 Alan Hayward <alan.hayward@arm.com>
4696 David Sherwood <david.sherwood@arm.com>
4698 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4699 machine_mode to scalar_int_mode.
4700 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4701 (rs6000_option_override_internal): Remove cast to int.
4702 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4703 machine_mode to scalar_int_mode.
4704 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4705 (rs6000_option_override_internal): Remove cast to int.
4706 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4707 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4709 * config/s390/s390.c (s390_expand_builtin): Likewise.
4710 * coretypes.h (scalar_int_mode): New type.
4711 (opt_scalar_int_mode): New typedef.
4712 * machmode.h (scalar_int_mode): New class.
4713 (scalar_int_mode::includes_p): New function.
4714 (byte_mode): Change type to scalar_int_mode.
4715 (word_mode): Likewise.
4716 (ptr_mode): Likewise.
4717 * emit-rtl.c (byte_mode): Likewise.
4718 (word_mode): Likewise.
4719 (ptr_mode): Likewise.
4720 (init_derived_machine_modes): Update accordingly.
4721 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4722 and MODE_PARTIAL_INT.
4723 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4724 opt_scalar_int_mode.
4726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4727 Alan Hayward <alan.hayward@arm.com>
4728 David Sherwood <david.sherwood@arm.com>
4730 * target.def (libgcc_floating_mode_supported_p): Take a
4732 * doc/tm.texi: Regenerate.
4733 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4735 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4736 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4740 Alan Hayward <alan.hayward@arm.com>
4741 David Sherwood <david.sherwood@arm.com>
4743 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4744 * doc/tm.texi: Regenerate.
4745 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4746 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4747 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4748 * targhooks.h (default_floatn_mode): Likewise.
4749 * targhooks.c (default_floatn_mode): Likewise.
4750 * tree.c (build_common_tree_nodes): Update accordingly.
4752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4753 Alan Hayward <alan.hayward@arm.com>
4754 David Sherwood <david.sherwood@arm.com>
4756 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4757 (mode_iterator::iterate_p): Likewise.
4758 (mode_iterator::get_wider): Likewise.
4759 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4762 Alan Hayward <alan.hayward@arm.com>
4763 David Sherwood <david.sherwood@arm.com>
4765 * coretypes.h (opt_scalar_float_mode): New typedef.
4766 * machmode.h (float_mode_for_size): New function.
4767 * emit-rtl.c (double_mode): Delete.
4768 (init_emit_once): Use float_mode_for_size.
4769 * stor-layout.c (layout_type): Likewise.
4770 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4773 Alan Hayward <alan.hayward@arm.com>
4774 David Sherwood <david.sherwood@arm.com>
4776 * output.h (assemble_real): Take a scalar_float_mode.
4777 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4778 * config/arm/arm.md (consttable_4): Likewise.
4779 (consttable_8): Likewise.
4780 (consttable_16): Likewise.
4781 * config/mips/mips.md (consttable_float): Likewise.
4782 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4783 * varasm.c (assemble_real): Take a scalar_float_mode.
4784 (output_constant_pool_2): Update accordingly.
4785 (output_constant): Likewise.
4787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4788 Alan Hayward <alan.hayward@arm.com>
4789 David Sherwood <david.sherwood@arm.com>
4791 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4792 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4793 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4794 (native_encode_real): Likewise.
4795 (native_interpret_real): Likewise.
4796 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4797 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4800 Alan Hayward <alan.hayward@arm.com>
4801 David Sherwood <david.sherwood@arm.com>
4803 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4804 <scalar_float_mode>. Simplify.
4805 (gen_extend_conv_libfunc): Likewise.
4807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4808 Alan Hayward <alan.hayward@arm.com>
4809 David Sherwood <david.sherwood@arm.com>
4811 * coretypes.h (scalar_float_mode): New type.
4812 * machmode.h (mode_traits::from_int): Use machine_mode if
4813 USE_ENUM_MODES is defined.
4814 (is_a): New function.
4816 (dyn_cast): Likewise.
4817 (scalar_float_mode): New class.
4818 (scalar_float_mode::includes_p): New function.
4819 (is_float_mode): Likewise.
4820 * gdbhooks.py (MachineModePrinter): New class.
4821 (build_pretty_printer): Use it for scalar_float_mode.
4822 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4823 (format_helper::format_helper): Turn into a template.
4824 * genmodes.c (get_mode_class): New function.
4825 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4826 or machine_mode if none.
4827 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4828 as_a <scalar_float_mode>.
4829 * dwarf2out.c (mem_loc_descriptor): Likewise.
4830 (insert_float): Likewise.
4831 (add_const_value_attribute): Likewise.
4832 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4833 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4834 (expand_unop): Update accordingly.
4835 (expand_abs_nojump): Likewise.
4836 (expand_copysign_absneg): Take a scalar_float_mode.
4837 (expand_copysign_bit): Likewise.
4838 (expand_copysign): Update accordingly.
4840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4841 Alan Hayward <alan.hayward@arm.com>
4842 David Sherwood <david.sherwood@arm.com>
4844 * coretypes.h (opt_mode): New class.
4845 * machmode.h (opt_mode): Likewise.
4846 (opt_mode::else_void): New function.
4847 (opt_mode::require): Likewise.
4848 (opt_mode::exists): Likewise.
4849 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4850 (GET_MODE_2XWIDER_MODE): Likewise.
4851 (mode_iterator::get_wider): Update accordingly.
4852 (mode_iterator::get_2xwider): Likewise.
4853 (mode_iterator::get_known_wider): Likewise, turning into a template.
4854 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4855 forcing a wider mode to exist.
4856 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4857 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4858 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4859 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4860 * lower-subreg.c (init_lower_subreg): Likewise.
4861 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4862 on the final iteration.
4863 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4864 a wider mode exists before asking for a move pattern.
4865 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4866 forcing a wider mode to exist.
4867 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4868 returning false if no such mode exists.
4869 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4870 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4871 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4872 Avoid checking for a MODE_INT if we already know the mode is not a
4874 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4875 forcing a wider mode to exist.
4876 (expmed_mult_highpart_optab): Likewise.
4877 (expmed_mult_highpart): Likewise.
4878 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4880 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4881 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4882 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4883 * internal-fn.c (expand_mul_overflow): Update use of
4884 GET_MODE_2XWIDER_MODE.
4885 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4886 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4887 GET_MODE_WIDER_MODE.
4888 (convert_plusminus_to_widen): Likewise.
4889 * tree-switch-conversion.c (array_value_type): Likewise.
4890 * var-tracking.c (emit_note_insn_var_location): Likewise.
4891 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4892 Return false inside rather than outside the loop if no wider mode
4894 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4895 and GET_MODE_2XWIDER_MODE
4896 (can_compare_p): Use else_void.
4897 * gdbhooks.py (OptMachineModePrinter): New class.
4898 (build_pretty_printer): Use it for opt_mode.
4900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4901 Alan Hayward <alan.hayward@arm.com>
4902 David Sherwood <david.sherwood@arm.com>
4904 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4905 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4908 Alan Hayward <alan.hayward@arm.com>
4909 David Sherwood <david.sherwood@arm.com>
4911 * machmode.h (mode_traits): New structure.
4912 (get_narrowest_mode): New function.
4913 (mode_iterator::start): Likewise.
4914 (mode_iterator::iterate_p): Likewise.
4915 (mode_iterator::get_wider): Likewise.
4916 (mode_iterator::get_known_wider): Likewise.
4917 (mode_iterator::get_2xwider): Likewise.
4918 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4919 (FOR_EACH_MODE): Likewise.
4920 (FOR_EACH_MODE_FROM): Likewise.
4921 (FOR_EACH_MODE_UNTIL): Likewise.
4922 (FOR_EACH_WIDER_MODE): Likewise.
4923 (FOR_EACH_2XWIDER_MODE): Likewise.
4924 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4925 * combine.c (simplify_comparison): Likewise
4926 * config/i386/i386.c (type_natural_mode): Likewise.
4927 * cse.c (cse_insn): Likewise.
4928 * dse.c (find_shift_sequence): Likewise.
4929 * emit-rtl.c (init_derived_machine_modes): Likewise.
4930 (init_emit_once): Likewise.
4931 * explow.c (hard_function_value): Likewise.
4932 * expmed.c (extract_fixed_bit_field_1): Likewise.
4933 (extract_bit_field_1): Likewise.
4934 (expand_divmod): Likewise.
4935 (emit_store_flag_1): Likewise.
4936 * expr.c (init_expr_target): Likewise.
4937 (convert_move): Likewise.
4938 (alignment_for_piecewise_move): Likewise.
4939 (widest_int_mode_for_size): Likewise.
4940 (emit_block_move_via_movmem): Likewise.
4941 (copy_blkmode_to_reg): Likewise.
4942 (set_storage_via_setmem): Likewise.
4943 (compress_float_constant): Likewise.
4944 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4945 * optabs-query.c (get_best_extraction_insn): Likewise.
4946 * optabs.c (expand_binop): Likewise.
4947 (expand_twoval_unop): Likewise.
4948 (expand_twoval_binop): Likewise.
4949 (widen_leading): Likewise.
4950 (widen_bswap): Likewise.
4951 (expand_parity): Likewise.
4952 (expand_unop): Likewise.
4953 (prepare_cmp_insn): Likewise.
4954 (prepare_float_lib_cmp): Likewise.
4955 (expand_float): Likewise.
4956 (expand_fix): Likewise.
4957 (expand_sfix_optab): Likewise.
4958 * postreload.c (move2add_use_add2_insn): Likewise.
4959 * reg-stack.c (reg_to_stack): Likewise.
4960 * reginfo.c (choose_hard_reg_mode): Likewise.
4961 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4962 * stor-layout.c (mode_for_size): Likewise.
4963 (smallest_mode_for_size): Likewise.
4964 (mode_for_vector): Likewise.
4965 (finish_bitfield_representative): Likewise.
4966 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4967 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4968 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4969 * var-tracking.c (prepare_call_arguments): Likewise.
4971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4972 Alan Hayward <alan.hayward@arm.com>
4973 David Sherwood <david.sherwood@arm.com>
4975 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4976 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4977 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4978 * machmode.h (mode_size): Move earlier in file.
4979 (mode_precision): Likewise.
4980 (mode_inner): Likewise.
4981 (mode_nunits): Likewise.
4982 (mode_unit_size): Likewise.
4983 (unit_unit_precision): Likewise.
4984 (mode_wider): Likewise.
4985 (mode_2xwider): Likewise.
4986 (machine_mode): New class.
4987 (mode_to_bytes): New function.
4988 (mode_to_bits): Likewise.
4989 (mode_to_precision): Likewise.
4990 (mode_to_inner): Likewise.
4991 (mode_to_unit_size): Likewise.
4992 (mode_to_unit_precision): Likewise.
4993 (mode_to_nunits): Likewise.
4994 (GET_MODE_SIZE): Use mode_to_bytes.
4995 (GET_MODE_BITSIZE): Use mode_to_bits.
4996 (GET_MODE_PRECISION): Use mode_to_precision.
4997 (GET_MODE_INNER): Use mode_to_inner.
4998 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4999 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5000 (GET_MODE_NUNITS): Use mode_to_nunits.
5001 * system.h (ALWAYS_INLINE): New macro.
5002 * config/powerpcspe/powerpcspe-c.c
5003 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5004 int for arg1_mode and arg2_mode.
5006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5007 Alan Hayward <alan.hayward@arm.com>
5008 David Sherwood <david.sherwood@arm.com>
5010 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5011 Prefix mode names with E_ in case statements.
5012 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5013 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5014 (aarch64_split_simd_move): Likewise.
5015 (aarch64_gen_storewb_pair): Likewise.
5016 (aarch64_gen_loadwb_pair): Likewise.
5017 (aarch64_gen_store_pair): Likewise.
5018 (aarch64_gen_load_pair): Likewise.
5019 (aarch64_get_condition_code_1): Likewise.
5020 (aarch64_constant_pool_reload_icode): Likewise.
5021 (get_rsqrte_type): Likewise.
5022 (get_rsqrts_type): Likewise.
5023 (get_recpe_type): Likewise.
5024 (get_recps_type): Likewise.
5025 (aarch64_gimplify_va_arg_expr): Likewise.
5026 (aarch64_simd_container_mode): Likewise.
5027 (aarch64_emit_load_exclusive): Likewise.
5028 (aarch64_emit_store_exclusive): Likewise.
5029 (aarch64_expand_compare_and_swap): Likewise.
5030 (aarch64_gen_atomic_cas): Likewise.
5031 (aarch64_emit_bic): Likewise.
5032 (aarch64_emit_atomic_swap): Likewise.
5033 (aarch64_emit_atomic_load_op): Likewise.
5034 (aarch64_evpc_trn): Likewise.
5035 (aarch64_evpc_uzp): Likewise.
5036 (aarch64_evpc_zip): Likewise.
5037 (aarch64_evpc_ext): Likewise.
5038 (aarch64_evpc_rev): Likewise.
5039 (aarch64_evpc_dup): Likewise.
5040 (aarch64_gen_ccmp_first): Likewise.
5041 (aarch64_gen_ccmp_next): Likewise.
5042 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5043 (alpha_emit_xfloating_libcall): Likewise.
5044 (emit_insxl): Likewise.
5045 (alpha_arg_type): Likewise.
5046 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5047 (arc_preferred_simd_mode): Likewise.
5048 (arc_secondary_reload): Likewise.
5049 (get_arc_condition_code): Likewise.
5050 (arc_print_operand): Likewise.
5051 (arc_legitimate_constant_p): Likewise.
5052 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5053 * config/arc/arc.md (casesi_load): Likewise.
5054 (casesi_compact_jump): Likewise.
5055 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5056 (cc_use_register): Likewise.
5057 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5058 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5059 (arm_init_iwmmxt_builtins): Likewise.
5060 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5061 (neon_expand_vector_init): Likewise.
5062 (arm_attr_length_move_neon): Likewise.
5063 (maybe_get_arm_condition_code): Likewise.
5064 (arm_emit_vector_const): Likewise.
5065 (arm_preferred_simd_mode): Likewise.
5066 (arm_output_iwmmxt_tinsr): Likewise.
5067 (thumb1_output_casesi): Likewise.
5068 (thumb2_output_casesi): Likewise.
5069 (arm_emit_load_exclusive): Likewise.
5070 (arm_emit_store_exclusive): Likewise.
5071 (arm_expand_compare_and_swap): Likewise.
5072 (arm_evpc_neon_vuzp): Likewise.
5073 (arm_evpc_neon_vzip): Likewise.
5074 (arm_evpc_neon_vrev): Likewise.
5075 (arm_evpc_neon_vtrn): Likewise.
5076 (arm_evpc_neon_vext): Likewise.
5077 (arm_validize_comparison): Likewise.
5078 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5079 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5080 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5081 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
5082 (c6x_preferred_simd_mode): Likewise.
5083 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
5084 (epiphany_rtx_costs): Likewise.
5085 * config/epiphany/predicates.md (proper_comparison_operator):
5087 * config/frv/frv.c (condexec_memory_operand): Likewise.
5088 (frv_emit_move): Likewise.
5089 (output_move_single): Likewise.
5090 (output_condmove_single): Likewise.
5091 (frv_hard_regno_mode_ok): Likewise.
5092 (frv_matching_accg_mode): Likewise.
5093 * config/h8300/h8300.c (split_adds_subs): Likewise.
5094 (h8300_rtx_costs): Likewise.
5095 (h8300_print_operand): Likewise.
5096 (compute_mov_length): Likewise.
5097 (output_logical_op): Likewise.
5098 (compute_logical_op_length): Likewise.
5099 (compute_logical_op_cc): Likewise.
5100 (h8300_shift_needs_scratch_p): Likewise.
5101 (output_a_shift): Likewise.
5102 (compute_a_shift_length): Likewise.
5103 (compute_a_shift_cc): Likewise.
5104 (expand_a_rotate): Likewise.
5105 (output_a_rotate): Likewise.
5106 * config/i386/i386.c (classify_argument): Likewise.
5107 (function_arg_advance_32): Likewise.
5108 (function_arg_32): Likewise.
5109 (function_arg_64): Likewise.
5110 (function_value_64): Likewise.
5111 (ix86_gimplify_va_arg): Likewise.
5112 (ix86_legitimate_constant_p): Likewise.
5113 (put_condition_code): Likewise.
5114 (split_double_mode): Likewise.
5115 (ix86_avx256_split_vector_move_misalign): Likewise.
5116 (ix86_expand_vector_logical_operator): Likewise.
5117 (ix86_split_idivmod): Likewise.
5118 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
5119 (ix86_build_const_vector): Likewise.
5120 (ix86_build_signbit_mask): Likewise.
5121 (ix86_match_ccmode): Likewise.
5122 (ix86_cc_modes_compatible): Likewise.
5123 (ix86_expand_branch): Likewise.
5124 (ix86_expand_sse_cmp): Likewise.
5125 (ix86_expand_sse_movcc): Likewise.
5126 (ix86_expand_int_sse_cmp): Likewise.
5127 (ix86_expand_vec_perm_vpermi2): Likewise.
5128 (ix86_expand_vec_perm): Likewise.
5129 (ix86_expand_sse_unpack): Likewise.
5130 (ix86_expand_int_addcc): Likewise.
5131 (ix86_split_to_parts): Likewise.
5132 (ix86_vectorize_builtin_gather): Likewise.
5133 (ix86_vectorize_builtin_scatter): Likewise.
5134 (avx_vpermilp_parallel): Likewise.
5135 (inline_memory_move_cost): Likewise.
5136 (ix86_tieable_integer_mode_p): Likewise.
5137 (x86_maybe_negate_const_int): Likewise.
5138 (ix86_expand_vector_init_duplicate): Likewise.
5139 (ix86_expand_vector_init_one_nonzero): Likewise.
5140 (ix86_expand_vector_init_one_var): Likewise.
5141 (ix86_expand_vector_init_concat): Likewise.
5142 (ix86_expand_vector_init_interleave): Likewise.
5143 (ix86_expand_vector_init_general): Likewise.
5144 (ix86_expand_vector_set): Likewise.
5145 (ix86_expand_vector_extract): Likewise.
5146 (emit_reduc_half): Likewise.
5147 (ix86_emit_i387_round): Likewise.
5148 (ix86_mangle_type): Likewise.
5149 (ix86_expand_round_sse4): Likewise.
5150 (expand_vec_perm_blend): Likewise.
5151 (canonicalize_vector_int_perm): Likewise.
5152 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5153 (expand_vec_perm_1): Likewise.
5154 (expand_vec_perm_interleave3): Likewise.
5155 (expand_vec_perm_even_odd_pack): Likewise.
5156 (expand_vec_perm_even_odd_1): Likewise.
5157 (expand_vec_perm_broadcast_1): Likewise.
5158 (ix86_vectorize_vec_perm_const_ok): Likewise.
5159 (ix86_expand_vecop_qihi): Likewise.
5160 (ix86_expand_mul_widen_hilo): Likewise.
5161 (ix86_expand_sse2_abs): Likewise.
5162 (ix86_expand_pextr): Likewise.
5163 (ix86_expand_pinsr): Likewise.
5164 (ix86_preferred_simd_mode): Likewise.
5165 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5166 * config/i386/sse.md (*andnot<mode>3): Likewise.
5167 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5168 (*<code><mode>3): Likewise.
5169 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5170 (ia64_expand_atomic_op): Likewise.
5171 (ia64_arg_type): Likewise.
5172 (ia64_mode_to_int): Likewise.
5173 (ia64_scalar_mode_supported_p): Likewise.
5174 (ia64_vector_mode_supported_p): Likewise.
5175 (expand_vec_perm_broadcast): Likewise.
5176 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5177 (iq2000_function_arg_advance): Likewise.
5178 (iq2000_function_arg): Likewise.
5179 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5180 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5181 (m68k_libcall_value): Likewise.
5182 (m68k_function_value): Likewise.
5183 (sched_attr_op_type): Likewise.
5184 * config/mcore/mcore.c (mcore_output_move): Likewise.
5185 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5187 (microblaze_function_arg): Likewise.
5188 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5189 (mips_print_operand): Likewise.
5190 (mips_mode_ok_for_mov_fmt_p): Likewise.
5191 (mips_vector_mode_supported_p): Likewise.
5192 (mips_preferred_simd_mode): Likewise.
5193 (mips_expand_vpc_loongson_even_odd): Likewise.
5194 (mips_expand_vec_unpack): Likewise.
5195 (mips_expand_vi_broadcast): Likewise.
5196 (mips_expand_vector_init): Likewise.
5197 (mips_expand_vec_reduc): Likewise.
5198 (mips_expand_msa_cmp): Likewise.
5199 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5200 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5201 (cc_flags_for_mode): Likewise.
5202 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5203 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5204 (nds32_output_casesi_pc_relative): Likewise.
5205 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5206 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5207 (nvptx_gen_unpack): Likewise.
5208 (nvptx_gen_pack): Likewise.
5209 (nvptx_gen_shuffle): Likewise.
5210 (nvptx_gen_wcast): Likewise.
5211 (nvptx_preferred_simd_mode): Likewise.
5212 * config/pa/pa.c (pa_secondary_reload): Likewise.
5213 * config/pa/predicates.md (base14_operand): Likewise.
5214 * config/powerpcspe/powerpcspe-c.c
5215 (altivec_resolve_overloaded_builtin): Likewise.
5216 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5218 (rs6000_preferred_simd_mode): Likewise.
5219 (output_vec_const_move): Likewise.
5220 (rs6000_expand_vector_extract): Likewise.
5221 (rs6000_split_vec_extract_var): Likewise.
5222 (reg_offset_addressing_ok_p): Likewise.
5223 (rs6000_legitimate_offset_address_p): Likewise.
5224 (rs6000_legitimize_address): Likewise.
5225 (rs6000_emit_set_const): Likewise.
5226 (rs6000_const_vec): Likewise.
5227 (rs6000_emit_move): Likewise.
5228 (spe_build_register_parallel): Likewise.
5229 (rs6000_darwin64_record_arg_recurse): Likewise.
5230 (swap_selector_for_mode): Likewise.
5231 (spe_init_builtins): Likewise.
5232 (paired_init_builtins): Likewise.
5233 (altivec_init_builtins): Likewise.
5234 (do_load_for_compare): Likewise.
5235 (rs6000_generate_compare): Likewise.
5236 (rs6000_expand_float128_convert): Likewise.
5237 (emit_load_locked): Likewise.
5238 (emit_store_conditional): Likewise.
5239 (rs6000_output_function_epilogue): Likewise.
5240 (rs6000_handle_altivec_attribute): Likewise.
5241 (rs6000_function_value): Likewise.
5242 (emit_fusion_gpr_load): Likewise.
5243 (emit_fusion_p9_load): Likewise.
5244 (emit_fusion_p9_store): Likewise.
5245 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5246 (fusion_gpr_mem_load): Likewise.
5247 (fusion_addis_mem_combo_load): Likewise.
5248 (fusion_addis_mem_combo_store): Likewise.
5249 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5250 (fusion_gpr_mem_load): Likewise.
5251 (fusion_addis_mem_combo_load): Likewise.
5252 (fusion_addis_mem_combo_store): Likewise.
5253 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5255 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5256 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5257 (rs6000_preferred_simd_mode): Likewise.
5258 (output_vec_const_move): Likewise.
5259 (rs6000_expand_vector_extract): Likewise.
5260 (rs6000_split_vec_extract_var): Likewise.
5261 (reg_offset_addressing_ok_p): Likewise.
5262 (rs6000_legitimate_offset_address_p): Likewise.
5263 (rs6000_legitimize_address): Likewise.
5264 (rs6000_emit_set_const): Likewise.
5265 (rs6000_const_vec): Likewise.
5266 (rs6000_emit_move): Likewise.
5267 (rs6000_darwin64_record_arg_recurse): Likewise.
5268 (swap_selector_for_mode): Likewise.
5269 (paired_init_builtins): Likewise.
5270 (altivec_init_builtins): Likewise.
5271 (rs6000_expand_float128_convert): Likewise.
5272 (emit_load_locked): Likewise.
5273 (emit_store_conditional): Likewise.
5274 (rs6000_output_function_epilogue): Likewise.
5275 (rs6000_handle_altivec_attribute): Likewise.
5276 (rs6000_function_value): Likewise.
5277 (emit_fusion_gpr_load): Likewise.
5278 (emit_fusion_p9_load): Likewise.
5279 (emit_fusion_p9_store): Likewise.
5280 * config/rx/rx.c (rx_gen_move_template): Likewise.
5281 (flags_from_mode): Likewise.
5282 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5283 (s390_slb_comparison): Likewise.
5284 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5285 (s390_vector_mode_supported_p): Likewise.
5286 (s390_cc_modes_compatible): Likewise.
5287 (s390_match_ccmode_set): Likewise.
5288 (s390_canonicalize_comparison): Likewise.
5289 (s390_emit_compare_and_swap): Likewise.
5290 (s390_branch_condition_mask): Likewise.
5291 (s390_rtx_costs): Likewise.
5292 (s390_secondary_reload): Likewise.
5293 (__SECONDARY_RELOAD_CASE): Likewise.
5294 (s390_expand_cs): Likewise.
5295 (s390_preferred_simd_mode): Likewise.
5296 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5297 * config/sh/sh.c (sh_print_operand): Likewise.
5298 (dump_table): Likewise.
5299 (sh_secondary_reload): Likewise.
5300 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5301 * config/sh/sh.md (casesi_worker_1): Likewise.
5302 (casesi_worker_2): Likewise.
5303 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5304 (fcc_comparison_operator): Likewise.
5305 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5306 (emit_soft_tfmode_cvt): Likewise.
5307 (sparc_preferred_simd_mode): Likewise.
5308 (output_cbranch): Likewise.
5309 (sparc_print_operand): Likewise.
5310 (sparc_expand_vec_perm_bmask): Likewise.
5311 (vector_init_bshuffle): Likewise.
5312 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5313 (spu_vector_mode_supported_p): Likewise.
5314 (spu_expand_insv): Likewise.
5315 (spu_emit_branch_or_set): Likewise.
5316 (spu_handle_vector_attribute): Likewise.
5317 (spu_builtin_splats): Likewise.
5318 (spu_builtin_extract): Likewise.
5319 (spu_builtin_promote): Likewise.
5320 (spu_expand_sign_extend): Likewise.
5321 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5322 (tilegx_simd_int): Likewise.
5323 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5324 (tilepro_simd_int): Likewise.
5325 * config/v850/v850.c (const_double_split): Likewise.
5326 (v850_print_operand): Likewise.
5327 (ep_memory_offset): Likewise.
5328 * config/vax/vax.c (vax_rtx_costs): Likewise.
5329 (vax_output_int_move): Likewise.
5330 (vax_output_int_add): Likewise.
5331 (vax_output_int_subtract): Likewise.
5332 * config/visium/predicates.md (visium_branch_operator): Likewise.
5333 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5334 (visium_print_operand_address): Likewise.
5335 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5336 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5337 (xtensa_expand_conditional_branch): Likewise.
5338 (xtensa_copy_incoming_a7): Likewise.
5339 (xtensa_output_literal): Likewise.
5340 * dfp.c (decimal_real_maxval): Likewise.
5341 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5344 Alan Hayward <alan.hayward@arm.com>
5345 David Sherwood <david.sherwood@arm.com>
5347 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5348 (mode_nunits_inline): Likewise.
5349 (mode_inner_inline): Likewise.
5350 (mode_unit_size_inline): Likewise.
5351 (mode_unit_precision_inline): Likewise.
5352 (emit_insn_modes_h): Likewise. Also emit a #define of the
5354 (emit_mode_wider): Add an E_ prefix to mode names.
5355 (emit_mode_complex): Likewise.
5356 (emit_mode_inner): Likewise.
5357 (emit_mode_adjustments): Likewise.
5358 (emit_mode_int_n): Likewise.
5359 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5360 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5361 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5362 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5363 (CRC32_BUILTIN, ENTRY): Likewise.
5364 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5365 (aarch64_pop_regs): Likewise.
5366 (aarch64_process_components): Likewise.
5367 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5368 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5369 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5370 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5371 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5372 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5374 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5375 (pcmpistr): Likewise.
5376 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5377 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5378 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5380 * config/rl78/rl78.c (mduc_regs): Likewise.
5381 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5382 (htm_expand_builtin): Likewise.
5383 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5384 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5385 * config/xtensa/xtensa.c (print_operand): Likewise.
5386 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5387 (NUM_MODE_VECTOR_INT): Likewise.
5388 * genoutput.c (null_operand): Likewise.
5389 (output_operand_data): Likewise.
5390 * genrecog.c (print_parameter_value): Likewise.
5391 * lra.c (debug_operand_data): Likewise.
5393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5395 * dwarf2out.c (scompare_loc_descriptor_wide)
5396 (scompare_loc_descriptor_narrow): New functions, split out from...
5397 (scompare_loc_descriptor): ...here.
5398 * expmed.c (emit_store_flag_int): New function, split out from...
5399 (emit_store_flag): ...here.
5401 2017-08-30 Richard Biener <rguenther@suse.de>
5403 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5404 (dwarf2out_early_finish): Move setting of AT_pubnames from
5405 early debug output to early finish.
5407 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5409 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5410 and -mdata-region to the assembler.
5412 2017-08-30 Richard Biener <rguenther@suse.de>
5414 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5416 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5417 (dwarf2out_early_global_decl): Only generate a DIE for the
5418 abstract origin if it doesn't already exist or is a declaration DIE.
5419 (resolve_addr): Do not add the linkage name twice when
5420 generating a stub DIE for the DW_TAG_GNU_call_site target.
5422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5424 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5425 Use machine_mode rather than int for arg1_mode.
5427 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5430 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5431 that the second argument of the built-in functions to unpack
5432 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5433 switch statement instead a lot of if statements.
5434 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5435 Allow 64-bit values to be in Altivec registers as well as
5436 traditional floating point registers.
5437 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5439 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5441 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5442 MAX_REGS_PER_ADDRESS == 1.
5444 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5446 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5447 * config/i386/i386.c (ix86_option_override_internal): Simplify
5448 setting of opts->x_flag_entry.
5450 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5451 Jakub Jelinek <jakub@redhat.com>
5452 Richard Biener <rguenther@suse.de>
5454 PR tree-optimization/81503
5455 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5456 folded constant fits in the target type; reorder tests for clarity.
5458 2017-08-29 Martin Liska <mliska@suse.cz>
5460 * passes.def: Include pass_lower_switch.
5461 * stmt.c (dump_case_nodes): Remove and move to
5462 tree-switch-conversion.
5463 (case_values_threshold): Likewise.
5464 (expand_switch_as_decision_tree_p): Likewise.
5465 (emit_case_decision_tree): Likewise.
5466 (expand_case): Likewise.
5467 (balance_case_nodes): Likewise.
5468 (node_has_low_bound): Likewise.
5469 (node_has_high_bound): Likewise.
5470 (node_is_bounded): Likewise.
5471 (emit_case_nodes): Likewise.
5472 (struct simple_case_node): New struct.
5473 (add_case_node): Remove.
5474 (emit_case_dispatch_table): Use vector instead of case_list.
5475 (reset_out_edges_aux): Remove.
5476 (compute_cases_per_edge): Likewise.
5477 (expand_case): Build list of simple_case_node.
5478 (expand_sjlj_dispatch_table): Use it.
5479 * tree-switch-conversion.c (struct case_node): Moved from
5480 stmt.c and adjusted.
5481 (emit_case_nodes): Likewise.
5482 (node_has_low_bound): Likewise.
5483 (node_has_high_bound): Likewise.
5484 (node_is_bounded): Likewise.
5485 (case_values_threshold): Likewise.
5486 (reset_out_edges_aux): Likewise.
5487 (compute_cases_per_edge): Likewise.
5488 (add_case_node): Likewise.
5489 (dump_case_nodes): Likewise.
5490 (balance_case_nodes): Likewise.
5491 (expand_switch_as_decision_tree_p): Likewise.
5492 (emit_jump): Likewise.
5493 (emit_case_decision_tree): Likewise.
5494 (try_switch_expansion): Likewise.
5495 (do_jump_if_equal): Likewise.
5496 (emit_cmp_and_jump_insns): Likewise.
5497 (fix_phi_operands_for_edge): New function.
5498 (record_phi_operand_mapping): Likewise.
5499 (class pass_lower_switch): New pass.
5500 (pass_lower_switch::execute): New function.
5501 (make_pass_lower_switch): Likewise.
5502 (conditional_probability):
5503 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5504 * tree-pass.h: Add make_pass_lower_switch.
5506 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5509 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5512 2017-08-29 Richard Biener <rguenther@suse.de>
5514 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5515 we do not add a DW_AT_inline attribute twice.
5516 (gen_subprogram_die): Remove code setting DW_AT_inline on
5517 DECL_ABSTRACT_P nodes.
5519 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5521 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5522 calls to internal functions.
5523 (gimplify_modify_expr): Likewise.
5524 * tree-call-cdce.c (use_internal_fn): Likewise.
5525 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5526 (convert_to_divmod): Set the nothrow flag.
5527 * tree-if-conv.c (predicate_mem_writes): Likewise.
5528 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5529 (vectorizable_call): Likewise.
5530 (vectorizable_store): Likewise.
5531 (vectorizable_load): Likewise.
5532 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5533 (vect_recog_mask_conversion_pattern): Likewise.
5535 2017-08-29 Martin Liska <mliska@suse.cz>
5538 * gcc.c (driver_handle_option): Add new argument.
5539 * opts-common.c (handle_option): Pass
5540 target_option_override_hook.
5541 * opts-global.c (lang_handle_option): Add new option.
5542 (set_default_handlers): Add new argument.
5543 (decode_options): Likewise.
5544 * opts.c (target_handle_option): Likewise.
5545 (common_handle_option): Call target_option_override_hook.
5546 * opts.h (struct cl_option_handler_func): Add hook for
5547 target option override.
5548 (struct cl_option_handlers): Likewise.
5549 (set_default_handlers): Add new argument.
5550 (decode_options): Likewise.
5551 (common_handle_option): Likewise.
5552 (target_handle_option): Likewise.
5553 * toplev.c (toplev::main): Pass targetm.target_option.override
5556 2017-08-29 Richard Biener <rguenther@suse.de>
5557 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5559 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5560 life to the active subtree.
5562 2017-08-28 Jeff Law <law@redhat.com>
5564 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5565 derive_equivalences.
5566 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5568 (edge_info::derive_equivalences): New private member function
5570 * tree-ssa-dom.c (class edge_info): Changed from a struct
5571 to a class. Add ctor/dtor, methods and data members.
5572 (edge_info::edge_info): Renamed from allocate_edge_info.
5573 Initialize additional members.
5574 (edge_info::~edge_info): New.
5575 (free_dom_edge_info): Delete the edge info.
5576 (record_edge_info): Use new class & associated member functions.
5577 Tighten forms for testing for edge equivalences.
5578 (record_temporary_equivalences): Iterate over the simple
5579 equivalences rather than assuming there's only one per edge.
5580 (cprop_into_successor_phis): Iterate over the simple
5581 equivalences rather than assuming there's only one per edge.
5582 (optimize_stmt): Use operand_equal_p rather than pointer
5583 equality for mini-DSE code.
5585 2017-08-28 Nathan Sidwell <nathan@acm.org>
5587 * gcc.c (execute): Fold SIGPIPE handling into switch
5588 statement. Adjust internal error message.
5590 2017-08-28 Richard Biener <rguenther@suse.de>
5593 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5594 Do nothing for removed DIEs.
5596 2017-08-28 Richard Biener <rguenther@suse.de>
5598 PR tree-optimization/81977
5599 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5602 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5605 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5606 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5607 using targetm.gen_mem_thread_fence.
5609 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5612 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5613 predicate to register_operand. Reorder operands.
5614 (*btr<mode>): Ditto.
5615 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5616 (*btr<mode>_mask): Ditto.
5618 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5620 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5621 * config/rs6000/xmmintrin.h: New file.
5622 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5624 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5627 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5628 parameter and_insn and return it.
5629 (recombine_lvx_pattern): Insert a copy to ensure availability of
5630 the base register of the copied masking operation at the point of
5631 the instruction replacement.
5632 (recombine_stvx_pattern): Likewise.
5634 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5636 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5637 undocumented switches.
5638 (-mpower9-dform-vector): Likewise.
5639 (-mpower9-dform): Likewise.
5640 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5641 comments to delete references to -mpower9-dform* switches.
5642 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5643 Delete reference to -mpower9-dform* switches, test for
5644 -mpower9-vector instead.
5645 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5646 (OTHER_P9_VECTOR_MASKS): Likewise.
5647 (POWERPC_MASKS): Likewise.
5648 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5649 tests against -mpower9-dform* to -mpower9-vector. Delete code
5650 that checked for -mpower9-dform* consistancy with other options.
5651 Add test for -mpower9-misc to enable other power9 switches.
5652 (rs6000_init_hard_regno_mode_ok): Likewise.
5653 (rs6000_option_override_internal): Likewise.
5654 (rs6000_emit_prologue): Likewise.
5655 (rs6000_emit_epilogue): Likewise.
5656 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5657 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5658 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5660 (emit_fusion_p9_store): Likewise.
5661 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5662 resetting these macros if the assembler does not support ISA 3.0
5664 (TARGET_P9_DFORM_VECTOR): Likewise.
5665 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5666 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5668 2017-08-25 Alan Modra <amodra@gmail.com>
5670 PR rtl-optimization/81747
5671 * cse.c (cse_extended_basic_block): Don't attempt to record
5672 equivalences for degenerate conditional branches that branch
5673 to their fall-through.
5675 2017-08-24 Martin Sebor <msebor@redhat.com>
5678 * gimple-fold.c (size_must_be_zero_p): New function.
5679 (gimple_fold_builtin_memory_op): Call it.
5681 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5683 * config/rs6000/mm_malloc.h: New file.
5685 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5687 PR tree-optimization/81913
5688 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5689 analysis when either IVs in condition can wrap.
5691 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5693 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5694 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5696 2017-08-24 Richard Biener <rguenther@suse.de>
5699 * targhooks.c (default_target_can_inline_p): Properly
5700 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5701 is present and always compare.
5702 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5703 infer -mfpmath=sse from TARGET_SSE_P.
5704 (ix86_can_inline_p): Properly use target_option_default_node when
5705 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5707 2017-08-24 Richard Biener <rguenther@suse.de>
5710 * dwarf2out.c (output_die): Handle flag_generate_offload like
5712 (output_comp_unit): Likewise.
5713 (gen_array_type_die): Likewise.
5714 (dwarf2out_early_finish): Likewise.
5715 (note_variable_value_in_expr): Likewise.
5716 (dwarf2out_finish): Likewise. Adjust assert.
5717 * cgraphunit.c (symbol_table::compile): Move setting of
5718 flag_generate_offload earlier ...
5719 (symbol_table::finalize_compilation_unit): ... here, before
5720 early debug finalization.
5722 2017-08-24 Richard Biener <rguenther@suse.de>
5724 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5725 and ipa-fnsummary.h.
5726 (ix86_can_inline_p): When ix86_fpmath flags do not match
5727 check whether the callee uses FP math at all.
5729 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5732 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5733 instead of TYPE_PRECISION.
5735 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5736 Alan Hayward <alan.hayward@arm.com>
5737 David Sherwood <david.sherwood@arm.com>
5739 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5741 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5743 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5745 (ix86_frame::stack_realign_allocate): New field.
5746 (struct machine_frame_state): Modify comments.
5747 (machine_frame_state::sp_realigned_fp_end): New field.
5748 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5750 (sp_valid_at): Add assertion to assure no attempt to access invalid
5751 offset of a realigned stack.
5752 (fp_valid_at): Likewise.
5753 (choose_baseaddr): Modify comments.
5754 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5755 ix86_expand_prologue.
5756 (ix86_expand_prologue): Modify stack realignment and allocation.
5757 (ix86_expand_epilogue): Modify comments.
5758 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5759 avx2_runtime, avx512f, and avx512f_runtime.
5761 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5763 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5764 (mstackrealign): Do not init to -1.
5765 * config/i386/i386.c (ix86_option_override_internal):
5766 Check opts_set, not opts when setting default value of
5767 opts->x_ix86_force_align_arg_pointer.
5769 2017-08-23 Richard Biener <rguenther@suse.de>
5771 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5772 lang_hooks.decl_printable_name.
5773 * print-rtl-function.c (print_rtx_function): Likewise.
5774 * tree-pretty-print.c (dump_function_header): Likewise.
5776 2017-08-23 Richard Biener <rguenther@suse.de>
5779 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5780 -g0 at compile-time.
5782 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5785 * doc/sourcebuild.texi (Other hardware attributes):
5788 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5791 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5792 Fix single-use check.
5794 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5796 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5798 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5800 * doc/install.texi: Modify to add more details on running selected
5803 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5805 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5806 is combined with -mabi=ms.
5807 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5810 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5812 PR tree-optimization/81488
5813 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5814 and cached_basis fields.
5815 (MAX_SPREAD): New constant.
5816 (alloc_cand_and_find_basis): Initialize new fields.
5817 (clear_visited): New function.
5818 (create_phi_basis_1): Rename from create_phi_basis, set visited
5819 and cached_basis fields.
5820 (create_phi_basis): New wrapper function.
5821 (phi_add_costs_1): Rename from phi_add_costs, add spread
5822 parameter, set visited field, short-circuit when limits reached.
5823 (phi_add_costs): New wrapper function.
5824 (record_phi_increments_1): Rename from record_phi_increments, set
5826 (record_phi_increments): New wrapper function.
5827 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5828 (phi_incr_cost): New wrapper function.
5829 (all_phi_incrs_profitable_1): Rename from
5830 all_phi_incrs_profitable, set visited field.
5831 (all_phi_incrs_profitable): New wrapper function.
5833 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5834 Alan Hayward <alan.hayward@arm.com>
5835 David Sherwood <david.sherwood@arm.com>
5837 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5838 that takes the outer and inner modes.
5839 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5840 comparison as the canonical test for a paradoxical subreg.
5841 * combine.c (simplify_set): Use paradoxical_subreg_p.
5842 (make_extraction): Likewise.
5843 (force_to_mode): Likewise.
5844 (rtx_equal_for_field_assignment_p): Likewise.
5845 (gen_lowpart_for_combine): Likewise.
5846 (simplify_comparison): Likewise.
5847 * cse.c (equiv_constant): Likewise.
5848 * expmed.c (store_bit_field_1): Likewise.
5849 * final.c (alter_subreg): Likewise.
5850 * fwprop.c (propagate_rtx): Likewise.
5851 (forward_propagate_subreg): Likewise.
5852 * ira-conflicts.c (ira_build_conflicts): Likewise.
5853 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5854 * lra-constraints.c (curr_insn_transform): Likewise.
5855 (split_reg): Likewise.
5856 * lra-eliminations.c (move_plus_up): Likewise.
5857 (lra_eliminate_regs_1): Likewise.
5858 * recog.c (general_operand): Likewise.
5859 * ree.c (combine_reaching_defs): Likewise.
5860 * reload.c (push_reload): Likewise.
5861 (find_reloads): Likewise.
5862 * reload1.c (elimination_effects): Likewise.
5863 (compute_reload_subreg_offset): Likewise.
5864 (choose_reload_regs): Likewise.
5865 * rtlanal.c (subreg_lsb_1): Likewise.
5866 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5867 (simplify_subreg): Likewise.
5868 * var-tracking.c (track_loc_p): Likewise.
5869 * emit-rtl.c (byte_lowpart_offset): Likewise.
5870 (paradoxical_subreg_p): Delete out-of-line definition.
5872 2017-08-22 Jeff Law <law@redhat.com>
5874 PR tree-optimization/81741
5875 PR tree-optimization/71947
5876 * tree-ssa-dom.c: Include tree-inline.h.
5877 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5878 equivalences if one is more expensive to compute than the other.
5879 * tree-ssa-scopedtables.h (class const_or_copies): Make
5880 record_const_or_copy_raw method private.
5881 (class avail_exprs_stack): New method simplify_binary_operation.
5882 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5883 avail_exprs_stack::simplify_binary_operation as needed.
5884 (avail_exprs_stack::simplify_binary_operation): New function.
5886 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5888 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5889 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5890 (DOT_SYMBOLS): Likewise.
5891 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5892 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5893 (RS6000_ABI_NAME): Likewise.
5894 (TARGET_CMODEL): Likewise.
5895 (TOC_SECTION_ASM_OP): Likewise.
5896 (SET_CMODEL): New macro.
5897 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5899 2017-08-22 Richard Biener <rguenther@suse.de>
5901 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5902 to allow for free-lang-data replacements similar to verify_type_variant.
5904 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5905 Alan Hayward <alan.hayward@arm.com>
5906 David Sherwood <david.sherwood@arm.com>
5908 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5909 VOIDmode for the LABEL_REF.
5911 2017-08-22 Richard Biener <rguenther@suse.de>
5913 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5915 2017-08-22 Richard Biener <rguenther@suse.de>
5917 * common.opt (feliminate-dwarf2-dups): Ignore.
5918 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5919 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5920 same_die_p_wrap, compute_section_prefix,
5921 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5922 (comdat_symbol_id, comdat_symbol_number): Likewise.
5923 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5925 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5926 (output_die): Mark unreachable path unreachable.
5927 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5928 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5929 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5930 (dwarf2out_early_finish): Likewise.
5932 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5934 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5936 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5939 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5940 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5941 (avr_attribute_table) <io, io_low, address>: Initialize
5942 .decl_required with true.
5944 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5946 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5947 undocumented debugging options.
5948 (-mvsx-scalar-double): Likewise.
5949 (-mallow-df-permute): Likewise.
5950 (-mvectorize-builtins): Likewise.
5951 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5952 (rs6000_builtin_vectorized_function): Likewise.
5953 (rs6000_builtin_md_vectorized_function): Likewise.
5954 (rs6000_opt_vars): Likewise.
5956 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5959 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5960 (*btrq_imm): Rename from *btrq.
5961 (*btcq_imm): Rename from *btcq.
5962 (btsc): New code attribute.
5963 (*<btsc><mode>): New insn pattern.
5964 (*btr<mode>): Ditto.
5965 (*<btsc><mode>_mask): New insn_and_split pattern.
5966 (*btr<mode>_mask): Ditto.
5968 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5969 Alan Hayward <alan.hayward@arm.com>
5970 David Sherwood <david.sherwood@arm.com>
5972 * function.c (pad_below): Simplify padding calculation.
5974 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5975 Alan Hayward <alan.hayward@arm.com>
5976 David Sherwood <david.sherwood@arm.com>
5978 * target.def (function_prologue): Remove frame size argument.
5979 (function_epilogue): Likewise.
5980 * doc/tm.texi: Regenerate.
5981 * final.c (final_start_function): Update call to function_prologue.
5982 (final_end_function): Update call to function_epilogue.
5983 (default_function_pro_epilogue): Remove frame size argument.
5984 * output.h (default_function_pro_epilogue): Likewise.
5985 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5986 (arm_output_function_prologue): Likewise.
5987 * config/frv/frv.c (frv_function_prologue): Likewise.
5988 (frv_function_epilogue): Likewise.
5989 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5990 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5991 (ia64_output_function_epilogue): Likewise.
5992 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5993 (m32r_output_function_epilogue): Likewise.
5994 * config/microblaze/microblaze.c (microblaze_function_prologue)
5995 (microblaze_function_epilogue): Likewise.
5996 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5997 (mips_output_function_epilogue): Likewise.
5998 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5999 (mmix_target_asm_function_epilogue): Likewise.
6000 * config/msp430/msp430.c (msp430_start_function): Likewise.
6001 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6002 (nds32_asm_function_epilogue): Likewise.
6003 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6004 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6005 (pa_output_function_epilogue): Likewise.
6006 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6007 (rs6000_output_function_epilogue): Likewise.
6008 * config/rl78/rl78.c (rl78_start_function): Likewise.
6009 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6010 (rs6000_output_function_epilogue): Likewise.
6011 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6012 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6013 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6014 (sparc_asm_function_epilogue): Likewise.
6016 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6018 * tree.h (type_has_mode_precision_p): New function.
6019 * convert.c (convert_to_integer_1): Use it.
6020 * expr.c (expand_expr_real_2): Likewise.
6021 (expand_expr_real_1): Likewise.
6022 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6023 * match.pd: Likewise.
6024 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6025 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6026 * tree-tailcall.c (process_assignment): Likewise.
6027 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6028 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6029 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6030 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6031 (vectorizable_assignment): Likewise.
6032 (vectorizable_shift): Likewise.
6033 (vectorizable_operation): Likewise.
6034 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6036 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6038 * match.pd: Add pow (C, x) simplification.
6040 2017-08-21 Richard Biener <rguenther@suse.de>
6042 PR tree-optimization/81900
6043 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6044 for blocks with abnormal predecessors.
6045 (compute_antic): Do not set visited flag prematurely.
6047 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6050 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6052 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6054 * stor-layout.h (vector_type_mode): Move to...
6055 * tree.h (vector_type_mode): ...here.
6056 * stor-layout.c (vector_type_mode): Move to...
6057 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6059 2017-08-21 Richard Biener <rguenther@suse.de>
6061 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6062 register_external_die hooks.
6063 (debug_false_tree_charstarstar_uhwistar): Declare.
6064 (debug_nothing_tree_charstar_uhwi): Likewise.
6065 * debug.c (do_nothing_debug_hooks): Adjust.
6066 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6067 (debug_nothing_tree_charstar_uhwi): Likewise.
6068 * dbxout.c (dbx_debug_hooks): Adjust.
6069 (xcoff_debug_hooks): Likewise.
6070 * sdbout.c (sdb_debug_hooks): Likewise.
6071 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6072 * dwarf2out.c (macinfo_label_base): New global.
6073 (dwarf2out_register_external_die): New function for the
6074 register_external_die hook.
6075 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6076 (dwarf2_debug_hooks): Use them.
6077 (dwarf2_lineno_debug_hooks): Adjust.
6078 (struct die_struct): Add with_offset flag.
6079 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6080 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6081 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
6082 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
6083 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
6084 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
6085 defining section names for the early LTO debug variants.
6086 (reset_indirect_string): New helper.
6087 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
6088 (print_dw_val): Add support for offsetted symbol references.
6089 (get_ultimate_context): Split out from is_cxx.
6090 (is_cxx): Use get_ultimate_context.
6091 (is_fortran): Add decl overload.
6092 (compute_comp_unit_symbol): Split out worker from
6093 compute_section_prefix.
6094 (compute_section_prefix): Call compute_comp_unit_symbol and
6095 set comdat_type_p here.
6096 (output_die): Skip DIE symbol output for the LTO added one.
6097 Handle DIE symbol references with offset.
6098 (output_comp_unit): Guard section name mangling properly.
6099 For LTO debug sections emit a symbol at the section beginning
6100 which we use to refer to its DIEs.
6101 (add_abstract_origin_attribute): For DIEs registered via
6102 dwarf2out_register_external_die directly refer to the early
6103 DIE rather than indirectly through the shadow one we created.
6104 Remove obsolete call to dwarf2out_abstract_function for
6105 non-function/block origins.
6106 (gen_array_type_die): When generating early LTO debug do
6107 not emit DW_AT_string_length.
6108 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
6109 late when in LTO. As suggested place a gcc_unreachable for
6110 the DECL_ABSTRACT_P case.
6111 (gen_subprogram_die): Avoid another specification DIE
6112 for early built declarations/definitions for the late LTO case.
6113 (gen_variable_die): Add type references for late duplicated VLA dies
6115 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
6116 we have the abstract instance already.
6117 (process_scope_var): Adjust decl DIE contexts in LTO which
6118 first puts them in limbo.
6119 (gen_decl_die): Do not generate type DIEs late apart from
6120 types for VLAs or for decls we do not yet have a DIE. Do not
6121 call dwarf2out_abstract_function late.
6122 (dwarf2out_early_global_decl): Make sure to create DIEs
6123 for abstract instances of a decl first.
6124 (dwarf2out_late_global_decl): Adjust comment.
6125 (output_macinfo_op): With multiple macro sections use
6126 macinfo_label_base to distinguish labels.
6127 (output_macinfo): Likewise. Update macinfo_label_base.
6128 Pass in the line info label.
6129 (note_variable_value_in_expr): When generating LTO resolve
6130 all variable values here by generating DIEs as needed.
6131 (init_sections_and_labels): Add early LTO debug flag parameter
6132 and generate different sections and names if set. Add generation
6133 counter for the labels so we can have multiple of them.
6134 (reset_dies): Helper to allow DIEs to be output multiple times.
6135 (dwarf2out_finish): When outputting DIEs to the fat part of an
6136 LTO object first reset DIEs.
6137 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6138 (modified_type_die): Check for decl_ultimate_origin being self
6140 (gen_type_die_with_usage): Likewise.
6141 (gen_typedef_die): Allow decl_ultimate_origin being self.
6142 (set_decl_abstract_flags): Remove.
6143 (set_block_abstract_flags): Likewise.
6144 (dwarf2out_abstract_function): Treat the early generated DIEs
6145 as the abstract copy and only add DW_AT_inline and
6146 DW_AT_artificial here and call set_decl_origin_self.
6147 If the DIE has an abstract origin don't do anything.
6148 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6149 if we have none yet (Go fails to build one, PR78628).
6150 (variably_modified_type_p): Prevent endless recursion for Ada
6151 cyclic pointer types.
6152 * lto-streamer-in.c: Include debug.h.
6153 (dref_queue): New global.
6154 (lto_read_tree_1): Stream in DIE references.
6155 (lto_input_tree): Register DIE references.
6156 (input_function): Stream DECL_DEBUG_ARGS.
6157 * lto-streamer-out.c: Include debug.h.
6158 (lto_write_tree_1): Output DIE references.
6159 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6160 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6161 (output_function): Stream DECL_DEBUG_ARGS.
6162 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6163 Stream DECL_ABSTRACT_ORIGIN.
6164 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6165 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6166 DECL_CONTEXT for file-scope decls.
6167 * lto-streamer.h (struct dref_entry): Declare.
6168 (dref_queue): Likewise.
6169 * cfgexpand.c (pass_expand::execute): Do not call the
6170 outlining_inline_function hook here.
6171 * lto-wrapper.c (debug_obj): New global.
6172 (tool_cleanup): Unlink it if required.
6173 (debug_objcopy): New function.
6174 (run_gcc): Handle early debug sections in the IL files by
6175 extracting them to separate files, partially linkin them and
6176 feeding the result back as result to the linker.
6177 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6178 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6179 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6180 sections into a separate segment.
6181 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6183 (darwin_asm_dwarf_section): Likewise.
6184 (darwin_asm_output_dwarf_offset): Likewise.
6185 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6187 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6188 Alan Hayward <alan.hayward@arm.com>
6189 David Sherwood <david.sherwood@arm.com>
6191 * read-md.h (md_reader::record_potential_iterator_use): Replace
6192 pointer argument with an rtx and an index.
6193 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6194 (apply_mode_iterator): Likewise.
6195 (apply_code_iterator): Likewise.
6196 (apply_int_iterator): Likewise.
6197 (apply_subst_iterator): Likewise.
6198 (record_iterator_use): Likewise.
6199 (record_attribute_use): Likewise.
6200 (md_reader::record_potential_iterator_use): Likewise. Update calls
6201 to record_iterator_use and apply_iterator.
6202 (iterator_use): Replace ptr with x and index.
6203 (attribute_use): Likewise.
6204 (apply_attribute_uses): Update calls to apply_iterator.
6205 (apply_iterators): Likewise. Update initialization of iterator_use.
6206 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6207 and record_potential_iterator_use.
6208 (rtx_reader::read_rtx_operand): Likewise.
6210 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6211 Alan Hayward <alan.hayward@arm.com>
6212 David Sherwood <david.sherwood@arm.com>
6214 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6217 2017-08-21 Richard Biener <rguenther@suse.de>
6220 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6221 at struct end conservatively when comparing common bases.
6223 2017-08-21 Richard Biener <rguenther@suse.de>
6225 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6226 (mem_ref_in_stmt): Remove.
6227 (determine_max_movement): Use ref index to get at the reference.
6228 (invariantness_dom_walker::before_dom_children): Deal with
6229 lim data already initialized.
6230 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6232 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6234 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6235 (TARGET_ISA_ROUND): Ditto.
6236 (TARGET_ROUND): Ditto.
6237 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6238 * config/i386/i386.md: Ditto.
6239 * config/i386/sse.md: Ditto.
6240 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6241 with OPTION_MASK_ISA_SSE4_1.
6243 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6246 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6247 __builtin_ia32_lzcnt_u16.
6249 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6252 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6253 (rs6000_set_current_function): Rewrite function to use it.
6255 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6258 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6259 and TYPE_WARN_IF_NOT_ALIGN.
6260 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6261 (handle_warn_if_not_align): New.
6262 (place_union_field): Call handle_warn_if_not_align.
6263 (place_field): Call handle_warn_if_not_align.
6264 Copy TYPE_WARN_IF_NOT_ALIGN.
6265 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6266 (layout_type): Likewise.
6267 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6269 (tree_decl_common): Add warn_if_not_align.
6270 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6271 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6272 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6273 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6274 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6275 * doc/extend.texi: Document warn_if_not_aligned attribute.
6276 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6278 2017-08-17 Martin Liska <mliska@suse.cz>
6281 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6282 (get_data_dependence): Use it as pointer type.
6283 (distribute_loop): Likewise.
6285 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6287 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6288 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6289 (p8_vmrgow_<mode>_direct): New define_insn.
6290 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6291 handle endianness for vmrgew and vmrgow permute patterns.
6293 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6295 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6296 * config/rs6000/rs6000-cpus.def: Remove comment.
6297 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6298 (POWERPC_MASKS): Likewise.
6299 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6300 use of TARGET_VSX_TIMODE.
6301 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6302 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6303 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6304 (rs6000_option_override_internal): Remove dead code.
6305 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6306 (rs6000_legitimize_reload_address): Likewise.
6307 (rs6000_legitimate_address_p): Likewise.
6308 (rs6000_opt_masks): Delete "vsx-timode".
6309 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6310 from function comment.
6311 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6312 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6313 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6315 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6316 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6317 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6320 (define_peephole2): Likewise.
6322 2017-08-17 Martin Sebor <msebor@redhat.com>
6325 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6326 past the end of an array.
6327 (test_pp_format): Add test cases.
6329 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6331 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6332 missing ECF_NOTHROW flags.
6334 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6337 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6338 operands residing in integer registers.
6339 (*vsx_le_perm_load_<mode>): Likewise.
6340 (*vsx_le_perm_store_<mode>): Likewise.
6341 (define_peephole2): Add peepholes to optimize the above.
6343 2017-08-17 Marek Polacek <polacek@redhat.com>
6346 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6347 to mimic what shorten_compare did. Change the return type to bool.
6348 (fold_cond_expr_with_comparison): Update call to
6349 operand_equal_for_comparison_p.
6350 (fold_ternary_loc): Likewise.
6352 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6354 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6356 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6358 2017-08-17 Richard Biener <rguenther@suse.de>
6360 * tree-ssa-structalias.c (solve_graph): When propagating
6361 to successors update the graphs succ edges and avoid duplicate work.
6363 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6366 * config/i386/i386.c (ix86_option_override_internal): Save target
6367 specific options after ix86_stack_protector_guard_reg was changed.
6369 2017-08-17 Richard Biener <rguenther@suse.de>
6371 PR tree-optimization/81827
6372 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6373 (new_var_info): Initialize it conservatively.
6374 (get_call_vi): Mark register vars.
6375 (new_scalar_tmp_constraint_exp): Likewise.
6376 (handle_rhs_call): Likewise.
6377 (handle_const_call): Likewise.
6378 (create_function_info_for): Likewise.
6379 (solve_constraints): Sort varinfos to separate register from
6380 non-register vars to pack points-to solution bitmaps during
6383 2017-08-17 Marek Polacek <polacek@redhat.com>
6385 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6387 2017-08-17 Richard Biener <rguenther@suse.de>
6389 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6390 to true when overflow is undefined and we saturated the result.
6392 2017-08-17 Alan Modra <amodra@gmail.com>
6395 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6396 Don't use store multiple if only one reg needs saving.
6397 (interesting_frame_related_regno): New function.
6398 (rs6000_frame_related): Don't emit frame info for regs that
6400 (rs6000_emit_epilogue): Likewise.
6402 2017-08-16 Nathan Sidwell <nathan@acm.org>
6404 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6405 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6406 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6407 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6408 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6409 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6410 (verify_type): Adjust for TYPE_BINFO move.
6411 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6412 process TYPE_BINFO directly.
6413 (hash_tree): Likewise.
6414 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6416 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6419 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6421 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6423 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6426 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6427 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6428 Add ix86_binary_operator_ok to insn constraint.
6429 (*iordi_1_bts): Ditto.
6430 (*xordi_1_btc): Ditto.
6431 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6432 Update corresponding peephole2 pattern.
6436 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6438 PR tree-optimization/81832
6439 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6440 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6442 2017-08-16 Marek Polacek <polacek@redhat.com>
6445 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6446 regarding size_zero_node.
6448 2017-08-16 Martin Liska <mliska@suse.cz>
6451 * config.gcc: Respect previously set extra_objs in case
6454 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6456 PR tree-optimization/81835
6457 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6458 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6459 not depend on the phi.
6461 2017-08-16 Alan Modra <amodra@gmail.com>
6463 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6466 2017-08-16 Alan Modra <amodra@gmail.com>
6468 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6469 (save_reg_p): ..into this. Update all callers.
6470 (first_reg_to_save): Simplify.
6472 2017-08-16 Alan Modra <amodra@gmail.com>
6474 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6477 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6481 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6482 constant count if that count is less than 32.
6484 2017-08-15 Nathan Sidwell <nathan@acm.org>
6486 * gcc.c (execute): Emit friendlier message if inferior is killed
6487 by an external cause.
6489 2017-08-15 Richard Biener <rguenther@suse.de>
6491 PR tree-optimization/81790
6492 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6493 CONSTRUCTORs from simplifying and VN.
6495 2017-08-14 Martin Sebor <msebor@redhat.com>
6497 * builtin-attrs.def: Add comments.
6499 2017-08-14 Martin Sebor <msebor@redhat.com>
6502 * doc/extend.texi (attribute nonstring): Document new attribute.
6504 2017-08-14 Martin Sebor <msebor@redhat.com>
6507 * tree-diagnostic.c (default_tree_printer): Handle %G.
6508 * gimple-pretty-print.h (percent_G_format): Declare new function.
6509 * gimple-pretty-print.c (percent_G_format): Define.
6510 * tree-pretty-print.c (percent_K_format): Add argument.
6512 2017-08-14 Martin Sebor <msebor@redhat.com>
6514 PR translation/79998
6515 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6516 Remove a stray space.
6518 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6521 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6522 (*iordi_1_bts): Ditto.
6523 (*xordi_1_btc): Ditto.
6525 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6528 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6529 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6531 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6532 quoted strings, and make more translator-friendly.
6533 (darwin_rs6000_override_options): Likewise.
6534 (rs6000_option_override_internal): Likewise.
6535 (rs6000_return_in_memory): Fix overlong line.
6536 (init_cmulative_args): Use quoted strings, and make more
6537 translator-friendly.
6538 (rs6000_pass_by_reference): Fix overlong line.
6539 (def_builtin): Use quoted strings.
6540 (altivec_expand_predicate_builtin): Use quoted strings, and make
6541 more translator-friendly.
6542 (htm_expand_builtin): Use quoted strings.
6543 (cpu_expand_builtin): Use quoted strings, and make more
6544 translator-friendly.
6545 (altivec_expand_builtin): Likewise.
6546 (paired_expand_predicate_builtin): Likewise.
6547 (rs6000_invalid_builtin): Likewise.
6548 (builtin_function_type): Use quoted strings.
6549 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6550 more translator-friendly.
6551 (rs6000_trampoline_init): Likewise.
6552 (rs6000_handle_altivec_attribute): Likewise.
6553 (rs6000_inner_target_options): Use quoted strings.
6554 (rs6000_disable_incompatible_switches): Likewise.
6555 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6556 strings, and make more translator-friendly.
6557 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6559 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6561 PR tree-optimization/81799
6562 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6563 cond_expr to simple gimple operand.
6565 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6568 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6570 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6574 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6575 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6576 TARGET_GASISR_PROLOGUES.
6577 * config/avr/avr.c (avr_option_override): Same.
6578 (avr_pass_pre_proep::execute): Same.
6580 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6583 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6584 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6586 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6588 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6589 define_insn_and_split. Split to a memory load from 0 in
6590 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6591 using PTR mode iterator.
6592 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6593 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6594 (*add_tp_<mode>): Redefine as define_insn_and_split.
6595 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6596 address space. Merge with *add_tp_x32 using PTR mode iterator.
6597 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6598 Split to an add with a memory load from 0 in
6599 DEFAULT_TLS_SEG_REG address space.
6601 2017-08-12 Andrew Pinski <apinski@cavium.com>
6603 * config/aarch64/aarch64-option-extensions.def (rdma):
6604 Fix feature string to what Linux prints out in /proc/cpuinfo.
6606 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6609 * dwarf2out.c (modified_type_die): For C typedef types that have
6610 an ultimate origin, process the ultimate origin instead of the
6612 (gen_typedef_die): Assert that input DECLs have no ultimate
6614 (gen_type_die_with_usage): For typedef variants that have an
6615 ultimate origin, just call gen_decl_die on the original DECL.
6616 (process_scope_var): Avoid creating DIEs for local typedefs and
6617 concrete static variables.
6619 2017-08-12 Alan Modra <amodra@gmail.com>
6623 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6624 match gnu-user.h startfile.
6625 (ENDFILE_LINUX_SPEC): Similarly.
6627 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6630 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6632 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6634 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6635 * config/aarch64/aarch64.md (mov<mode>): Change.
6636 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6637 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6638 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6640 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6642 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6643 for storage order barriers.
6645 2017-08-11 Martin Liska <mliska@suse.cz>
6648 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6649 variables of void type.
6651 2017-08-11 Martin Liska <mliska@suse.cz>
6653 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6654 TARGET_SUPPORTS_ALIASES.
6655 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6656 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6657 (optimize_weakref): Likewise.
6658 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6659 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6660 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6662 2017-08-11 Martin Liska <mliska@suse.cz>
6665 * config/i386/i386.c (make_resolver_func): Do complete
6666 refactoring of the function.
6668 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6671 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6672 * config/i386/i386.c (ix86_stack_protect_guard): Use
6673 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6674 * doc/invoke.texi (x86 Options): Document
6675 -mstack-protector-guard-symbol= option.
6677 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6679 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6680 * config/i386/i386.c (ix86_split_stack_guard): New function.
6681 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6682 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6683 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6684 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6685 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6686 (split_stack_space_check): Call ix86_split_stack_guard.
6688 2017-08-10 Martin Sebor <msebor@redhat.com>
6690 * print-tree.c (print_node): Print location using the established
6692 Replace spaces with colons.
6693 (debug_raw, debug): Ditto.
6695 2017-08-10 Martin Sebor <msebor@redhat.com>
6698 * pretty-print.c (pp_format): Correct the handling of %s precision.
6700 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6703 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6705 (ix86_finalize_stack_frame_flags): This. Also clear
6706 frame_pointer_needed if -fno-omit-frame-pointer is used without
6708 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6709 with ix86_finalize_stack_frame_flags.
6710 (ix86_expand_epilogue): Likewise.
6711 (ix86_expand_split_stack_prologue): Likewise.
6712 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6714 2017-08-10 Martin Liska <mliska@suse.cz>
6717 * c-attribs.c (handle_target_attribute):
6718 Report warning for an empty string argument of target attribute.
6720 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6723 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6725 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6726 or DECL_NONLOCAL labels.
6727 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6728 or DECL_NONLOCAL labels here.
6730 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6732 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6733 to indicate when early gimple folding has been disabled.
6734 (rs6000_gimple_fold_builtin): Add debug content.
6735 (rs6000_invalid_builtin): Fix whitespace.
6736 (rs6000_expand_builtin): Fix whitespace.
6737 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6739 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6742 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6743 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6745 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6747 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6748 (qdf24xx): Likewise.
6749 * config/aarch64/aarch64.md: Include falkor.md.
6750 * config/aarch64/falkor.md: New.
6752 2017-08-09 Marek Polacek <polacek@redhat.com>
6755 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6756 * diagnostic.c (emit_diagnostic): Add a comment.
6757 (emit_diagnostic_valist): New function.
6759 2017-08-09 Marek Polacek <polacek@redhat.com>
6762 * input.c (make_location): New overload.
6763 * input.h (make_location): Declare.
6765 2017-08-08 Alan Modra <amodra@gmail.com>
6766 H.J. Lu <hongjiu.lu@intel.com>
6769 * gcc.c (NO_PIE_SPEC): Delete.
6770 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6772 (LINK_PIE_SPEC): ..to here.
6773 (LINK_COMMAND_SPEC): Support -no-pie.
6774 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6775 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6777 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6778 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6779 (ENDFILE_CRTEND_SPEC): Similarly.
6781 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6784 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6785 (mstack-protector-guard-offset=): Ditto.
6786 * config/i386/i386.c (ix86_option_override): Handle
6787 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6789 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6790 ix86_stack_protect_guard_offset variables.
6791 (TARGET_STACK_PROTECT_GUARD): Always define.
6792 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6793 and -mstack-protector-guard-offset= options.
6795 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6797 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6798 boundary case for the last candidate.
6800 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6802 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6803 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6805 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6808 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6809 * config/aarch64/aarch64-builtins.c
6810 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6811 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6812 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6814 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6815 Andrew Pinski <pinskia@gmail.com>
6818 * internal-fn.def (XORSIGN): New.
6819 * optabs.def (xorsign_optab): New.
6820 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6821 (convert_expand_mult_copysign): New.
6822 (pass_optimize_widening_mul::execute): Call
6823 convert_expand_mult_copysign.
6825 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6827 PR tree-optimization/81354
6828 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6829 Insert on edges rather than explicitly creating landing pads.
6830 (analyze_candidates_and_replace): Commit edge inserts.
6832 2017-08-08 Richard Biener <rguenther@suse.de>
6835 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6836 (expand_simple_operations): Also look through ADDR_EXPRs with
6837 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6839 2017-08-08 Richard Biener <rguenther@suse.de>
6841 PR tree-optimization/81723
6842 * tree-vect-slp.c (struct bst_traits): New hash traits.
6843 (bst_fail): New global.
6844 (vect_build_slp_tree_2): New worker, split out from ...
6845 (vect_build_slp_tree): ... this now wrapping it with using
6846 bst_fail set to cache SLP tree build fails. Properly handle
6848 (vect_analyze_slp_instance): Allocate and free bst_fail.
6850 2017-08-08 Martin Liska <mliska@suse.cz>
6853 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6854 LABEL_DECLs that can be from a different function.
6856 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6858 PR tree-optimization/81744
6859 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6860 loop's number of iterations.
6862 2017-08-08 Martin Liska <mliska@suse.cz>
6864 * asan.c: Include header files.
6865 * attribs.c (build_decl_attribute_variant): New function moved
6867 (build_type_attribute_qual_variant): Likewise.
6868 (cmp_attrib_identifiers): Likewise.
6869 (simple_cst_list_equal): Likewise.
6870 (omp_declare_simd_clauses_equal): Likewise.
6871 (attribute_value_equal): Likewise.
6872 (comp_type_attributes): Likewise.
6873 (build_type_attribute_variant): Likewise.
6874 (lookup_ident_attribute): Likewise.
6875 (remove_attribute): Likewise.
6876 (merge_attributes): Likewise.
6877 (merge_type_attributes): Likewise.
6878 (merge_decl_attributes): Likewise.
6879 (merge_dllimport_decl_attributes): Likewise.
6880 (handle_dll_attribute): Likewise.
6881 (attribute_list_equal): Likewise.
6882 (attribute_list_contained): Likewise.
6883 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6884 (lookup_attribute_by_prefix): Likewise.
6885 * bb-reorder.c: Include header files.
6886 * builtins.c: Likewise.
6887 * calls.c: Likewise.
6888 * cfgexpand.c: Likewise.
6889 * cgraph.c: Likewise.
6890 * cgraphunit.c: Likewise.
6891 * convert.c: Likewise.
6892 * dwarf2out.c: Likewise.
6893 * final.c: Likewise.
6894 * fold-const.c: Likewise.
6895 * function.c: Likewise.
6896 * gimple-expr.c: Likewise.
6897 * gimple-fold.c: Likewise.
6898 * gimple-pretty-print.c: Likewise.
6899 * gimple.c: Likewise.
6900 * gimplify.c: Likewise.
6901 * hsa-common.c: Likewise.
6902 * hsa-gen.c: Likewise.
6903 * internal-fn.c: Likewise.
6904 * ipa-chkp.c: Likewise.
6905 * ipa-cp.c: Likewise.
6906 * ipa-devirt.c: Likewise.
6907 * ipa-fnsummary.c: Likewise.
6908 * ipa-inline.c: Likewise.
6909 * ipa-visibility.c: Likewise.
6911 * lto-cgraph.c: Likewise.
6912 * omp-expand.c: Likewise.
6913 * omp-general.c: Likewise.
6914 * omp-low.c: Likewise.
6915 * omp-offload.c: Likewise.
6916 * omp-simd-clone.c: Likewise.
6917 * opts-global.c: Likewise.
6918 * passes.c: Likewise.
6919 * predict.c: Likewise.
6920 * sancov.c: Likewise.
6921 * sanopt.c: Likewise.
6922 * symtab.c: Likewise.
6923 * toplev.c: Likewise.
6924 * trans-mem.c: Likewise.
6925 * tree-chkp.c: Likewise.
6926 * tree-eh.c: Likewise.
6927 * tree-into-ssa.c: Likewise.
6928 * tree-object-size.c: Likewise.
6929 * tree-parloops.c: Likewise.
6930 * tree-profile.c: Likewise.
6931 * tree-ssa-ccp.c: Likewise.
6932 * tree-ssa-live.c: Likewise.
6933 * tree-ssa-loop.c: Likewise.
6934 * tree-ssa-sccvn.c: Likewise.
6935 * tree-ssa-structalias.c: Likewise.
6936 * tree-ssa.c: Likewise.
6937 * tree-streamer-in.c: Likewise.
6938 * tree-vectorizer.c: Likewise.
6939 * tree-vrp.c: Likewise.
6941 * ubsan.c: Likewise.
6942 * varasm.c: Likewise.
6943 * varpool.c: Likewise.
6944 * tree.c: Remove functions moved to attribs.[ch].
6946 * config/aarch64/aarch64.c: Add attrs.h header file.
6947 * config/alpha/alpha.c: Likewise.
6948 * config/arc/arc.c: Likewise.
6949 * config/arm/arm.c: Likewise.
6950 * config/avr/avr.c: Likewise.
6951 * config/bfin/bfin.c: Likewise.
6952 * config/c6x/c6x.c: Likewise.
6953 * config/cr16/cr16.c: Likewise.
6954 * config/cris/cris.c: Likewise.
6955 * config/darwin.c: Likewise.
6956 * config/epiphany/epiphany.c: Likewise.
6957 * config/fr30/fr30.c: Likewise.
6958 * config/frv/frv.c: Likewise.
6959 * config/ft32/ft32.c: Likewise.
6960 * config/h8300/h8300.c: Likewise.
6961 * config/i386/winnt.c: Likewise.
6962 * config/ia64/ia64.c: Likewise.
6963 * config/iq2000/iq2000.c: Likewise.
6964 * config/lm32/lm32.c: Likewise.
6965 * config/m32c/m32c.c: Likewise.
6966 * config/m32r/m32r.c: Likewise.
6967 * config/m68k/m68k.c: Likewise.
6968 * config/mcore/mcore.c: Likewise.
6969 * config/microblaze/microblaze.c: Likewise.
6970 * config/mips/mips.c: Likewise.
6971 * config/mmix/mmix.c: Likewise.
6972 * config/mn10300/mn10300.c: Likewise.
6973 * config/moxie/moxie.c: Likewise.
6974 * config/msp430/msp430.c: Likewise.
6975 * config/nds32/nds32-isr.c: Likewise.
6976 * config/nds32/nds32.c: Likewise.
6977 * config/nios2/nios2.c: Likewise.
6978 * config/nvptx/nvptx.c: Likewise.
6979 * config/pa/pa.c: Likewise.
6980 * config/pdp11/pdp11.c: Likewise.
6981 * config/powerpcspe/powerpcspe.c: Likewise.
6982 * config/riscv/riscv.c: Likewise.
6983 * config/rl78/rl78.c: Likewise.
6984 * config/rx/rx.c: Likewise.
6985 * config/s390/s390.c: Likewise.
6986 * config/sh/sh.c: Likewise.
6987 * config/sol2.c: Likewise.
6988 * config/sparc/sparc.c: Likewise.
6989 * config/spu/spu.c: Likewise.
6990 * config/stormy16/stormy16.c: Likewise.
6991 * config/tilegx/tilegx.c: Likewise.
6992 * config/tilepro/tilepro.c: Likewise.
6993 * config/v850/v850.c: Likewise.
6994 * config/vax/vax.c: Likewise.
6995 * config/visium/visium.c: Likewise.
6996 * config/xtensa/xtensa.c: Likewise.
6998 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7001 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7002 constraints since the -mupper-regs-* switches have been
7004 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7005 into a vector from a double word element that was extracted from
7006 another vector, and eliminate extra XXPERMDI instructions.
7007 (vsx_concat_<mode>_2): Likewise.
7008 (vsx_concat_<mode>_3): Likewise.
7009 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7010 concat to allow optimizing inserts from previous extracts.
7012 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7014 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7015 memory reference to a SSP offset in TLS address space.
7016 (ix86_print_operand) <case '@'>: Remove.
7017 (ix86_print_operand_punct_valid_p): Remove '@' code.
7018 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7020 (stack_tls_protect_set_<mode>): Remove.
7021 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7022 (stack_tls_protect_test_<mode>): Remove.
7023 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7025 2017-08-07 Olivier Hainque <hainque@adacore.com>
7028 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7030 2017-08-07 Douglas Rupp <rupp@adacore.com>
7032 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7033 variable was referenced as multidir in command.
7035 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7038 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7041 2017-08-07 Martin Liska <mliska@suse.cz>
7043 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7044 * config/rl78/rl78.c: Add include of attribs.h.
7045 * config/sh/sh.c: Likewise.
7046 * config/v850/v850.c: Likewise.
7048 2017-08-07 Tom de Vries <tom@codesourcery.com>
7051 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7053 2017-08-07 Martin Liska <mliska@suse.cz>
7055 * config/mips/mips.c: Include attribs.h.
7057 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7060 * doc/invoke.texi: Document change in behvaior for -Ofast for
7063 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7065 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7067 (aarch64_pop_regs): Likewise.
7068 (aarch64_gen_load_pair): Likewise.
7069 (aarch64_save_callee_saves): Likewise.
7070 (aarch64_restore_callee_saves): Likewise.
7072 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7074 * config/i386/i386.c: Revert the last change.
7076 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7079 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7081 (ix86_finalize_stack_frame_flags): This. Also clear
7082 frame_pointer_needed if -fno-omit-frame-pointer is used without
7084 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7085 with ix86_finalize_stack_frame_flags.
7086 (ix86_expand_epilogue): Likewise.
7087 (ix86_expand_split_stack_prologue): Likewise.
7089 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7092 * config/i386/i386.c (get_builtin_code_for_version): Set priority
7093 to P_AES for Westmere.
7095 2017-08-07 Jonathan Yong <10walls@gmail.com>
7097 * config/i386/mingw.opt (fset-stack-executable): Removed.
7098 * config/i386/cygming.opt (fset-stack-executable): Moved
7100 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7102 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7104 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7106 2017-08-07 Marek Polacek <polacek@redhat.com>
7109 * fold-const.c (fold_indirect_ref_1): Check type_domain.
7111 2017-08-07 Martin Liska <mliska@suse.cz>
7113 * attribs.h (canonicalize_attr_name): New function.
7114 (cmp_attribs): Move from c-format.c and adjusted.
7115 (is_attribute_p): Moved from tree.h.
7116 * tree-inline.c: Add new includes.
7117 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
7118 (private_is_attribute_p): Remove.
7119 (private_lookup_attribute): Likewise.
7120 (private_lookup_attribute_by_prefix): Simplify.
7121 (remove_attribute): Use is_attribute_p.
7122 * tree.h: Remove removed declarations.
7124 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7127 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
7128 instead of computing it in the function. Formatting fix.
7129 (expand_case): Don't rely on default_edge being the first edge,
7130 clear it if removing it, pass default_edge to
7131 emit_case_dispatch_table.
7132 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7135 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7137 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7138 insn in the function, emit NOP after the insn.
7140 2017-08-06 Tom de Vries <tom@codesourcery.com>
7142 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7145 2017-08-06 Tom de Vries <tom@codesourcery.com>
7147 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7150 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7152 PR tree-optimization/57371
7153 * match.pd: New pattern.
7155 2017-08-04 Marek Polacek <polacek@redhat.com>
7158 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7159 perform the computation in offset_int.
7161 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7163 PR tree-optimization/81136
7164 * tree-vectorizer.h: Include tree-hash-traits.h.
7165 (vec_base_alignments): New typedef.
7166 (vec_info): Add a base_alignments field.
7167 (vect_record_base_alignments): Declare.
7168 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7170 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7171 (create_data_ref): Add an is_conditional_in_stmt argument.
7172 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7173 the is_conditional_in_stmt field.
7174 (data_ref_loc): Add an is_conditional_in_stmt field.
7175 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7176 (find_data_references_in_stmt): Update call to create_data_ref.
7177 (graphite_find_data_references_in_stmt): Likewise.
7178 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7179 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7180 (vect_record_base_alignment): New function.
7181 (vect_record_base_alignments): Likewise.
7182 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7183 for nested statements even if we fail to compute a misalignment.
7184 Use pooled base alignments for unconditional references.
7185 (vect_find_same_alignment_drs): Compare base addresses instead
7187 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7188 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7190 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7192 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7193 Add an explicit name for the enum. Use auto_vec for slp_instances
7195 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7197 (_bb_vec_info): Add a constructor and destructor.
7198 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7199 (destroy_loop_vec_info): Delete.
7200 (vect_destroy_datarefs): Likewise.
7201 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7202 (vec_info::vec_info): New function.
7203 (vec_info::~vec_info): Likewise.
7204 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7205 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7206 destroy_loop_vec_info.
7207 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7208 (_loop_vec_info::_loop_vec_info): ...this.
7209 (destroy_loop_vec_info): Replace with...
7210 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7211 the stmt_vec_infos. Leave handling of vec_info information to its
7212 destructor. Remove explicit vector releases.
7213 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7214 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7215 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7216 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7217 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7218 (destroy_bb_vec_info): Replace with...
7219 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7220 information to its destructor.
7221 (vect_slp_analyze_bb_1): Use new and delete instead of
7222 new_bb_vec_info and destroy_bb_vec_info.
7223 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7226 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7228 * tree-data-ref.h (subscript): Add access_fn field.
7229 (data_dependence_relation): Add could_be_independent_p.
7230 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7231 (same_access_functions): Move to tree-data-ref.c.
7232 * tree-data-ref.c (ref_contains_union_access_p): New function.
7233 (access_fn_component_p): Likewise.
7234 (access_fn_components_comparable_p): Likewise.
7235 (dr_analyze_indices): Add a reference to access_fn_component_p.
7236 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7238 (constant_access_functions): Likewise.
7239 (add_other_self_distances): Likewise.
7240 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7241 (initialize_data_dependence_relation): Use XCNEW and remove
7242 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7243 of access functions that have the same type. Allow the
7244 subsequence to end with different bases in some circumstances.
7245 Record the chosen access functions in SUB_ACCESS_FN.
7246 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7247 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7248 (subscript_dependence_tester_1): Likewise dra and drb.
7249 (build_classic_dist_vector): Update calls accordingly.
7250 (subscript_dependence_tester): Likewise.
7251 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7252 DDR_COULD_BE_INDEPENDENT_P.
7253 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7254 comp_alias_ddrs instead of may_alias_ddrs.
7255 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7257 (vect_analyze_data_ref_dependence): Use it if
7258 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7259 distance vectors if that fails.
7260 (dependence_distance_ge_vf): New function.
7261 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7262 LOOP_VINFO_MAY_ALIAS_DDRS.
7264 2017-08-04 Richard Biener <rguenther@suse.de>
7267 * fold-const.c (fold_binary_loc): Properly restrict
7268 minus_var0 && minus_var1 case when associating undefined overflow
7271 2017-08-04 Tom de Vries <tom@codesourcery.com>
7273 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7275 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7277 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7278 Don't start diagnostic messages with a capital letter.
7279 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7281 (rs6000_invalid_builtin): Likewise.
7282 (rs6000_trampoline_init): Likewise.
7284 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7287 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7288 after setting changeable df flags.
7290 2017-08-03 Richard Biener <rguenther@suse.de>
7292 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7293 up if the use is in USE - X.
7295 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7297 * toplev.c (dumpfile.h): New include.
7298 (internal_error_reentered): New static function. Use it...
7299 (internal_error_function): ...here to handle reentered internal_error.
7301 2017-08-03 Richard Biener <rguenther@suse.de>
7304 * fold-const.c (split_tree): Add minus_var and minus_con
7305 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7306 here but always use minus_*.
7307 (associate_trees): Assert we never associate with MINUS_EXPR
7308 and NULL first operand. Do not recurse for PLUS_EXPR operands
7309 when associating as MINUS_EXPR either.
7310 (fold_binary_loc): Track minus_var and minus_con.
7312 2017-08-03 Tom de Vries <tom@codesourcery.com>
7315 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7316 ACCEL_COMPILER, apply finish_options on
7317 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7319 2017-08-03 Tom de Vries <tom@codesourcery.com>
7322 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7323 function_entry_patch_area_size > 0.
7325 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7328 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7329 instead of 10??LU, perform unit multiplication in wide_int,
7330 don't change alloc_object_size_limit if the limit is larger
7333 PR tree-optimization/81655
7334 PR tree-optimization/81588
7335 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7336 the case when ranges[i].low and high are 1 for unsigned type with
7340 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7341 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7343 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7345 * tree-vrp.h: Add include guard.
7347 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7350 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7351 (ud2): New insn pattern.
7352 * config/i386/i386.c (ix86_expand_epilogue):
7353 For naked functions, generate ud2 instead of trap insn.
7355 2017-08-02 Marek Polacek <polacek@redhat.com>
7358 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7360 2017-08-02 Tom de Vries <tom@codesourcery.com>
7361 Cesar Philippidis <cesar@codesourcery.com>
7363 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7364 Add missing edge probabilities.
7366 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7368 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7371 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7374 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7375 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7376 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7377 prologue_seq sequences - if any.
7379 2017-08-02 Richard Biener <rguenther@suse.de>
7381 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7382 via vectors if supported, integer extracts via punning if supported
7383 or otherwise vector extracts.
7385 2017-08-02 Richard Biener <rguenther@suse.de>
7387 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7389 (bitmap_insert_into_set): ... this.
7391 2017-08-02 Richard Biener <rguenther@suse.de>
7393 PR tree-optimization/81633
7395 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7397 PR tree-optimization/71752
7398 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7400 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7402 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7403 (machine_function::call_ms2sysv_pad_out): Remove field.
7404 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7405 (ix86_compute_frame_layout): Likewise.
7407 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7410 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7411 attribute with interrupt attribute.
7413 2017-08-01 Andrew Pinski <apinski@cavium.com>
7415 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7416 BIT_INSERT_EXPR's operand 1
7417 to see if the types precision matches.
7419 2017-08-01 Martin Liska <mliska@suse.cz>
7422 * builtins.c (expand_builtin_memcpy_args): Remove.
7423 (expand_builtin_memcpy): Call newly added function
7424 expand_builtin_memory_copy_args.
7425 (expand_builtin_memcpy_with_bounds): Likewise.
7426 (expand_builtin_mempcpy): Remove last argument.
7427 (expand_builtin_mempcpy_with_bounds): Likewise.
7428 (expand_builtin_memory_copy_args): New function created from
7429 expand_builtin_mempcpy_args with small modifications.
7430 (expand_builtin_mempcpy_args): Remove.
7431 (expand_builtin_stpcpy): Remove unused argument.
7432 (expand_builtin): Likewise.
7433 (expand_builtin_with_bounds): Likewise.
7435 2017-08-01 Martin Liska <mliska@suse.cz>
7438 Make mempcpy more optimal (PR middle-end/70140).
7440 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7443 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7444 __builtin_vec_cmpne verify both arguments are compatible vectors
7445 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7446 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7447 move computation of aligned to after checking the argument types.
7451 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7452 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7455 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7456 Doug Rupp <rupp@adacore.com>
7457 Olivier Hainque <hainque@adacore.com>
7459 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7460 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7462 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7463 for TARGET_OS_CPP_BUILTIN.
7464 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7465 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7467 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7468 passing required abi options to the assembler for EABI configurations.
7469 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7470 of .text.hot and .text.unlikely sections for kernel modules when
7471 using ARM style exceptions.
7472 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7473 options. Add EXTRA_CC1_SPEC.
7474 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7476 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7478 (ARM_TARGET2_DWARF_FORMAT): Define.
7479 * config/arm/t-vxworks: Adjust multilib control to removal of the
7480 Diab command line options.
7482 2017-08-01 Martin Liska <mliska@suse.cz>
7484 PR gcov-profile/81561
7485 * gcov.c (unblock): Make unblocking safe as we need to preserve
7486 index correspondence of blocks and block_lists.
7488 2017-08-01 Richard Biener <rguenther@suse.de>
7490 PR tree-optimization/81181
7491 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7492 (compute_antic): ... end of iteration here.
7494 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7496 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7497 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7498 (ftree-slp-vectorize): Likewise.
7499 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7500 can no longer be set independent of flag_tree_loop_vectorize.
7501 * omp-general.c (emp_max_vf): Likewise.
7502 * opts.c (enable_fdo_optimizations): Remove references to
7503 flag_tree_vectorize, these are now implicit.
7504 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7505 and leave it for the options machinery.
7507 2017-08-01 Martin Liska <mliska@suse.cz>
7510 * builtins.c (expand_builtin_memcpy_args): Remove.
7511 (expand_builtin_memcpy): Call newly added function
7512 expand_builtin_memory_copy_args.
7513 (expand_builtin_memcpy_with_bounds): Likewise.
7514 (expand_builtin_mempcpy): Remove last argument.
7515 (expand_builtin_mempcpy_with_bounds): Likewise.
7516 (expand_builtin_memory_copy_args): New function created from
7517 expand_builtin_mempcpy_args with small modifications.
7518 (expand_builtin_mempcpy_args): Remove.
7519 (expand_builtin_stpcpy): Remove unused argument.
7520 (expand_builtin): Likewise.
7521 (expand_builtin_with_bounds): Likewise.
7523 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7526 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7527 print "ds:" only for immediates in generic address space.
7529 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7532 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7533 (ix86_function_ok_for_sibcall): Return false for naked functions.
7535 2017-08-01 Richard Biener <rguenther@suse.de>
7537 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7538 (compute_antic): Seed worklist with exit block predecessors.
7539 * cfganal.c (dfs_find_deadend): For a cycle return the source
7540 of the edge closing it.
7542 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7544 * config/aarch64/aarch64.c
7545 (aarch64_can_const_movi_rtx_p): Move 0 check.
7547 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7549 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7550 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7552 * match.pd: Ditto in address comparison pattern.
7554 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7556 PR tree-optimization/81627
7557 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7558 closed ssa form for store-store chain.
7560 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7562 PR tree-optimization/81620
7563 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7564 for store-store chain.
7566 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7568 PR tree-optimization/81588
7569 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7570 ranges[i].in_p, invert comparison code ccode. For >/>=,
7571 swap rhs1 and rhs2 and comparison code unconditionally,
7572 for </<= don't do that. Don't swap rhs1/rhs2 again if
7573 ranges[i].in_p, instead invert comparison code ccode if
7574 opcode or oe->rank is BIT_IOR_EXPR.
7577 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7578 a direct optab to conversion optab.
7579 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7580 with GET_MODE_INNER as last argument instead of optab_handler.
7581 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7582 vector extraction if possible and optab is available.
7583 * expr.c (store_constructor): Use convert_optab_handler instead
7584 of optab_handler. Use vector initialization from smaller
7585 vectors if possible and optab is available.
7586 * tree-vect-stmts.c (vectorizable_load): Likewise.
7587 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7589 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7590 of vec_init from half-sized vectors with the same element mode.
7591 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7592 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7593 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7594 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7595 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7596 after mode in gen_vec_extract* calls.
7597 (vec_extract<mode>): Renamed to ...
7598 (vec_extract<mode><ssescalarmodelower>): ... this.
7599 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7600 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7601 element mode after mode in gen_vec_init* calls.
7602 (VEC_INIT_HALF_MODE): New mode iterator.
7603 (vec_init<mode>): Renamed to ...
7604 (vec_init<mode><ssescalarmodelower>): ... this.
7605 (vec_init<mode><ssehalfvecmodelower>): New expander.
7606 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7607 (vec_extractv2sfsf): ... this.
7608 (vec_initv2sf): Renamed to ...
7609 (vec_initv2sfsf): ... this.
7610 (vec_extractv2si): Renamed to ...
7611 (vec_extractv2sisi): ... this.
7612 (vec_initv2si): Renamed to ...
7613 (vec_initv2sisi): ... this.
7614 (vec_extractv4hi): Renamed to ...
7615 (vec_extractv4hihi): ... this.
7616 (vec_initv4hi): Renamed to ...
7617 (vec_initv4hihi): ... this.
7618 (vec_extractv8qi): Renamed to ...
7619 (vec_extractv8qiqi): ... this.
7620 (vec_initv8qi): Renamed to ...
7621 (vec_initv8qiqi): ... this.
7622 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7623 (vec_init<mode>): Renamed to ...
7624 (vec_init<mode><VEC_base_l>): ... this.
7625 (vec_extract<mode>): Renamed to ...
7626 (vec_extract<mode><VEC_base_l>): ... this.
7627 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7628 (vec_initv2sfsf): ... this.
7629 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7630 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7631 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7632 element mode after mode in gen_vec_init* calls.
7633 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7634 (vec_init<mode><Vel>): ... this.
7635 (vec_extract<mode>): Renamed to ...
7636 (vec_extract<mode><Vel>): ... this.
7637 * config/aarch64/iterators.md (Vel): New mode attribute.
7638 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7639 Add element mode after mode in gen_vec_extract* calls.
7640 * config/s390/vector.md (non_vec_l): New mode attribute.
7641 (vec_extract<mode>): Renamed to ...
7642 (vec_extract<mode><non_vec_l>): ... this.
7643 (vec_init<mode>): Renamed to ...
7644 (vec_init<mode><non_vec_l>): ... this.
7645 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7646 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7648 * config/arm/iterators.md (V_elem_l): New mode attribute.
7649 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7650 (vec_extract<mode><V_elem_l>): ... this.
7651 (vec_extractv2di): Renamed to ...
7652 (vec_extractv2didi): ... this.
7653 (vec_init<mode>): Renamed to ...
7654 (vec_init<mode><V_elem_l>): ... this.
7655 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7656 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7657 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7658 Add element mode after gen_vec_extract* calls.
7659 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7660 (vec_init<mode><unitmode>): ... this.
7661 (vec_extract<mode>): Renamed to ...
7662 (vec_extract<mode><unitmode>): ... this.
7663 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7664 (vec_init<mode><unitmode>): ... this.
7665 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7666 (vec_initv2sfsf): ... this.
7667 (vec_extractv2sf): Renamed to ...
7668 (vec_extractv2sfsf): ... this.
7669 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7670 Add element mode after gen_vec_extract* calls.
7671 * config/mips/mips.md (unitmode): New mode iterator.
7672 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7673 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7674 * config/spu/spu.md (inner_l): New mode attribute.
7675 (vec_init<mode>): Renamed to ...
7676 (vec_init<mode><inner_l>): ... this.
7677 (vec_extract<mode>): Renamed to ...
7678 (vec_extract<mode><inner_l>): ... this.
7679 * config/sparc/sparc.md (veltmode): New mode iterator.
7680 (vec_init<VMALL:mode>): Renamed to ...
7681 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7682 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7683 (vec_initv2sisi): ... this.
7684 (vec_initv2sf): Renamed to ...
7685 (vec_initv2sfsf): ... this.
7686 (vec_extractv2sf): Renamed to ...
7687 (vec_extractv2sfsf): ... this.
7688 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7689 (vec_init<mode>): Renamed to ...
7690 (vec_init<mode><VEC_base_l>): ... this.
7691 (vec_extract<mode>): Renamed to ...
7692 (vec_extract<mode><VEC_base_l>): ... this.
7693 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7694 (vec_initv2sfsf): ... this.
7695 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7696 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7697 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7698 gen_vec_init* calls.
7700 2017-08-01 Richard Biener <rguenther@suse.de>
7702 PR tree-optimization/81297
7703 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7704 TREE_OVERFLOW from INTEGER_CSTs.
7706 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7708 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7710 2017-07-31 Carl Love <cel@us.ibm.com>
7712 * config/rs6000/rs6000-c: Add support for built-in functions
7713 vector signed char vec_xl_be (signed long long, signed char *);
7714 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7715 vector signed int vec_xl_be (signed long long, signed int *);
7716 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7717 vector signed long long vec_xl_be (signed long long, signed long long *);
7718 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7719 vector signed short vec_xl_be (signed long long, signed short *);
7720 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7721 vector double vec_xl_be (signed long long, double *);
7722 vector float vec_xl_be (signed long long, float *);
7723 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7724 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7725 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7727 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7728 (altivec_expand_builtin): Add switch statement to call
7729 altivec_expand_xl_be for each builtin.
7730 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7731 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7732 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7733 __builtin_vsx_le_be_v16qi.
7734 * doc/extend.texi: Update the built-in documentation file for the
7735 new built-in functions.
7737 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7740 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7742 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7744 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7746 * config.gcc: Add z14.
7747 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7748 CPU model numbers for z13s and z14.
7749 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7751 * config/s390/s390-opts.h (enum processor_type): Rename
7752 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7753 * config/s390/s390.c (processor_table): Add field for CPU name to
7754 be passed to Binutils.
7755 (s390_asm_output_machine_for_arch): Use the new field in
7756 processor_table for Binutils.
7757 (s390_expand_builtin): Replace arch12 with z14.
7758 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7759 (s390_get_sched_attrmask): Likewise.
7760 (s390_get_unit_mask): Likewise.
7761 * config/s390/s390.opt: Add z14 to processor_type enum.
7763 2017-07-31 Martin Jambor <mjambor@suse.cz>
7766 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7767 regardless of optimization level.
7769 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7770 Martin Liska <mliska@suse.cz>
7772 * predict.def: Remove old comment and adjust probability.
7773 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7776 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7779 * config/i386/i386.c (ix86_function_naked): New function.
7780 (ix86_can_use_return_insn_p): Return false for naked functions.
7781 (ix86_expand_prologue): Skip prologue for naked functions.
7782 (ix86_expand_epilogue): Skip epilogue for naked functions
7783 and emit trap instruction.
7784 (ix86_warn_func_return): New function.
7785 (ix86_attribute_table): Add "naked" attribute specification.
7786 (TARGET_WARN_FUNC_RETURN): Define.
7787 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7789 2017-07-31 Martin Liska <mliska@suse.cz>
7791 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7792 (dump_gimple_bb_header): Always dump BB info.
7793 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7795 2017-07-31 Martin Liska <mliska@suse.cz>
7798 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7799 also with current_function_decl non-null equality.
7801 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7804 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7805 change type to the element type, instead add eltype variable and
7806 use it where we are interested in the element type.
7808 PR tree-optimization/81603
7809 * ipa-polymorphic-call.c
7810 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7811 offset arithmetic in offset_int, bail out if the resulting bit offset
7812 doesn't fit into shwi.
7814 2017-07-31 Martin Liska <mliska@suse.cz>
7816 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7817 (gimplify_save_expr): Fix comment.
7819 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7822 * config/i386/i386.c (ix86_function_arg): Update arguments for
7824 (ix86_compute_frame_layout): Set the initial stack offset to
7825 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7826 INCOMING_FRAME_SP_OFFSET.
7827 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7828 stack before exception handler returns.
7829 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7830 the 'ERROR_CODE' for exception handler.
7832 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7834 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7835 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7836 (ASM_OUTPUT_REG_POP): Ditto.
7837 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7838 instead of asm_fprintf to output pure string.
7840 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7842 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7843 to imported_module_or_decl hook.
7844 (debug_nothing_tree_tree_tree_bool): Remove.
7845 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7846 * debug.c (do_nothing_debug_hooks): Use
7847 debug_nothing_tree_tree_tree_bool_bool instead of
7848 debug_nothing_tree_tree_tree_bool.
7849 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7850 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7851 * sdbout.c (sdb_debug_hooks): Likewise.
7852 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7853 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7855 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7856 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7857 attribute, don't add anything.
7859 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7861 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7862 (fold_build2_stat_loc): Likewise.
7863 (fold_build3_stat_loc): Likewise.
7864 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7865 (fold_build1_loc): Remove macro.
7866 (fold_build2_loc): Likewise.
7867 (fold_build3_loc): Likewise.
7869 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7871 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7872 (gimple_build_debug_bind_source_stat): Likewise.
7873 * gimple.h (gimple_build_debug_bind): Remove macro.
7874 (gimple_build_debug_bind_source): Likewise.
7876 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7878 * bitmap.c (bitmap_alloc): Adjust.
7879 (bitmap_gc_alloc): Likewise.
7880 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7882 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7884 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7885 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7886 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7887 (bitmap_gc_alloc_stat): Likewise.
7888 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7890 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7892 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7893 * rtl.h (shallow_copy_rtx): Remove macro.
7895 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7897 * emit-rtl.c (gen_raw_REG): Adjust.
7898 * gengenrtl.c (gendef): Likewise.
7899 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7900 * rtl.h (rtx_alloc): Remove macro.
7902 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7904 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7905 (build_tree_list_stat): Likewise.
7906 * tree.h (build_tree_list): Remove macro.
7907 (build_tree_list_vec): Likewise.
7909 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7911 * tree.c (make_vector_stat): Remove _stat from name.
7912 (build_vector_stat): Likewise.
7913 * tree.h (make_vector_stat): Remove macro.
7914 (build_vector_stat): Likewise.
7916 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7918 * tree.h (build_var_debug_value): Remove prototype.
7920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7922 * tree.c (tree_cons_stat): Remove _stat from name.
7923 * tree.h (tree_cons): Remove macro.
7925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7927 * tree.c (build_vl_exp_stat): Remove _stat from name.
7928 * tree.h (build_vl_exp): Remove macro.
7930 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7932 * tree.c (build_decl_stat): Remove _stat from name.
7933 * tree.h (build_decl): Remove macro.
7935 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7937 * gimple.c (gimple_build_with_ops_stat): Adjust.
7938 (gimple_alloc_stat): Remove _stat from name.
7939 * gimple.h (gimple_alloc): Remove macro.
7941 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7943 * tree.c (make_tree_vec_stat): Remove _stat from name.
7944 (grow_tree_vec_stat): Likewise.
7945 * tree.h (make_tree_vec_stat): Adjust prototype.
7946 (grow_tree_vec_stat): Likewise.
7947 (make_tree_vec): Remove macro.
7948 (grow_tree_vec): Likewise.
7950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7952 * fold-const.c (fold_build1_stat_loc): Adjust.
7953 (fold_build2_stat_loc): Likewise.
7954 (fold_build3_stat_loc): Likewise.
7955 * tree.c (build0_stat): Remove _stat from name.
7956 (build1_stat): Likewise.
7957 (build2_stat): Likewise.
7958 (build3_stat): Likewise.
7959 (build4_stat): Likewise.
7960 (build5_stat): Likewise.
7961 * tree.h (build1_loc): Remove macro, and rename _stat function
7963 (build2_loc): Likewise.
7964 (build3_loc): Likewise.
7965 (build4_loc): Likewise.
7966 (build5_loc): Likewise.
7968 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7970 * tree.c (make_int_cst_stat): Remove _stat from name.
7971 * tree.h (make_int_cst_stat): Adjust prototype.
7972 (make_int_cst): Remove macro.
7974 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7976 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7977 * tree.h (make_tree_binfo_stat): Adjust prototype.
7978 (make_tree_binfo): Remove.
7980 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7982 * tree.c (copy_node_stat): Rename to copy_node.
7983 (build_distinct_type_copy): Adjust.
7984 * tree.h (copy_node_stat): Adjust prototype.
7985 (copy_node): Remove macro.
7987 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7989 * tree.c (make_node_stat): rename to make_node.
7990 (build_tree_list_stat): Adjust.
7991 (build0_stat): Likewise.
7992 (build2_stat): Likewise.
7993 (build3_stat): Likewise.
7994 (build4_stat): Likewise.
7995 (build5_stat): Likewise.
7996 (build_decl_stat): Likewise.
7997 * tree.h (make_node_stat): Adjust prototype.
7998 (make_node): remove macro.
8000 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8002 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8003 (PPC_FEATURE2_SCV): Likewise.
8004 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8006 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8008 * config/aarch64/aarch64.c
8009 (aarch64_internal_mov_immediate): Add new special pattern.
8010 * config/aarch64/aarch64.md (*movdi_aarch64):
8011 Add reg/32bit const mov case.
8013 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8014 Richard Sandiford <richard.sandiford@linaro.org>
8016 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8017 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8018 Add integer and movi cases.
8019 (movi-split-hf-df-sf split, fp16): New.
8020 (enabled): Added TARGET_FP_F16INST.
8021 * config/aarch64/iterators.md (GPF_HF): New.
8022 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8024 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8026 * config/aarch64/aarch64.c
8027 (aarch64_simd_container_mode): Add prototype.
8028 (aarch64_expand_mov_immediate): Add HI support.
8029 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8030 (aarch64_can_const_movi_rtx_p): New.
8031 (aarch64_preferred_reload_class):
8032 Remove restrictions of using FP registers for certain SIMD operations.
8033 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8034 (aarch64_valid_floating_const): Add integer move validation.
8035 (aarch64_simd_imm_scalar_p): Remove.
8036 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8037 (aarch64_legitimate_constant_p): Expand list of supported cases.
8038 * config/aarch64/aarch64-protos.h
8039 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8040 (aarch64_reinterpret_float_as_int): New.
8041 (aarch64_simd_imm_scalar_p): Remove.
8042 * config/aarch64/constraints.md (Uvi): New.
8043 (Dd): Split into Ds and new Dd.
8044 * config/aarch64/aarch64.md (*movsi_aarch64):
8046 (*movdi_aarch64): Add SIMD mov case.
8048 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8050 * tree-predcom.c: (struct chain): Handle store-store chain in which
8051 stores for elimination only store loop invariant values.
8052 (execute_pred_commoning_chain): Ditto.
8053 (prepare_initializers_chain_store_elim): Ditto.
8054 (prepare_finalizers): Ditto.
8055 (is_inv_store_elimination_chain): New function.
8056 (initialize_root_vars_store_elim_1): New function.
8058 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8060 * tree-predcom.c: Revise general description of the pass.
8061 (enum chain_type): New enum type for store elimination.
8062 (struct chain): New field supporting store elimination.
8063 (struct component): Ditto.
8064 (dump_chain): Dump store-stores chain.
8065 (release_chain): Release resources.
8066 (split_data_refs_to_components): Compute and create component
8067 contains only stores for elimination.
8068 (get_chain_last_ref_at): New function.
8069 (make_invariant_chain): Initialization.
8070 (make_rooted_chain): Specify chain type in parameter and record it.
8071 (add_looparound_copies): Skip for store-stores chain.
8072 (determine_roots_comp): Compute type of chain and pass it to
8074 (initialize_root_vars_store_elim_2): New function.
8075 (finalize_eliminated_stores): New function.
8076 (remove_stmt): Handle store for elimination.
8077 (execute_pred_commoning_chain): Execute predictive commoning on
8079 (determine_unroll_factor): Skip unroll for store-stores chain.
8080 (prepare_initializers_chain_store_elim): New function.
8081 (prepare_initializers_chain): Hanlde store-store chain.
8082 (prepare_finalizers_chain, prepare_finalizers): New function.
8083 (tree_predictive_commoning_loop): Return integer value indicating
8084 if loop is unrolled or lcssa form is corrupted.
8085 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
8087 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8089 * tree-predcom.c (initialize_root): Delete.
8090 (execute_pred_commoning_chain): Initialize root vars and replace
8091 reference of non-combined chain directly, rather than call above
8094 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8096 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
8097 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
8099 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8101 * tree-predcom.c (struct chain): New field init_seq.
8102 (release_chain): Release init_seq.
8103 (prepare_initializers_chain): Record intialization stmts in above
8105 (insert_init_seqs): New function.
8106 (tree_predictive_commoning_loop): Call insert_init_seqs.
8108 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8110 * tree-predcom.c (determine_roots_comp): Skip trivial components.
8112 2017-07-28 Richard Biener <rguenther@suse.de>
8114 * match.pd: Remove superfluous :c.
8115 * genmatch.c (simplify::id): Add member.
8116 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
8118 (current_id): New global.
8119 (dt_node::parent): Move from ...
8120 (dt_operand::parent): ... here. Add for_id member.
8121 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
8122 (decision_tree::find_node): Relax order requirement when
8123 merging DT_TRUE nodes to ones inbetween the current simplify
8124 and the one we try to merge with. Add diagnostic whenever
8125 we need to enforce pattern order by not merging.
8126 (decision_tree::insert): Set current_id.
8127 (decision_tree::print_node): Dump parent node and for_id.
8128 (parser::last_id): Add member.
8129 (parser::push_simplify): Assign unique id.
8130 (parser::parser): Initialize last_id.
8132 2017-07-28 Martin Liska <mliska@suse.cz>
8135 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8136 gimple_build_debug_bind.
8138 2017-07-28 Richard Biener <rguenther@suse.de>
8140 PR tree-optimization/81502
8141 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8143 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8145 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8146 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8147 for BIT_FIELD_REF args.
8148 * fold-const.c (make_bit_field_ref): Likewise.
8149 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8151 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8154 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8155 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8156 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8157 Or it into SANITIZER_UNDEFINED.
8158 * ubsan.c: Include gimple-fold.h and varasm.h.
8159 (ubsan_expand_ptr_ifn): New function.
8160 (instrument_pointer_overflow): New function.
8161 (maybe_instrument_pointer_overflow): New function.
8162 (instrument_object_size): Formatting fix.
8163 (pass_ubsan::execute): Call instrument_pointer_overflow
8164 and maybe_instrument_pointer_overflow.
8165 * internal-fn.c (expand_UBSAN_PTR): New function.
8166 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8167 * sanitizer.def (__ubsan_handle_pointer_overflow,
8168 __ubsan_handle_pointer_overflow_abort): New builtins.
8169 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8170 * internal-fn.def (UBSAN_PTR): New internal function.
8171 * opts.c (sanitizer_opts): Add pointer-overflow.
8172 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8173 * fold-const.c (build_range_check): Compute pointer range check in
8174 integral type if pointer arithmetics would be needed. Formatting
8177 2017-07-28 Martin Liska <mliska@suse.cz>
8180 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8181 parameters that are of a variable-length.
8183 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8185 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8187 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8188 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8189 (CRT_CALL_STATIC_FUNCTION): Likewise.
8190 (ASM_DEFAULT_SPEC): New define.
8191 (ASM_SPEC32): Likewise.
8192 (ASM_SPEC64): Likewise.
8193 (ASM_SPEC_COMMON): Likewise.
8194 (ASM_SPEC): Likewise.
8195 (INVALID_64BIT): Likewise.
8196 (LINK_OS_DEFAULT_SPEC): Likewise.
8197 (LINK_OS_SPEC32): Likewise.
8198 (LINK_OS_SPEC64): Likewise.
8199 (POWERPC_LINUX): Likewise.
8200 (PTRDIFF_TYPE): Likewise.
8201 (RESTORE_FP_PREFIX): Likewise.
8202 (RESTORE_FP_SUFFIX): Likewise.
8203 (SAVE_FP_PREFIX): Likewise.
8204 (SAVE_FP_SUFFIX): Likewise.
8205 (SIZE_TYPE): Likewise.
8206 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8207 (TARGET_64BIT): Likewise.
8208 (TARGET_64BIT): Likewise.
8209 (TARGET_AIX): Likewise.
8210 (WCHAR_TYPE_SIZE): Likewise.
8211 (WCHAR_TYPE): Undefine.
8212 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8213 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8214 (CPP_OS_RTEMS_SPEC): Delete.
8215 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8216 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8218 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8220 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8222 PR tree-optimization/81578
8223 * tree-parloops.c (build_new_reduction): Bail out if
8224 reduction_code isn't one of the standard OpenMP reductions.
8225 Move the details printing after that decision.
8227 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8229 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8230 related to reload_in_progress.
8231 (splat_input_operand): Likewise.
8232 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8234 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8236 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8237 (TARGET_INSTANTIATE_DECLS): Likewise.
8238 (legitimate_indexed_address_p): Delete reload_in_progress code.
8239 (rs6000_debug_legitimate_address_p): Likewise.
8240 (rs6000_eliminate_indexed_memrefs): Likewise.
8241 (rs6000_emit_le_vsx_store): Likewise.
8242 (rs6000_emit_move_si_sf_subreg): Likewise.
8243 (rs6000_emit_move): Likewise.
8244 (register_to_reg_type): Likewise.
8245 (rs6000_pre_atomic_barrier): Likewise.
8246 (rs6000_machopic_legitimize_pic_address): Likewise.
8247 (rs6000_allocate_stack_temp): Likewise.
8248 (rs6000_address_for_fpconvert): Likewise.
8249 (rs6000_address_for_altivec): Likewise.
8250 (rs6000_secondary_memory_needed_rtx): Delete function.
8251 (rs6000_check_sdmode): Likewise.
8252 (rs6000_alloc_sdmode_stack_slot): Likewise.
8253 (rs6000_instantiate_decls): Likewise.
8254 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8255 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8256 Delete reload_in_progress.
8257 (*vec_reload_and_plus_<mptrsize>): Likewise.
8258 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8259 (vsx_div_v2di): Likewise.
8260 (vsx_udiv_v2di): Likewise.
8262 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8264 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8265 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8266 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8267 (rs6000_debug_reg_global): Delete print of LRA status.
8268 (rs6000_option_override_internal): Delete dead LRA related code.
8269 (rs6000_lra_p): Delete function.
8270 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8272 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8274 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8275 * config/riscv/rtems.h: New file.
8277 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8278 Sudakshina Das <sudi.das@arm.com>
8280 * config/aarch64/aarch64.md
8281 (define_split for and<mode>3nr_compare): Move
8282 non aarch64_logical_operand to a register.
8283 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8284 register immediate operand to a register.
8285 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8287 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8290 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8292 2017-07-27 Richard Biener <rguenther@suse.de>
8294 PR tree-optimization/81573
8295 PR tree-optimization/81494
8296 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8297 multi defuse cycle case.
8299 2017-07-27 Richard Biener <rguenther@suse.de>
8301 PR tree-optimization/81571
8302 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8305 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8307 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8308 earlier and only if MASK_FPU is set. Adjust formatting.
8310 2017-07-27 Martin Liska <mliska@suse.cz>
8312 * opt-functions.awk: Add validation of value of Init.
8313 * optc-gen.awk: Pass new argument.
8315 2017-07-27 Martin Liska <mliska@suse.cz>
8317 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8318 Fix wrong condition.
8320 2017-07-27 Martin Liska <mliska@suse.cz>
8322 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8323 BBs and edges seen by autoFDO.
8325 2017-07-27 Richard Biener <rguenther@suse.de>
8327 PR tree-optimization/81502
8328 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8329 with incompatible but same sized type.
8330 (execute_update_addresses_taken): Likewise.
8332 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8334 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8335 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8337 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8340 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8341 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8342 Change s_operand to memory_operand.
8344 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8346 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8347 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8348 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8349 Emit instructions rather than returning an expression. Handle TFmode
8350 and KFmode by casting to TImode.
8351 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8352 (rs6000_emit_le_vsx_store): Likewise.
8353 * config/rs6000/vsx.md (VSX_TI): New iterator.
8354 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8355 (*vsx_le_undo_permute_<mode>): Likewise.
8356 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8357 emit the split sequence.
8358 (*vsx_le_perm_store_<mode>): Likewise.
8360 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8362 PR tree-optimization/81555
8363 PR tree-optimization/81556
8364 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8365 if true, force CHANGED for the recursive invocation.
8366 (reassociate_bb): Remember original length of ops array, pass
8367 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8369 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8370 attributes for noipa attribute. For naked attribute use
8371 lookup_attribute first before lookup_attribute_spec.
8372 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8374 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8375 for functions with noipa attribute.
8376 (cgraph_externally_visible_p): Return true for functions with noipa
8378 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8379 for functions with noipa attribute.
8380 * doc/extend.texi: Document noipa function attribute.
8381 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8382 also for functions with noipa attribute.
8383 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8385 2017-07-26 Andrew Pinski <apinski@cavium.com>
8387 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8388 vec_unalign_load_cost and vec_unalign_store_cost.
8390 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8392 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8393 -mvsx-small-integer option.
8394 (ISA_3_0_MASKS_IEEE): Likewise.
8395 (OTHER_VSX_VECTOR_MASKS): Likewise.
8396 (POWERPC_MASKS): Likewise.
8397 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8398 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8399 code, only testing for DImode being allowed in non-VSX floating
8401 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8402 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8404 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8405 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8406 TARGET_P8_VECTOR test.
8407 (rs6000_secondary_reload_simple_move): Likewise.
8408 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8409 since TARGET_P9_VECTOR was already tested.
8410 (rs6000_opt_masks): Remove -mvsx-small-integer.
8411 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8412 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8414 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8415 test for TARGET_VEXTRACTUB was used, and that uses
8417 (p9 extract splitter): Likewise.
8418 (vsx_extract_<mode>_di_p9): Likewise.
8419 (vsx_extract_<mode>_store_p9): Likewise.
8420 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8421 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8422 the elimination of TARGET_VSX_SMALL_INTEGER.
8423 (vsx_extract_<mode>_p8): Likewise.
8424 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8425 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8426 (vsx_set_<mode>_p9): Likewise.
8427 (vsx_set_v4sf_p9): Likewise.
8428 (vsx_set_v4sf_p9_zero): Likewise.
8429 (vsx_insert_extract_v4sf_p9): Likewise.
8430 (vsx_insert_extract_v4sf_p9_2): Likewise.
8431 * config/rs6000/rs6000.md (sign extend splitter): Change
8432 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8433 (floatsi<mode>2_lfiwax_mem): Likewise.
8434 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8435 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8436 since a test for TARGET_P9_VECTOR was used.
8437 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8438 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8439 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8440 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8441 TARGET_P8_VECTOR test.
8442 (fix_trunc<mode>si2_stfiwx): Likewise.
8443 (fix_trunc<mode>si2_internal): Likewise.
8444 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8445 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8447 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8448 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8449 TARGET_P8_VECTOR test.
8450 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8451 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8452 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8454 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8455 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8456 since a test for TARGET_P9_VECTOR was used.
8457 (splitter for loading small constants): Likewise.
8459 2017-07-26 Andrew Pinski <apinski@cavium.com>
8461 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8464 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8467 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8468 (fp_valid_at): Likewise.
8470 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8472 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8473 (qdf24xx_addrcost_table): Likewise.
8474 (cortexa57_tunings): Update to use generic_branch_cost.
8475 (cortexa72_tunings): Likewise.
8476 (cortexa73_tunings): Likewise.
8477 (qdf24xx_tunings): Likewise.
8479 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8481 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8482 (thunderx2t99_branch_cost): Likewise.
8483 (cortexa35_tunings): Update to use generic_branch_cost.
8484 (cortexa53_tunings): Likewise.
8485 (cortexa57_tunings): Likewise.
8486 (cortexa72_tunings): Likewise.
8487 (cortexa73_tunings): Likewise.
8488 (thunderx2t99_tunings): Likewise.
8490 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8492 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8493 (sparc_option_override): Honour MASK_FSMULD.
8494 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8495 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8496 * config/sparc/sparc.opt (mfsmuld): New option.
8497 * doc/invoke.texi (mfsmuld): Document option.
8499 2017-07-26 Marek Polacek <polacek@redhat.com>
8502 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8504 2017-07-26 Richard Biener <rguenther@suse.de>
8506 * gimple-match-head.c (do_valueize): Return OP if valueize
8508 (get_def): New helper to get at the def stmt of a SSA name
8510 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8511 do_valueize to get at the def stmt.
8512 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8514 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8517 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8519 2017-07-26 Martin Liska <mliska@suse.cz>
8522 * function.c (expand_function_start): Make expansion of
8523 nonlocal_goto_save_area after parm_birth_insn.
8525 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8527 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8528 from all CPU target flags enable members.
8530 2017-07-26 Richard Biener <rguenther@suse.de>
8532 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8533 (decision_tree::gen): Make 'type' const.
8534 (write_predicate): Likewise.
8536 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8538 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8539 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8540 (rs6000_option_override_internal): Likewise.
8541 (rs6000_expand_vector_set): Likewise.
8542 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8543 (TARGET_UPPER_REGS_SF): Likewise.
8544 (TARGET_UPPER_REGS_DI): Likewise.
8545 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8546 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8547 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8548 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8549 (Splitters for DI constants in Altivec registers): Likewise.
8550 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8551 (vsx_set_v4sf_p9): Likewise.
8552 (vsx_set_v4sf_p9_zero): Likewise.
8553 (vsx_insert_extract_v4sf_p9): Likewise.
8554 (vsx_insert_extract_v4sf_p9_2): Likewise.
8556 2017-07-25 Carl Love <cel@us.ibm.com>
8558 * doc/extend.texi: Update the built-in documentation file for the
8559 existing built-in functions
8560 vector signed char vec_cnttz (vector signed char);
8561 vector unsigned char vec_cnttz (vector unsigned char);
8562 vector signed short vec_cnttz (vector signed short);
8563 vector unsigned short vec_cnttz (vector unsigned short);
8564 vector signed int vec_cnttz (vector signed int);
8565 vector unsigned int vec_cnttz (vector unsigned int);
8566 vector signed long long vec_cnttz (vector signed long long);
8567 vector unsigned long long vec_cnttz (vector unsigned long long);
8569 2017-07-25 Andrew Pinski <apinski@cavium.com>
8571 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8572 accesses where the use is for the first operand of a BIT_INSERT.
8574 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8577 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8578 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8580 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8582 * config/i386/gstabs.h: Delete.
8583 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8585 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8587 * config/i386/i386.c (ix86_decompose_address): Do not check for
8588 register RTX when looking at index_reg or base_reg.
8589 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8591 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8593 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8594 to update EH info here.
8596 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8598 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8600 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8602 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8604 2017-07-25 Torsten Duwe <duwe@suse.de>
8606 * common.opt: Introduce -fpatchable-function-entry
8607 command line option, and its variables function_entry_patch_area_size
8608 and function_entry_patch_area_start.
8609 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8610 including a two-value parser.
8611 * target.def (print_patchable_function_entry): New target hook.
8612 * targhooks.h (default_print_patchable_function_entry): New function.
8613 * targhooks.c (default_print_patchable_function_entry): Likewise.
8614 * toplev.c (process_options): Switch off IPA-RA if
8615 patchable function entries are being generated.
8616 * varasm.c (assemble_start_function): Look at the
8617 patchable-function-entry command line switch and current
8618 function attributes and maybe generate NOP instructions by
8619 calling the print_patchable_function_entry hook.
8620 * doc/extend.texi: Document patchable_function_entry attribute.
8621 * doc/invoke.texi: Document -fpatchable_function_entry
8622 command line option.
8623 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8625 * doc/tm.texi: Re-generate.
8627 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8630 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8631 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8633 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8635 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8637 2017-07-25 Richard Biener <rguenther@suse.de>
8639 PR tree-optimization/81455
8640 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8641 not walk in cycles when looking for guards.
8643 2017-07-25 Richard Biener <rguenther@suse.de>
8645 PR tree-optimization/81529
8646 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8647 when optimizing backedge uses.
8649 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8651 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8653 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8654 to dl_section_ref. On AIX, append an expression to subtract
8655 the size of the section length to dl_section_ref.
8657 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8659 * configure.ac: If any of the config.* scripts fail, exit 1.
8660 * configure: Regenerate.
8662 2017-07-25 Richard Biener <rguenther@suse.de>
8665 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8666 of immediate uses, be more verbose on errors.
8668 2017-07-25 Richard Biener <rguenther@suse.de>
8670 PR tree-optimization/81510
8671 * tree-vect-loop.c (vect_is_simple_reduction): When the
8672 reduction stmt is not inside the loop bail out.
8674 2017-07-25 Richard Biener <rguenther@suse.de>
8676 PR tree-optimization/81303
8677 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8678 profitability check against LOOP_VINFO_NITERSM1.
8680 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8682 * domwalk.c (cmp_bb_postorder): Simplify.
8683 (sort_bbs_postorder): New function. Use it...
8684 (dom_walker::walk): ...here to optimize common cases.
8686 2017-07-25 Martin Liska <mliska@suse.cz>
8689 * ipa-visibility.c (function_and_variable_visibility): Make the
8690 redirection just on target that supports aliasing.
8691 Fix GNU coding style.
8693 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8696 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8698 * config/arm/rtems.h: Likewise.
8699 * config/bfin/rtems.h: Likewise.
8700 * config/i386/rtemself.h: Likewise.
8701 * config/lm32/rtems.h: Likewise.
8702 * config/m32c/rtems.h: Likewise.
8703 * config/m68k/rtemself.h: Likewise.
8704 * config/microblaze/rtems.h: Likewise.
8705 * config/mips/rtems.h: Likewise.
8706 * config/moxie/rtems.h: Likewise.
8707 * config/nios2/rtems.h: Likewise.
8708 * config/powerpcspe/rtems.h: Likewise.
8709 * config/rs6000/rtems.h: Likewise.
8710 * config/rtems.h: Likewise.
8711 * config/sh/rtems.h: Likewise.
8712 * config/sh/rtemself.h: Likewise.
8713 * config/sparc/rtemself.h: Likewise.
8715 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8718 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8719 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8720 * tree-ssa-structalias.c (alias_get_name): Same.
8722 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8725 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8726 instructions if no du chain is found.
8728 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8730 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8732 2017-07-25 Richard Biener <rguenther@suse.de>
8735 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8738 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8740 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8742 (ISA_2_7_MASKS_SERVER): Likewise.
8743 (ISA_3_0_MASKS_IEEE): Likewise.
8744 (OTHER_P8_VECTOR_MASKS): Likewise.
8745 (OTHER_VSX_VECTOR_MASKS): Likewise.
8746 (POWERPC_MASKS): Likewise.
8747 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8748 duplicate list of options.
8749 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8750 explicit -mupper-regs options.
8751 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8752 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8753 alias for -mupper-regs-df.
8754 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8755 (rs6000_init_hard_regno_mode_ok): Likewise.
8756 (rs6000_option_override_internal): Likewise.
8757 (rs6000_opt_masks): Likewise.
8758 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8759 options in terms of whether -mvsx or -mpower8-vector was used.
8760 (TARGET_UPPER_REGS_DI): Likewise.
8761 (TARGET_UPPER_REGS_SF): Likewise.
8762 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8763 -mupper-regs-* options.
8765 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8767 * passes.c (emergency_dump_function): Print some empty lines and a
8768 header before the RTL dump.
8770 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8772 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8774 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8777 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8778 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8780 2017-07-24 Carl Love <cel@us.ibm.com>
8782 * config/rs6000/rs6000-c.c: Add support for built-in functions
8783 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8784 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8785 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8786 vec_extract_fp_from_shortl): Add defines for the two builtins.
8787 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8788 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8790 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8791 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8792 * doc/extend.texi: Update the built-in documentation file for the
8793 new built-in function.
8795 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8798 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8800 * doc/generic.texi: Likewise.
8801 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8802 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8804 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8806 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8807 (aarch64_mls_elt_merge<mode>): Likewise.
8809 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8811 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8812 having __cxa_atexit.
8814 2017-07-23 Michael Collison <michael.collison@arm.com>
8816 * config/arm/arm.c (arm_option_override): Deprecate
8817 use of -mstructure-size-boundary.
8818 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8819 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8821 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8824 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8825 Reduce cost estimate for direct moves.
8827 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8830 * config/i386/i386.c (ix86_option_override_internal): Disable
8831 BMI, BMI2 and TBM instructions for -m16.
8833 2017-07-21 Carl Love <cel@us.ibm.com>
8835 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8836 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8837 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8838 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8839 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8840 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8841 VMULOSW): New enum "unspec" values.
8842 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8843 altivec_vmulosw): New patterns.
8844 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8845 VMULOSW): Add definitions.
8847 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8849 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8850 (qdf24xx): Likewise.
8851 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8852 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8853 (AARCH64_FL_V8_1): Renumber.
8854 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8855 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8856 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8857 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8858 rdma to feature modifiers list.
8860 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8863 * ipa-visibility (function_and_variable_visibility): Convert
8864 recursive PLT call to direct call if appropriate.
8866 2017-07-21 Andrew Pinski <apinski@cavium.com>
8868 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8869 operand 1 to see if the types precision matches.
8870 * fold-const.c (operand_equal_p): Likewise.
8872 2017-07-21 Richard Biener <rguenther@suse.de>
8874 PR tree-optimization/81303
8875 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8876 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8877 (vect_peeling_hash_get_lowest_cost): Adjust.
8878 (vect_enhance_data_refs_alignment): Likewise. Use
8879 vect_get_peeling_costs_all_drs to compute the penalty for no
8880 peeling to match up costs.
8882 2017-07-21 Richard Biener <rguenther@suse.de>
8884 PR tree-optimization/81500
8885 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8886 we didn't identify a reduction path.
8888 2017-07-21 Tom de Vries <tom@codesourcery.com>
8889 Cesar Philippidis <cesar@codesourcery.com>
8891 PR gcov-profile/81442
8892 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8895 2017-07-21 Tom de Vries <tom@codesourcery.com>
8898 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8900 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8901 nvptx_override_options_after_change.
8903 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8905 * dwarf2out.c (output_file_names): Avoid double testing for
8908 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8910 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8912 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8914 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8916 (try_crossjump_to_edge): Do not punt on partitioned functions.
8918 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8920 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8921 Put all BBs reachable only via paths crossing cold region to cold
8923 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8925 2016-07-21 Richard Biener <rguenther@suse.de>
8927 PR tree-optimization/81303
8928 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8929 into account prologue and epilogue iterations when raising
8930 min_profitable_iters to sth at least covering one vector iteration.
8932 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8934 * config/arm/arm.c (arm_test_cpu_arch_dat):
8937 2017-07-20 Nathan Sidwell <nathan@acm.org>
8939 Remove TYPE_METHODS.
8940 * tree.h (TYPE_METHODS): Delete.
8941 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8942 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8943 (dbxout_type_methods): Scan TYPE_FIELDS.
8944 (dbxout_type): Don't check TYPE_METHODS here.
8945 * function.c (use_register_for_decl): Always ignore register for
8946 class types when not optimizing.
8947 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8948 * tree.c (free_lang_data_in_type): Stitch out member functions and
8949 templates from TYPE_FIELDS.
8950 (build_distinct_type_copy, verify_type_variant,
8951 verify_type): Member fns are on TYPE_FIELDS.
8952 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8953 * tree-pretty-print.c (dump_generic_node): Likewise.
8955 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8958 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8959 V2TImode and V4TImode.
8960 (ix86_expand_vector_extract): Likewise.
8961 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8962 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8963 (ssescalarmode): Handle V4TImode and V2TImode.
8964 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8965 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8966 (VEXTRACTI128_MODE): New mode iterator.
8967 (splitter for *vec_extractv?ti first element): New.
8968 (VEC_INIT_MODE): New mode iterator.
8969 (vec_init<mode>): Consolidate 3 expanders into one using
8970 VEC_INIT_MODE mode iterator.
8972 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8974 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8975 non_spilled_static_chain_regno_p.
8977 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8979 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8981 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8983 * bb-reorder.c (connect_traces): Allow copying of blocks within
8986 2017-07-20 Richard Biener <rguenther@suse.de>
8988 * gimple.h (gimple_phi_result): Add gphi * overload.
8989 (gimple_phi_result_ptr): Likewise.
8990 (gimple_phi_arg): Likewise. Adjust index assert to only
8991 allow actual argument accesses rather than all slots available
8993 (gimple_phi_arg_def): Add gphi * overload.
8994 * tree-phinodes.c (make_phi_node): Initialize only actual
8996 (resize_phi_node): Clear memory not covered by old node,
8997 do not initialize excess argument slots.
8998 (reserve_phi_args_for_new_edge): Initialize new argument slot
9001 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9003 PR tree-optimization/81388
9005 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9007 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9008 by removing computation of may_be_zero.
9010 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9011 Tom de Vries <tom@codesourcery.com>
9014 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9015 when gimple level profile disagrees with what RTL expander did.
9017 2017-07-20 Richard Biener <rguenther@suse.de>
9019 PR tree-optimization/61171
9020 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9021 (vect_analyze_stmt): Add slp instance parameter.
9022 (vectorizable_reduction): Likewise.
9023 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9024 (vect_is_simple_reduction): Deal with chains not detected
9025 as SLP reduction chain, specifically not properly associated
9026 chains containing a mix of plus/minus.
9027 (get_reduction_op): Remove.
9028 (get_initial_defs_for_reduction): Simplify, pass in whether
9029 this is a reduction chain, pass in the SLP node for the PHIs.
9030 (vect_create_epilog_for_reduction): Get the SLP instance as
9032 (vectorizable_reduction): Get the SLP instance as arg.
9033 During analysis remember the SLP node with the PHIs in the
9034 instance. Simplify getting at the vectorized reduction PHIs.
9035 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9036 through SLP instance.
9037 (vect_slp_analyze_operations): Likewise.
9038 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9039 (vect_transform_stmt): Likewise.
9041 2017-07-20 Tom de Vries <tom@codesourcery.com>
9043 PR tree-optimization/81489
9044 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9045 read of phi arg location to before loop that modifies phi.
9047 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9049 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9052 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9055 * except.c (execute): Fix ordering issue.
9057 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9059 PR rtl-optimization/81423
9060 * combine.c (make_compound_operation_int): Don't try to optimize
9061 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9063 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9065 PR rtl-optimization/81423
9066 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9067 with a constant that is -1 in the truncated to mode.
9069 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9071 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9072 (determine_unlikely_bbs): ... here.
9073 * predict.h (propagate_unlikely_bbs_forward): Declare.
9074 * cfgexpand.c (pass_expand::execute): Use it.
9075 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9077 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9078 propagate_unlikely_bbs_forward.
9080 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9083 * except.c (maybe_add_nop_after_section_switch): New function.
9086 2017-07-19 Tom de Vries <tom@codesourcery.com>
9088 * gimple.h (gimple_phi_set_arg): Make assert more strict.
9090 2017-07-19 Tom de Vries <tom@codesourcery.com>
9092 * gimple.h (gimple_phi_arg): Make assert more strict.
9094 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
9096 * config.gcc (powerpc*-*-*): Add mmintrin.h.
9097 * config/rs6000/mmintrin.h: New file.
9098 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
9100 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9102 PR tree-optimization/81346
9103 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9105 2017-07-19 Tom de Vries <tom@codesourcery.com>
9107 * config/nvptx/nvptx.md (VECIM): Add V2DI.
9109 2017-07-19 Tom de Vries <tom@codesourcery.com>
9111 * config/nvptx/nvptx-modes.def: Add V2DImode.
9112 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
9113 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
9114 (nvptx_output_mov_insn): Handle lack of mov.b128.
9115 (nvptx_print_operand): Handle 'H' and 'L' codes.
9116 (nvptx_vector_mode_supported): Allow V2DImode.
9117 (nvptx_preferred_simd_mode): New function.
9118 (nvptx_data_alignment): New function.
9119 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
9120 nvptx_preferred_simd_mode.
9121 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
9123 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
9125 2017-07-19 Tom de Vries <tom@codesourcery.com>
9127 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
9128 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
9129 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9130 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9131 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9132 (mov<VECIM>_insn): New define_insn.
9133 (define_expand "mov<VECIM>): New define_expand.
9135 2017-07-19 Tom de Vries <tom@codesourcery.com>
9137 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9139 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9141 PR tree-optimization/81346
9142 * fold-const.h (fold_div_compare, range_check_type): Declare.
9143 * fold-const.c (range_check_type): New function.
9144 (build_range_check): Use range_check_type.
9145 (fold_div_compare): No longer static, rewritten into
9146 a match.pd helper function.
9147 (fold_comparison): Don't call fold_div_compare here.
9148 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9151 2017-07-19 Nathan Sidwell <nathan@acm.org>
9153 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9154 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9155 * tree.c (find_decls_types_r, verify_type): Use
9156 TYPE_{MIN,MAX}_VALUE_RAW.
9157 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9158 (hash_tree): Likewise.
9159 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9161 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9164 2017-07-18 Tom de Vries <tom@codesourcery.com>
9167 * omp-expand.c (expand_omp_for_static_chunk): Handle
9168 equal-argument loop exit phi.
9170 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9173 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9174 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9175 operand 2 predicate.
9176 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9177 operand 2 predicate.
9178 (ror,rol -> rorx splitters): Use const_int_operand as
9179 operand 2 predicate.
9181 2017-06-18 Richard Biener <rguenther@suse.de>
9183 PR tree-optimization/81410
9184 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9185 the gap in the ! slp_perm SLP case after each group.
9187 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9190 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9193 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9196 * predict.c (set_even_probabilities): Cleanup; do not affect
9197 probabilities that are already known.
9198 (combine_predictions_for_bb): Call even when count is set.
9200 2017-07-18 Nathan Sidwell <nathan@acm.org>
9202 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9205 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9208 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9209 optimization for loop niter analysis.
9211 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9214 * config/avr/avr.c (avr_optimize_casesi): Don't use
9215 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9217 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9219 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9220 body_cost_vec from _vect_peel_extended_info.
9221 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9222 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9225 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9227 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9229 2017-07-18 Richard Biener <rguenther@suse.de>
9231 PR tree-optimization/80620
9232 PR tree-optimization/81403
9233 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9234 info when re-using a VN table entry.
9236 2017-07-18 Richard Biener <rguenther@suse.de>
9238 PR tree-optimization/81418
9239 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9240 vectype_in. Verify that with lane-reducing reduction operations
9241 we have a single def-use cycle.
9243 2017-07-17 Carl Love <cel@us.ibm.com>
9245 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9247 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9248 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9249 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9250 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9251 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9252 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9253 VMULOSW): New enum "unspec" values.
9254 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9255 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9256 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9257 altivec_vmulosw): New patterns.
9258 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9259 VMULOSW): Add definitions.
9260 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9262 * config/alpha/alpha.c: Include predict.h.
9264 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9266 * tree-vrp.c (compare_assert_loc): Fix comparison function
9267 to return predictable results.
9269 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9271 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9274 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9275 * doc/invoke.texi (mexpand-adddi): Update text.
9277 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9279 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9280 that also clobbers the CC register. The old expand code is moved
9282 (*arc_clzsi2): ... here.
9283 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9284 the CC register. The old expand code is moved to ...
9285 (arc_ctzsi2): ... here.
9287 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9289 * config/arc/arc.opt (mindexed-loads): Use initial value
9290 TARGET_INDEXED_LOADS_DEFAULT.
9291 (mauto-modify-reg): Use initial value
9292 TARGET_AUTO_MODIFY_REG_DEFAULT.
9293 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9294 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9295 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9296 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9298 2017-07-17 Martin Liska <mliska@suse.cz>
9301 * opts.c (finish_options): Do not allow -fgnu-tm
9302 w/ -fsanitize={kernel-,}address. Say sorry.
9304 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9307 * tree-loop-distribution.c (classify_partition): Only assert on
9308 numer of iterations.
9309 (merge_dep_scc_partitions): Delete prameter. Update function call.
9310 (distribute_loop): Remove code handling loop with unknown niters.
9311 (pass_loop_distribution::execute): Skip loop with unknown niters.
9313 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9316 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9317 function sort_partitions_by_post_order.
9319 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9321 PR tree-optimization/81374
9322 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9323 the max index of basic blocks, rather than number of basic blocks.
9325 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9327 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9329 (arc_legitimate_pic_operand_p): Likewise.
9330 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9332 (arc_needs_pcl_p): Likewise.
9333 (arc_legitimate_pc_offset_p): Likewise.
9334 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9335 function is also used in constrains.md.
9336 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9337 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9338 PLUS. Only return true/false in known cases, otherwise assert.
9339 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9340 is already called in arc_legitimate_constant_p.
9341 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9343 (LEGITIMATE_PIC_OPERAND_P): Use
9344 arc_raw_symbolic_reference_mentioned_p function.
9345 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9350 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9351 Andrew Burgess <andrew.burgess@embecosm.com>
9353 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9354 (arc_return_address_register): New function.
9355 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9356 (arc_handle_fndecl_attribute): Add naked attribute.
9357 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9358 (TARGET_WARN_FUNC_RETURN): Likewise.
9359 (arc_allocate_stack_slots_for_args): New function.
9360 (arc_warn_func_return): Likewise.
9361 (machine_function): Change type fn_type.
9362 (arc_compute_function_type): Consider new naked function type,
9363 change function return type.
9364 (arc_must_save_register): Adapt to handle new
9365 arc_compute_function_type's return type.
9366 (arc_expand_prologue): Likewise.
9367 (arc_expand_epilogue): Likewise.
9368 (arc_return_address_regs): Delete.
9369 (arc_return_address_register): New function.
9370 (arc_epilogue_uses): Use above function.
9371 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9372 (arc_function_type): Change encoding, add naked type.
9373 (ARC_INTERRUPT_P): Change to handle the new encoding.
9374 (ARC_FAST_INTERRUPT_P): Likewise.
9375 (ARC_NORMAL_P): Define.
9376 (ARC_NAKED_P): Likewise.
9377 (arc_compute_function_type): Delete prototype.
9378 * config/arc/arc.md (in_ret_delay_slot): Use
9379 arc_return_address_register function.
9380 (simple_return): Likewise.
9381 (p_return_i): Likewise.
9383 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9385 PR tree-optimization/81428
9386 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9387 can't be built for those types.
9389 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9391 Remove stuff dead since r239246.
9393 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9394 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9395 (avr_inform_devices): Remove dead stuff.
9397 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9399 * config/arm/arm_neon.h: Fix softp typo.
9401 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9403 PR tree-optimization/81365
9404 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9405 aggregate moves onto bb predecessor edges, make sure there are no
9406 loads that could alias the lhs in between the start of bb and the
9409 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9412 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9413 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9414 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9416 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9418 PR tree-optimization/81396
9419 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9420 (init_symbolic_number): Initialize it to 1.
9421 (perform_symbolic_merge): Add n_ops from both operands into the new
9423 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9424 without base_addr as useless if they need more than one operation.
9425 (bswap_replace): Handle !bswap case for NULL base_addr.
9427 2017-07-17 Tom de Vries <tom@codesourcery.com>
9430 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9433 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9435 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9436 conditional builtin define __FIX_LEON3FT_B2BST.
9438 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9440 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9441 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9444 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9446 PR rtl-optimization/81424
9447 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9448 to remove potential trapping from operands if -fnon-call-exceptions.
9450 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9452 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9453 profile_proability for scalling.
9454 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9456 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9458 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9460 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9462 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9463 fixpoint arithmetics.
9465 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9467 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9468 fixpoint arithmetics.
9470 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9472 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9473 fixpoint arithmetics.
9475 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9477 * profile-count.h (profile_probability::from_reg_br_prob_note,
9478 profile_probability::to_reg_br_prob_note): New functions.
9479 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9480 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9481 * predict.c (probability_reliable_p): Update.
9482 (edge_probability_reliable_p): Update.
9483 (br_prob_note_reliable_p): Update.
9484 (invert_br_probabilities): Update.
9485 (add_reg_br_prob_note): New function.
9486 (combine_predictions_for_insn): Update.
9487 * asan.c (asan_clear_shadow): Update.
9488 * cfgbuild.c (compute_outgoing_frequencies): Update.
9489 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9490 (update_br_prob_note): Update.
9491 (rtl_verify_edges): Update.
9492 (purge_dead_edges): Update.
9493 (fixup_reorder_chain): Update.
9494 * emit-rtl.c (try_split): Update.
9495 * ifcvt.c (cond_exec_process_insns): Update.
9496 (cond_exec_process_if_block): Update.
9497 (dead_or_predicable): Update.
9498 * internal-fn.c (expand_addsub_overflow): Update.
9499 (expand_neg_overflow): Update.
9500 (expand_mul_overflow): Update.
9501 * loop-doloop.c (doloop_modify): Update.
9502 * loop-unroll.c (compare_and_jump_seq): Update.
9503 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9504 * predict.h: Update.
9505 * reorg.c (mostly_true_jump): Update.
9507 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9508 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9509 * config/arc/arc.c: (emit_unlikely_jump): Update.
9510 * config/arm/arm.c: (emit_unlikely_jump): Update.
9511 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9512 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9513 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9514 (ix86_print_operand): Update.
9515 (ix86_split_fp_branch): Update.
9516 (predict_jump): Update.
9517 * config/ia64/ia64.c (ia64_print_operand): Update.
9518 * config/mmix/mmix.c (mmix_print_operand): Update.
9519 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9520 (rs6000_expand_split_stack_prologue): Update.
9521 * config/rs6000/rs6000.c: Update.
9522 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9523 (s390_expand_vec_movstr): Update.
9524 (s390_expand_cs_tdsi): Update.
9525 (s390_expand_split_stack_prologue): Update.
9526 * config/sh/sh.c (sh_print_operand): Update.
9527 (expand_cbranchsi4): Update.
9528 (expand_cbranchdi4): Update.
9529 * config/sparc/sparc.c (output_v9branch): Update.
9530 * config/spu/spu.c (get_branch_target): Update.
9531 (ea_load_store_inline): Update.
9532 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9533 * config/tilepro/tilepro.c: Update.
9535 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9537 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9538 (gimplify_save_expr): Likewise.
9540 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9542 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9544 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9546 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9548 * timevar.def (TV_IPA_FNSUMMARY): Define.
9550 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9552 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9553 to back store errata sensitive sequence from being generated.
9554 (sqrtdf2_fix): Likewise.
9556 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9558 * tree-ssa-threadupdate.c (compute_path_counts,
9559 update_joiner_offpath_counts): Use profile_probability.
9561 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9564 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9566 * config/arm/arm-c.c (arm_cpu_builtins): Define
9567 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9569 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9571 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9572 array entries to represent __ieee128 versions of the
9573 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9574 scalar_extract_sig, and scalar_insert_exp built-in functions.
9575 (altivec_resolve_overloaded_builtin): Add special case handling
9576 for the __builtin_scalar_insert_exp function, as represented by
9577 the P9V_BUILTIN_VEC_VSIEDP constant.
9578 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9579 exponent support for __ieee128 argument.
9580 (VSESQP): Add scalar extract signature support for __ieee128
9582 (VSTDCNQP): Add scalar test negative support for __ieee128
9584 (VSIEQP): Add scalar insert exponent support for __int128 argument
9585 with __ieee128 result.
9586 (VSIEQPF): Add scalar insert exponent support for __ieee128
9587 argument with __ieee128 result.
9588 (VSTDCQP): Add scalar test data class support for __ieee128
9590 (VSTDCNQP): Add overload support for scalar test negative with
9592 (VSTDCQP): Add overload support for scalar test data class
9594 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9596 (UNSPEC_VSX_SIEXPQP): New constant.
9597 (xsxexpqp): New insn for VSX scalar extract exponent quad
9599 (xsxsigqp): New insn for VSX scalar extract significand quad
9601 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9602 precision with floating point argument.
9603 (xststdcqp): New expand for VSX scalar test data class quad
9605 (xststdcnegqp): New expand for VSX scalar test negative quad
9607 (xststdcqp): New insn to match expansions for VSX scalar test data
9608 class quad precision and VSX scalar test negative quad precision.
9609 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9610 special case operand checking to enforce that second operand of
9611 VSX scalar test data class with quad precision argument is a 7-bit
9613 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9614 prototypes and descriptions of __ieee128 versions of
9615 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9616 scalar_test_data_class, and scalar_test_neg built-in functions.
9618 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9620 PR tree-optimization/81162
9621 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9622 replace a negate with an add.
9624 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9626 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9628 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9630 * config/arm/arm-c.c (arm_cpu_builtins): Define
9631 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9633 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9635 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9636 (armv8-r): Set ARM Cortex-R52 as default CPU.
9637 * config/arm/arm-tables.opt: Regenerate.
9638 * config/arm/arm-tune.md: Regenerate.
9639 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9641 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9642 extension for -mcpu=cortex-r52.
9644 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9646 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9647 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9648 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9649 (fp-armv8): Define it as FP_ARMv8 only.
9650 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9651 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9653 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9654 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9655 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9656 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9657 than TARGET_FPU_ARMV8.
9658 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9659 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9660 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9662 * config/arm/neon.md (neon_vrint): Likewise.
9663 (neon_vcvt): Likewise.
9664 (neon_<fmaxmin_op><mode>): Likewise.
9665 (<fmaxmin><mode>3): Likewise.
9666 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9667 * config/arm/predicates.md (arm_cond_move_operator): Check against
9668 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9670 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9672 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9675 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9677 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9678 loop in comment with memset.
9680 2017-07-14 Martin Liska <mliska@suse.cz>
9682 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9683 * dwarf2out.c (is_java): Remove the function.
9684 (output_pubname): Remove usage of the function.
9685 (lower_bound_default): Remove usage of DW_LANG_Java.
9686 (gen_compile_unit_die): Likewise.
9687 * gcc.c: Remove compiler defaults for .java and .zip files.
9688 * gimple-expr.c (remove_suffix): Change as there's no longer
9689 extension than 4-letter one.
9690 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9691 (gimplify_save_expr): Likewise.
9692 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9693 as it's possible even for other languages than Java.
9694 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9695 * lto-opts.c (lto_write_options): Remove reference to Java.
9696 * opts.c (strip_off_ending): Update file extension handling.
9697 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9698 * tree-eh.c (lower_resx): Likewise.
9699 * tree.c (free_lang_data_in_type): Remove dead code.
9700 (find_decls_types_r): Likewise.
9701 (build_common_builtin_nodes): Remove Java from a comment.
9702 (verify_type): Remove dead code.
9703 * varasm.c (assemble_external): Remove Java from a comment.
9705 2017-07-14 Martin Liska <mliska@suse.cz>
9707 * opts.c (finish_options): Add quotes.
9708 (common_handle_option): Likewise.
9710 2017-07-14 Martin Liska <mliska@suse.cz>
9712 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9713 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9715 * dwarf2out.c (lower_bound_default): Do not handle
9717 (gen_compile_unit_die): Likewise.
9718 * gcc.c: Remove default extension binding for GNU Pascal.
9719 * stmt.c: Remove Pascal language from a comment.
9720 * xcoffout.c: Likewise.
9722 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9725 * diagnostic-show-locus.c (fixit_cmp): New function.
9726 (layout::layout): Sort m_fixit_hints.
9727 (column_range::column_range): Assert that the values are valid.
9728 (struct char_span): New struct.
9729 (correction::overwrite): New method.
9730 (struct source_line): New struct.
9731 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9732 calls in terms of classes source_line and char_span, and
9733 correction::overwrite.
9734 (selftest::test_overlapped_fixit_printing_2): New function.
9735 (selftest::diagnostic_show_locus_c_tests): Call it.
9737 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9739 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9740 early if there is no lhs.
9742 2017-07-13 Martin Liska <mliska@suse.cz>
9744 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9745 (gen_reference_type_die): Likewise.
9746 * stor-layout.c: Remove Pascal-related comment.
9748 2017-07-13 Martin Liska <mliska@suse.cz>
9750 * opts.c (finish_options): Add quotes to error messages.
9751 (parse_sanitizer_options): Likewise.
9753 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9755 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9757 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9759 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9761 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9763 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9765 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9767 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9770 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9771 provides the hardware capability bits, define the macro
9772 __BUILTIN_CPU_SUPPORTS__.
9773 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9774 if GLIBC does not provide the hardware capability bits. Add a
9775 gcc_unreachable call if the built-in cpu function is neither
9776 __builtin_cpu_is nor __builtin_cpu_supports.
9777 (rs6000_get_function_versions_dispatcher): Change the warning
9778 that an old GLIBC is used which does not export the capability
9779 bits to be an error.
9780 * doc/extend.texi (target_clones attribute): Document the
9781 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9782 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9783 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9784 the macros defined by GCC if the newer GLIBC is available.
9786 2017-07-12 Jeff Law <law@redhat.com>
9788 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9789 remaining includes slightly.
9790 * config/riscv/riscv-builtins.c: Include profile-count.h.
9792 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9795 * config/avr/avr.c (avr_set_current_function): In diagnostic
9796 messages: Quote keywords and (parts of) identifiers.
9797 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9800 2017-07-12 Carl Love <cel@us.ibm.com>
9802 * config/rs6000/rs6000-c.c: Add support for built-in functions
9803 vector bool char vec_revb (vector bool char);
9804 vector bool short vec_revb (vector short char);
9805 vector bool int vec_revb (vector bool int);
9806 vector bool long long vec_revb (vector bool long long);
9807 * doc/extend.texi: Update the built-in documentation file for the
9808 new built-in functions.
9810 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9812 * config/s390/s390.md: Remove movcc splitter.
9814 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9816 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9817 load/store on condition.
9819 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9822 * config/avr/avr.c (avr_encode_section_info)
9823 [progmem && !TREE_READONLY]: Error if progmem object needs
9826 2017-07-11 Michael Collison <michael.collison@arm.com>
9828 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9831 2017-07-11 Carl Love <cel@us.ibm.com>
9833 * config/rs6000/rs6000-c.c: Add support for builtins
9834 vector unsigned int vec_parity_lsbb (vector signed int);
9835 vector unsigned int vec_parity_lsbb (vector unsigned int);
9836 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9837 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9838 vector unsigned long long vec_parity_lsbb (vector signed long long);
9839 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9840 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9841 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9842 * doc/extend.texi: Update the built-in documentation file for the
9843 new built-in functions.
9845 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9847 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9848 (layout::m_primary_loc): New field.
9849 (layout::layout): Initialize new field. Move location filtering
9850 logic from here to...
9851 (layout::maybe_add_location_range): ...this new method. Add
9852 support for filtering to just the lines already specified by other
9854 (layout::will_show_line_p): New method.
9855 (gcc_rich_location::add_location_if_nearby): New method.
9856 (selftest::test_add_location_if_nearby): New test function.
9857 (selftest::diagnostic_show_locus_c_tests): Call it.
9858 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9861 2017-07-11 Tom de Vries <tom@codesourcery.com>
9863 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9864 (bb_first_real_insn): New function.
9865 (nvptx_single): Add extra initialization of broadcasted condition
9868 2017-07-11 Nathan Sidwell <nathan@acm.org>
9870 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9872 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9874 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9875 Binutils doc as TEXI will mess them up.
9876 * doc/invoke.texi (AVR Options): Same here.
9878 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9880 * config/sparc/sparc.opt (mfix-ut700): New option.
9881 (mfix-gr712rc): Likewise.
9882 (sparc_fix_b2bst): New variable.
9883 * doc/invoke.texi (SPARC options): Document them.
9884 (ARM options): Fix warnings.
9885 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9886 instructions to prevent sequences that can trigger the store-store
9887 errata for certain LEON3FT processors.
9888 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9889 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9890 * config/sparc/sparc.md (fix_b2bst): New attribute.
9891 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9893 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9896 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9898 (*rsqrtsf2_sse): Ditto.
9900 (div<mode>3): Macroize insn from divdf3 and divsf3
9901 using MODEF mode iterator.
9903 2017-07-10 Martin Sebor <msebor@redhat.com>
9905 PR tree-optimization/80397
9906 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9907 instead of testing for equality to INTEGER_TYPE.
9909 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9911 * config.gcc: Remove uclibc from arc target spec.
9913 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9915 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9917 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9920 * lto-wrapper.c (remove_option): New function.
9921 (merge_and_complain): Merge PIC/PIE options more realistically.
9923 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9925 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9929 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9930 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9931 * config.in: Regenerate.
9932 * configure: Regenerate.
9933 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9934 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9935 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9936 (TARGET_GASISR_PROLOGUES): ...target mask.
9937 * common/config/avr/avr-common.c
9938 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9939 Set -mgas-isr-prologues.
9940 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9941 INSERT_PASS_BEFORE for it.
9942 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9943 * config/avr/avr.c (avr_option_override)
9944 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9945 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9946 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9947 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9948 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9950 (make_avr_pass_pre_proep): New function.
9951 (emit_push_sfr) <treg>: Add argument to function and use it
9953 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9954 and set machine->gasisr.yes.
9955 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9956 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9957 __gcc_isr.n_pushed to .L__stack_usage.
9958 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9959 (avr_asm_final_postscan_insn): ...this new static function.
9960 * config/avr/avr.h (machine_function)
9961 <is_no_gccisr, use_L__stack_usage>: New fields.
9962 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9963 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9964 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9965 (gasisr, *gasisr): New expander and insn.
9966 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9967 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9968 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9970 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9972 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9975 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9977 Move jump-tables out of .text again.
9980 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9981 (ASM_OUTPUT_ADDR_VEC): New function.
9982 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9983 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9984 INSN_ADDRESSes as asm comment.
9985 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9986 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9987 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9988 * config/avr/avr.md (*tablejump): Adjust comment.
9989 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9990 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9992 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9993 (avr_output_addr_vec): New proto.
9994 (avr_log_t) <insn_addresses>: New field.
9996 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9999 * config/i386/i386.c (ix86_function_arg_advance): Set
10000 outgoing_args_on_stack to true if there are outgoing arguments
10002 (ix86_function_arg): Likewise.
10003 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10004 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10005 * config/i386/i386.h (machine_function): Add
10006 outgoing_args_on_stack.
10008 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10010 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10011 supporting pthreds.
10012 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10014 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10016 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10017 (REAL_H): Remove $(MACHMODE_H).
10018 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10020 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10021 $(MACHMODE_H) and double-int.h.
10022 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10024 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10027 2017-07-07 Andrew Pinski <apinski@cavium.com>
10029 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10030 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10032 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10034 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10035 Add warning if GCC was not configured to link against a GLIBC that
10036 exports the hardware capability bits.
10037 (make_resolver_func): Make resolver function private and not a
10038 COMDAT function. Create the name with clone_function_name instead
10039 of make_unique_name.
10042 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10043 correct operand in doing the split.
10045 2017-07-07 Carl Love <cel@us.ibm.com>
10047 * config/rs6000/rs6000-c: Add support for built-in function
10048 vector unsigned short vec_pack_to_short_fp32 (vector float,
10050 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10051 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10052 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10053 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10054 (convert_4f32_8i16): Add define_expand.
10055 * doc/extend.texi: Update the built-in documentation file for the
10056 new built-in function.
10058 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10060 * config/sparc/m8.md: New file.
10061 * config/sparc/sparc.md: Include m8.md.
10063 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10065 * config/sparc/sparc.opt: New option -mvis4b.
10066 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10067 (sparc_option_override): Handle VIS4B.
10068 (enum sparc_builtins): Define
10069 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10070 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10071 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10072 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10073 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10074 (check_constant_argument): New function.
10075 (sparc_vis_init_builtins): Define builtins
10076 __builtin_vis_dictunpack{8,16,32},
10077 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10078 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10079 __builtin_vis_fpcmpde{8,16,32}shl and
10080 __builtin_vis_fpcmpur{8,16,32}shl.
10081 (sparc_expand_builtin): Check that the constant operands to
10082 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
10083 constant and in range.
10084 * config/sparc/sparc-c.c (sparc_target_macros): Handle
10086 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
10087 (SPARC_IMM5_P): Likewise.
10088 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
10089 (enabled): Handle vis4b.
10090 (UNSPEC_DICTUNPACK): New unspec.
10091 (UNSPEC_FPCMPSHL): Likewise.
10092 (UNSPEC_FPUCMPSHL): Likewise.
10093 (UNSPEC_FPCMPDESHL): Likewise.
10094 (UNSPEC_FPCMPURSHL): Likewise.
10095 (cpu_feature): New CPU feature `vis4b'.
10096 (dictunpack{8,16,32}): New insns.
10097 (FPCSMODE): New mode iterator.
10098 (fpcscond): New code iterator.
10099 (fpcsucond): Likewise.
10100 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
10101 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10102 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10103 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10104 * config/sparc/constraints.md: Define constraints `q' for unsigned
10105 2-bit integer constants and `t' for unsigned 5-bit integer
10107 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
10109 (imm5_operand_dictunpack16): Likewise.
10110 (imm5_operand_dictunpack32): Likewise.
10111 (imm2_operand): Likewise.
10112 * doc/invoke.texi (SPARC Options): Document -mvis4b.
10113 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
10114 ditunpack* and fpcmp*shl builtins.
10116 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10118 * config.gcc: Handle m8 in --with-{cpu,tune} options.
10119 * config.in: Add HAVE_AS_SPARC6 define.
10120 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
10122 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
10124 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
10125 (CPP_CPU_SPEC): Handle m8.
10126 (ASM_CPU_SPEC): Likewise.
10127 * config/sparc/sparc-opts.h (enum processor_type): Add
10129 * config/sparc/sparc.c (m8_costs): New struct.
10130 (sparc_option_override): Handle TARGET_CPU_m8.
10131 (sparc32_initialize_trampoline): Likewise.
10132 (sparc64_initialize_trampoline): Likewise.
10133 (sparc_issue_rate): Likewise.
10134 (sparc_register_move_cost): Likewise.
10135 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10136 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10137 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10138 (CPP_CPU_SPEC): Handle M8.
10139 (ASM_CPU_SPEC): Likewise.
10140 (AS_M8_FLAG): Define.
10141 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10142 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10143 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10145 * configure: Regenerate.
10146 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10149 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10151 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10153 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10154 ("*movdi_insn_sp32"): Do not set v3pipe.
10155 ("*movsi_insn"): Likewise.
10156 ("*movdi_insn_sp64"): Likewise.
10157 ("*movsf_insn"): Likewise.
10158 ("*movdf_insn_sp32"): Likewise.
10159 ("*movdf_insn_sp64"): Likewise.
10160 ("*zero_extendsidi2_insn_sp64"): Likewise.
10161 ("*sign_extendsidi2_insn"): Likewise.
10162 ("*mov<VM32:mode>_insn"): Likewise.
10163 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10164 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10165 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10166 ("<vlop:code><VL:mode>3"): Likewise.
10167 ("*not_<vlop:code><VL:mode>3"): Likewise.
10168 ("*nand<VL:mode>_vis"): Likewise.
10169 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10170 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10171 ("one_cmpl<VL:mode>2"): Likewise.
10172 ("faligndata<VM64:mode>_vis"): Likewise.
10173 ("alignaddrsi_vis"): Likewise.
10174 ("alignaddrdi_vis"): Likweise.
10175 ("alignaddrlsi_vis"): Likewise.
10176 ("alignaddrldi_vis"): Likewise.
10177 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10178 ("bmaskdi_vis"): Likewise.
10179 ("bmasksi_vis"): Likewise.
10180 ("bshuffle<VM64:mode>_vis"): Likewise.
10181 ("cmask8<P:mode>_vis"): Likewise.
10182 ("cmask16<P:mode>_vis"): Likewise.
10183 ("cmask32<P:mode>_vis"): Likewise.
10184 ("pdistn<P:mode>_vis"): Likewise.
10185 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10187 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10189 * config/sparc/sparc.md ("subtype"): New insn attribute.
10190 ("*wrgsr_sp64"): Set insn subtype.
10191 ("*rdgsr_sp64"): Likewise.
10192 ("alignaddrsi_vis"): Likewise.
10193 ("alignaddrdi_vis"): Likewise.
10194 ("alignaddrlsi_vis"): Likewise.
10195 ("alignaddrldi_vis"): Likewise.
10196 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10197 ("fexpand_vis"): Likewise.
10198 ("fpmerge_vis"): Likewise.
10199 ("faligndata<VM64:mode>_vis"): Likewise.
10200 ("bshuffle<VM64:mode>_vis"): Likewise.
10201 ("cmask8<P:mode>_vis"): Likewise.
10202 ("cmask16<P:mode>_vis"): Likewise.
10203 ("cmask32<P:mode>_vis"): Likewise.
10204 ("fchksm16_vis"): Likewise.
10205 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10206 ("fmean16_vis"): Likewise.
10207 ("fp<plusminus_insn>64_vis"): Likewise.
10208 ("<plusminus_insn>v8qi3"): Likewise.
10209 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10210 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10211 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10212 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10213 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10214 ("*movqi_insn"): Likewise.
10215 ("*movhi_insn"): Likewise.
10216 ("*movsi_insn"): Likewise.
10217 ("movsi_pic_gotdata_op"): Likewise.
10218 ("*movdi_insn_sp32"): Likewise.
10219 ("*movdi_insn_sp64"): Likewise.
10220 ("movdi_pic_gotdata_op"): Likewise.
10221 ("*movsf_insn"): Likewise.
10222 ("*movdf_insn_sp32"): Likewise.
10223 ("*movdf_insn_sp64"): Likewise.
10224 ("*zero_extendhisi2_insn"): Likewise.
10225 ("*zero_extendqihi2_insn"): Likewise.
10226 ("*zero_extendqisi2_insn"): Likewise.
10227 ("*zero_extendqidi2_insn"): Likewise.
10228 ("*zero_extendhidi2_insn"): Likewise.
10229 ("*zero_extendsidi2_insn_sp64"): Likewise.
10230 ("ldfsr"): Likewise.
10231 ("prefetch_64"): Likewise.
10232 ("prefetch_32"): Likewise.
10233 ("tie_ld32"): Likewise.
10234 ("tie_ld64"): Likewise.
10235 ("*tldo_ldub_sp32"): Likewise.
10236 ("*tldo_ldub1_sp32"): Likewise.
10237 ("*tldo_ldub2_sp32"): Likewise.
10238 ("*tldo_ldub_sp64"): Likewise.
10239 ("*tldo_ldub1_sp64"): Likewise.
10240 ("*tldo_ldub2_sp64"): Likewise.
10241 ("*tldo_ldub3_sp64"): Likewise.
10242 ("*tldo_lduh_sp32"): Likewise.
10243 ("*tldo_lduh1_sp32"): Likewise.
10244 ("*tldo_lduh_sp64"): Likewise.
10245 ("*tldo_lduh1_sp64"): Likewise.
10246 ("*tldo_lduh2_sp64"): Likewise.
10247 ("*tldo_lduw_sp32"): Likewise.
10248 ("*tldo_lduw_sp64"): Likewise.
10249 ("*tldo_lduw1_sp64"): Likewise.
10250 ("*tldo_ldx_sp64"): Likewise.
10251 ("*mov<VM32:mode>_insn"): Likewise.
10252 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10253 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10255 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10257 * config/sparc/sparc.md ("type"): New insn type viscmp.
10258 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10260 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10261 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10262 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10263 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10265 ("n7_vis_logical_11cycle"): Likewise.
10266 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10267 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10268 * config/sparc/niagara.md ("niag_vis"): Likewise.
10269 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10270 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10272 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10274 * config/sparc/sparc.md: New instruction type `bmask'.
10275 (bmaskdi_vis): Use the `bmask' type.
10276 (bmasksi_vis): Likewise.
10277 * config/sparc/ultra3.md (us3_array): Likewise.
10278 * config/sparc/niagara7.md (n7_array): Likewise.
10279 * config/sparc/niagara4.md (n4_array): Likewise.
10280 * config/sparc/niagara2.md (niag2_vis): Likewise.
10281 (niag3_vis): Likewise.
10282 * config/sparc/niagara.md (niag_vis): Likewise.
10284 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10286 * ipa-comdats.c: Remove optimize check from gate.
10287 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10288 for functions not optimized.
10289 (ipa_fn_summary_read): Skip optimize check.
10290 (ipa_fn_summary_write): Likewise.
10291 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10293 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10295 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10296 (check_callers): Give up on caller which is not optimized.
10297 (inline_small_functions): Likewise.
10298 (ipa_inline): Do not give up when not optimizing.
10299 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10300 away unoptimizes cdtors.
10301 (whole_program_function_and_variable_visibility): Do
10302 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10303 * ipa.c (process_references): Do not check optimize.
10304 (symbol_table::remove_unreachable_nodes): Update optimize check.
10305 (set_writeonly_bit): Update optimize check.
10306 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10307 (pass_ipa_single_use::gate): Remove.
10309 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10311 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10312 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10313 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10314 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10315 permute_load, permute_store, adjust_extract, adjust_splat,
10316 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10317 replace_swap_with_copy, dump_swap_insn_table,
10318 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10319 recombine_lvx_pattern, recombine_stvx_pattern,
10320 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10321 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10322 to file rs6000-p8swap.c.
10323 * config/rs6000/rs6000-p8swap.c: New file.
10324 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10325 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10326 and rs6000*-*-* targets.
10328 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10330 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10332 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10334 * lto-wrapper.c (merge_and_complain): Do not merge
10335 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10336 fsigned_zeros, ftrapping_math, fwrapv.
10337 (append_compiler_options): Do not track these options.
10338 (append_linker_options): Likewie
10340 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10342 * cgraphunit.c (cgraph_node::finalize_function): When
10343 !flag_toplevel_reorde set no_reorder flag.
10344 (varpool_node::finalize_decl): Likewise.
10345 (symbol_table::compile): Drop no toplevel reorder path.
10347 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10349 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10350 edges; zero probability is not better than uninitialized.
10352 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10354 * asan.h (asan_sanitize_allocas_p): Declare.
10355 * asan.c (asan_sanitize_allocas_p): New function.
10356 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10357 (handle_builtin_alloca): Likewise.
10358 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10359 if !asan_sanitize_allocas_p.
10360 * params.def (asan-instrument-allocas): Add new option.
10361 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10362 * opts.c (common_handle_option): Disable allocas sanitization for
10365 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10367 * asan.c: Include gimple-fold.h.
10368 (get_last_alloca_addr): New function.
10369 (handle_builtin_stackrestore): Likewise.
10370 (handle_builtin_alloca): Likewise.
10371 (asan_emit_allocas_unpoison): Likewise.
10372 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10373 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10374 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10375 (instrument_builtin_call): Pass gimple iterator to
10376 get_mem_refs_of_builtin_call.
10377 (last_alloca_addr): New global.
10378 * asan.h (asan_emit_allocas_unpoison): Declare.
10379 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10380 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10381 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10382 if function calls alloca.
10383 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10384 * gimple-fold.h (replace_call_with_value): Declare.
10385 * internal-fn.c: Include asan.h.
10386 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10387 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10389 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10391 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10392 (C_SELFTEST_FLAGS): New.
10393 (CPP_SELFTEST_FLAGS): New.
10394 (SELFTEST_DEPS): New, from deps of s-selftest.
10395 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10396 (CPP_SELFTEST_DEPS): New.
10397 (selftest): Add dependency on s-selftest-c++.
10398 (s-selftest): Rename to...
10399 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10400 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10401 than SELFTEST_FLAGS.
10402 (selftest-gdb): Rename to...
10403 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10405 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10406 (selftest-valgrind): Rename to...
10407 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10409 (selftest-valgrind): Reintroduce as an alias for
10410 selftest-c-valgrind.
10411 (s-selftest-c++): New.
10412 (selftest-c++-gdb): New.
10413 (selftest-c++-valgrind): New.
10415 2017-07-06 Olivier Hainque <hainque@adacore.com>
10417 * gcc.c (process_command): When deciding if undefined variables
10418 should be ignored when processing specs, accept "gcc -v" as well.
10420 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10422 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10423 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10425 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10427 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10428 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10429 * config/arm/arm-tables.opt: Regenerate.
10430 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10432 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10434 2017-07-06 Carl Love <cel@us.ibm.com>
10436 * ChangeLog: Clean up from mid air collision
10438 2017-07-06 Carl Love <cel@us.ibm.com>
10440 * config/rs6000/rs6000-c.c: Add support for built-in functions
10441 vector signed int vec_subc (vector signed int, vector signed int);
10442 vector signed __int128 vec_subc (vector signed __int128,
10443 vector signed __int128);
10444 vector unsigned __int128 vec_subc (vector unsigned __int128,
10445 vector unsigned __int128);
10446 vector signed int vec_sube (vector signed int, vector signed int,
10447 vector signed int);
10448 vector unsigned int vec_sube (vector unsigned int,
10449 vector unsigned int,
10450 vector unsigned int);
10451 vector signed __int128 vec_sube (vector signed __int128,
10452 vector signed __int128,
10453 vector signed__int128);
10454 vector unsigned __int128 vec_sube (vector unsigned __int128,
10455 vector unsigned __int128,
10456 vector unsigned __int128);
10457 vector signed int vec_subec (vector signed int, vector signed int,
10458 vector signed int);
10459 vector unsigned int vec_subec (vector unsigned int,
10460 vector unsigned int,
10461 vector unsigned int);
10462 vector signed __int128 vec_subec (vector signed __int128,
10463 vector signed __int128,
10464 vector signed__int128);
10465 vector unsigned __int128 vec_subec (vector unsigned __int128,
10466 vector unsigned __int128,
10467 vector unsigned __int128);
10468 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10469 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10470 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10471 BU_ALTIVEC_OVERLOAD_X definitions.
10472 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10473 * doc/extend.texi: Update the built-in documentation file for the new
10474 built-in functions.
10476 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10479 * diagnostic-show-locus.c (layout::layout): Use start and finish
10480 spelling location for the start and finish of each range.
10481 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10482 Add unused aspect param.
10483 * input.c (expand_location_1): Add "aspect" param, and use it
10484 to access the correct part of the location.
10485 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10487 (expand_location_to_spelling_point): Likewise.
10488 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10489 param, and pass it to expand_location_1.
10491 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10493 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10494 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10495 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10496 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10497 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10498 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10499 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10500 _mm_maskz_getmant_ss): New intrinsics.
10501 (__builtin_ia32_getexpss128_mask): Changed to ...
10502 __builtin_ia32_getexpss128_round ... this.
10503 (__builtin_ia32_getexpsd128_mask): Changed to ...
10504 __builtin_ia32_getexpsd128_round ... this.
10505 * config/i386/i386-builtin-types.def
10506 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10507 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10508 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10509 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10510 __builtin_ia32_getmantss_mask_round): New builtins.
10511 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10512 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10513 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10514 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10515 * config/i386/sse.md
10516 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10517 avx512f_sgetexp<mode><mask_scalar_name>
10518 <round_saeonly_scalar_name> ... this.
10519 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10520 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10521 vgetexp<ssescalarmodesuffix>
10522 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10523 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10524 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10525 avx512f_vgetmant<mode><mask_scalar_name>
10526 <round_saeonly_scalar_name> ... this.
10527 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10528 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10529 vgetmant<ssescalarmodesuffix>
10530 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10531 %0<mask_scalar_operand4>, %1, %2
10532 <round_saeonly_scalar_mask_op4>, %3} ... this.
10533 * config/i386/subst.md (mask_scalar_operand4,
10534 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10535 round_saeonly_scalar_nimm_predicate): New subst attributes.
10537 2017-07-06 Julia Koval <julia.koval@intel.com>
10539 * config/i386/i386.c (ix86_erase_embedded_rounding):
10540 Remove code for old rounding pattern.
10542 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10544 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10546 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10548 * doc/sourcebuild.texi (Test Directives, Variants of
10549 dg-require-support): Add documentation for dg-require-stack-check.
10551 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10553 * config/i386/subst.md (mask_scalar, round_scalar,
10554 round_saeonly_scalar): New meta-templates.
10555 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10556 round_scalar_mask_operand3, round_scalar_mask_op3,
10557 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10558 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10559 round_saeonly_scalar_constraint,
10560 round_saeonly_scalar_prefix): New subst attribute.
10561 * config/i386/sse.md
10562 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10563 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10564 <round_scalar_name> ... this.
10565 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10566 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10567 <round_scalar_name> ... this.
10568 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10569 <sse>_vm<code><mode>3<mask_scalar_name>
10570 <round_saeonly_scalar_name> ... this.
10571 (v<plusminus_mnemonic><ssescalarmodesuffix>
10572 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10573 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10574 v<plusminus_mnemonic><ssescalarmodesuffix>
10575 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10576 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10577 (v<multdiv_mnemonic><ssescalarmodesuffix>
10578 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10579 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10580 v<multdiv_mnemonic><ssescalarmodesuffix>
10581 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10582 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10583 (v<maxmin_float><ssescalarmodesuffix>
10584 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10585 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10586 v<maxmin_float><ssescalarmodesuffix>
10587 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10588 %0<mask_scalar_operand3>, %1, %<iptr>2
10589 <round_saeonly_scalar_mask_op3>} ... this.
10591 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10593 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10594 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10596 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10597 Alan Hayward <alan.hayward@arm.com>
10598 David Sherwood <david.sherwood@arm.com>
10600 * combine.c (simplify_if_then_else): Remove "enum" before
10602 * compare-elim.c (can_eliminate_compare): Likewise.
10603 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10605 (aarch64_lookup_simd_builtin_type): Likewise.
10606 (aarch64_simd_builtin_type): Likewise.
10607 (aarch64_init_simd_builtin_types): Likewise.
10608 (aarch64_simd_expand_args): Likewise.
10609 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10611 (aarch64_reverse_mask): Likewise.
10612 (aarch64_simd_emit_reg_reg_move): Likewise.
10613 (aarch64_gen_adjusted_ldpstp): Likewise.
10614 (aarch64_ccmp_mode_to_code): Likewise.
10615 (aarch64_operands_ok_for_ldpstp): Likewise.
10616 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10617 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10619 (aarch64_min_divisions_for_recip_mul): Likewise.
10620 (aarch64_reassociation_width): Likewise.
10621 (aarch64_get_condition_code_1): Likewise.
10622 (aarch64_simd_emit_reg_reg_move): Likewise.
10623 (aarch64_simd_attr_length_rglist): Likewise.
10624 (aarch64_reverse_mask): Likewise.
10625 (aarch64_operands_ok_for_ldpstp): Likewise.
10626 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10627 (aarch64_gen_adjusted_ldpstp): Likewise.
10628 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10630 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10631 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10632 (arm_lookup_simd_builtin_type): Likewise.
10633 (arm_simd_builtin_type): Likewise.
10634 (arm_init_simd_builtin_types): Likewise.
10635 (arm_expand_builtin_args): Likewise.
10636 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10637 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10638 (ft32_setup_incoming_varargs): Likewise.
10639 (ft32_function_arg): Likewise.
10640 (ft32_function_arg_advance): Likewise.
10641 (ft32_pass_by_reference): Likewise.
10642 (ft32_arg_partial_bytes): Likewise.
10643 (ft32_valid_pointer_mode): Likewise.
10644 (ft32_addr_space_pointer_mode): Likewise.
10645 (ft32_addr_space_legitimate_address_p): Likewise.
10646 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10648 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10649 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10650 (iamcu_alignment): Likewise.
10651 (canonicalize_vector_int_perm): Likewise.
10652 (ix86_noce_conversion_profitable_p): Likewise.
10653 (ix86_mpx_bound_mode): Likewise.
10654 (ix86_operands_ok_for_move_multiple): Likewise.
10655 * config/microblaze/microblaze-protos.h
10656 (microblaze_expand_conditional_branch_reg): Likewise.
10657 * config/microblaze/microblaze.c
10658 (microblaze_expand_conditional_branch_reg): Likewise.
10659 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10661 (rs6000_reassociation_width): Likewise.
10662 (rs6000_invalid_binary_op): Likewise.
10663 (fusion_p9_p): Likewise.
10664 (emit_fusion_p9_load): Likewise.
10665 (emit_fusion_p9_store): Likewise.
10666 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10668 (riscv_hard_regno_mode_ok_p): Likewise.
10669 (riscv_address_insns): Likewise.
10670 (riscv_split_symbol): Likewise.
10671 (riscv_legitimize_move): Likewise.
10672 (riscv_function_value): Likewise.
10673 (riscv_hard_regno_nregs): Likewise.
10674 (riscv_expand_builtin): Likewise.
10675 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10676 (riscv_build_integer): Likewise.
10677 (riscv_split_integer): Likewise.
10678 (riscv_legitimate_constant_p): Likewise.
10679 (riscv_cannot_force_const_mem): Likewise.
10680 (riscv_regno_mode_ok_for_base_p): Likewise.
10681 (riscv_valid_base_register_p): Likewise.
10682 (riscv_valid_offset_p): Likewise.
10683 (riscv_valid_lo_sum_p): Likewise.
10684 (riscv_classify_address): Likewise.
10685 (riscv_legitimate_address_p): Likewise.
10686 (riscv_address_insns): Likewise.
10687 (riscv_load_store_insns): Likewise.
10688 (riscv_force_binary): Likewise.
10689 (riscv_split_symbol): Likewise.
10690 (riscv_force_address): Likewise.
10691 (riscv_legitimize_address): Likewise.
10692 (riscv_move_integer): Likewise.
10693 (riscv_legitimize_const_move): Likewise.
10694 (riscv_legitimize_move): Likewise.
10695 (riscv_address_cost): Likewise.
10696 (riscv_subword): Likewise.
10697 (riscv_output_move): Likewise.
10698 (riscv_canonicalize_int_order_test): Likewise.
10699 (riscv_emit_int_order_test): Likewise.
10700 (riscv_function_arg_boundary): Likewise.
10701 (riscv_pass_mode_in_fpr_p): Likewise.
10702 (riscv_pass_fpr_single): Likewise.
10703 (riscv_pass_fpr_pair): Likewise.
10704 (riscv_get_arg_info): Likewise.
10705 (riscv_function_arg): Likewise.
10706 (riscv_function_arg_advance): Likewise.
10707 (riscv_arg_partial_bytes): Likewise.
10708 (riscv_function_value): Likewise.
10709 (riscv_pass_by_reference): Likewise.
10710 (riscv_setup_incoming_varargs): Likewise.
10711 (riscv_print_operand): Likewise.
10712 (riscv_elf_select_rtx_section): Likewise.
10713 (riscv_save_restore_reg): Likewise.
10714 (riscv_for_each_saved_reg): Likewise.
10715 (riscv_register_move_cost): Likewise.
10716 (riscv_hard_regno_mode_ok_p): Likewise.
10717 (riscv_hard_regno_nregs): Likewise.
10718 (riscv_class_max_nregs): Likewise.
10719 (riscv_memory_move_cost): Likewise.
10720 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10721 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10722 (rl78_addr_space_address_mode): Likewise.
10723 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10725 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10726 (rs6000_reassociation_width): Likewise.
10727 (rs6000_invalid_binary_op): Likewise.
10728 (fusion_p9_p): Likewise.
10729 (emit_fusion_p9_load): Likewise.
10730 (emit_fusion_p9_store): Likewise.
10731 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10732 (ok_for_simple_move_operands): Likewise.
10733 (ok_for_simple_move_strict_operands): Likewise.
10734 (ok_for_simple_arith_logic_operands): Likewise.
10735 (visium_legitimize_reload_address): Likewise.
10736 (visium_select_cc_mode): Likewise.
10737 (output_cbranch): Likewise.
10738 (visium_split_double_move): Likewise.
10739 (visium_expand_copysign): Likewise.
10740 (visium_expand_int_cstore): Likewise.
10741 (visium_expand_fp_cstore): Likewise.
10742 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10743 (visium_function_arg): Likewise.
10744 (visium_function_arg_advance): Likewise.
10745 (visium_libcall_value): Likewise.
10746 (visium_setup_incoming_varargs): Likewise.
10747 (visium_legitimate_constant_p): Likewise.
10748 (visium_legitimate_address_p): Likewise.
10749 (visium_legitimize_address): Likewise.
10750 (visium_secondary_reload): Likewise.
10751 (visium_register_move_cost): Likewise.
10752 (visium_memory_move_cost): Likewise.
10753 (prepare_move_operands): Likewise.
10754 (ok_for_simple_move_operands): Likewise.
10755 (ok_for_simple_move_strict_operands): Likewise.
10756 (ok_for_simple_arith_logic_operands): Likewise.
10757 (visium_function_value_1): Likewise.
10758 (rtx_ok_for_offset_p): Likewise.
10759 (visium_legitimize_reload_address): Likewise.
10760 (visium_split_double_move): Likewise.
10761 (visium_expand_copysign): Likewise.
10762 (visium_expand_int_cstore): Likewise.
10763 (visium_expand_fp_cstore): Likewise.
10764 (visium_split_cstore): Likewise.
10765 (visium_select_cc_mode): Likewise.
10766 (visium_split_cbranch): Likewise.
10767 (output_cbranch): Likewise.
10768 (visium_print_operand_address): Likewise.
10769 * expmed.c (flip_storage_order): Likewise.
10770 * expmed.h (emit_cstore): Likewise.
10771 (flip_storage_order): Likewise.
10772 * genrecog.c (validate_pattern): Likewise.
10773 * hsa-gen.c (gen_hsa_addr): Likewise.
10774 * internal-fn.c (expand_arith_overflow): Likewise.
10775 * ira-color.c (allocno_copy_cost_saving): Likewise.
10776 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10777 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10778 (process_invariant_for_inheritance): Likewise.
10779 * lra-eliminations.c (move_plus_up): Likewise.
10780 * omp-low.c (lower_oacc_reductions): Likewise.
10781 * simplify-rtx.c (simplify_subreg): Likewise.
10782 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10783 (TARGET_CHKP_BOUND_MODE): Likewise..
10784 * targhooks.c (default_chkp_bound_mode): Likewise.
10785 (default_setup_incoming_vararg_bounds): Likewise.
10786 * targhooks.h (default_chkp_bound_mode): Likewise.
10787 (default_setup_incoming_vararg_bounds): Likewise.
10788 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10789 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10790 (have_whole_vector_shift): Likewise.
10791 * tree-vect-stmts.c (vectorizable_load): Likewise.
10792 * doc/tm.texi: Regenerate.
10794 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10796 Graceful degrade if Binutils PR21472 is not available.
10799 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10800 .rodata in flash test fails.
10801 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10802 * confgure: Regenerate.
10803 * config.in: Regenerate.
10804 * config/avr/avr.c (avr_asm_named_section)
10805 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10806 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10807 (avr_asm_init_sections): Same.
10809 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10811 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10812 (fma<VH:mode>4_intrinsic): Likewise.
10813 (*fmsub<VCVTF:mode>4): Likewise.
10814 (*fmsub<VH:mode>4_intrinsic): Likewise.
10816 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10819 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10820 Don't depend on "optimize > 0".
10821 (out_movhi_r_mr, out_movqi_mr_r): Same.
10822 (out_movhi_mr_r, out_movqi_r_mr): Same.
10823 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10824 io_address_operand on "optimize > 0".
10826 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10828 * tree-loop-distribution.c: Add general explanantion on the pass.
10829 (generate_loops_for_partition): Mark distributed loop.
10830 (pg_add_dependence_edges): New parameter. Handle alias data
10831 dependence specially and record it in the parameter if asked.
10832 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10833 (init_partition_graph_vertices, add_partition_graph_edge): New.
10834 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10835 (free_partition_graph_vdata, build_partition_graph): New.
10836 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10837 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10838 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10839 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10840 (version_for_distribution_p, finalize_partitions): New.
10841 (distribute_loop): Handle alias data dependence specially. Factor
10842 out loop fusion code as functions and call these functions.
10844 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10846 * tree-loop-distribution.c (classify_partition): New parameter and
10847 better handle reduction statement.
10848 (rdg_build_partitions): Revise comment.
10849 (distribute_loop): Compute statements in all partitions and pass it
10850 to classify_partition.
10852 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10854 * tree-loop-distribution.c (enum partition_type): New.
10855 (struct partition): New field type.
10856 (partition_merge_into): Add parameter. Update partition type.
10857 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10858 (build_rdg_partition_for_vertex): Compute partition type.
10859 (rdg_build_partitions): Dump partition type.
10860 (distribute_loop): Update calls to partition_merge_into.
10862 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10864 * tree-loop-distribution.c (struct ddr_hasher): New.
10865 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10867 (classify_partition): Call get_data_dependence.
10868 (pg_add_dependence_edges): Ditto.
10869 (distribute_loop): Release data dependence hash table.
10871 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10873 * tree-loop-distribution.c (ref_base_address): Delete.
10874 (similar_memory_accesses): Rename ...
10875 (share_memory_accesses): ... to this. Check if partitions access
10876 the same memory reference.
10877 (distribute_loop): Call share_memory_accesses.
10879 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10881 * tree-loop-distribution.c (struct partition): New field recording
10882 its data reference.
10883 (partition_alloc, partition_free): Init and release data refs.
10884 (partition_merge_into): Merge data refs.
10885 (build_rdg_partition_for_vertex): Collect data refs for partition.
10886 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10888 (distribute_loop): Remve data refs from vertice data of partition
10891 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10893 * tree-loop-distribution.c (params.h): Include header file.
10894 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10895 (datarefs_vec): New global var.
10896 (create_rdg_vertices): Use datarefs_vec directly.
10897 (free_rdg): Don't free data references.
10898 (build_rdg): Update use. Don't free data references.
10899 (distribute_loop): Compute global variable for data references.
10900 Bail out if there are too many data references.
10902 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10904 * tree-loop-distribution.c (loop_nest): New global var.
10905 (build_rdg): Use loop directly, rather than loop nest.
10906 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10908 (distribute_loop): Compute global variable loop nest. Update use.
10910 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10912 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10913 (partition_merge_into): New parameter. Dump reason for fusion.
10914 (distribute_loop): Update use of partition_merge_into.
10916 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10918 * tree-loop-distribution.c (bb_top_order_index): New.
10919 (bb_top_order_index_size, bb_top_order_cmp): New.
10920 (stmts_from_loop): Use topological order.
10921 (pass_loop_distribution::execute): Compute and release topological
10922 order for basic blocks.
10924 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10926 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10929 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10931 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10932 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10933 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10934 * internal-fn.def (LOOP_DIST_ALIAS): New.
10935 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10936 (fold_loop_internal_call): ... this.
10937 (vect_loop_dist_alias_call): New function.
10938 (set_uid_loop_bbs): Call fold_loop_internal_call.
10939 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10942 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10945 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10946 Require dead FLAGS_REG at the beginning of a peephole.
10948 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10951 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10952 arguments in the call to __builtin_ia32_sbb_u32.
10953 (_subborrow_u64): Swap _X and _Y arguments in the call to
10954 __builtin_ia32_sbb_u64.
10956 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10959 * tree-vrp.c (compare_assert_loc): Turn into a function template
10960 with stable template parameter. Only test if a->e is NULL,
10961 !a->e == !b->e has been verified already. Use e == NULL or
10962 e != NULL instead of e or ! e tests. If stable is true, don't use
10963 iterative_hash_expr, on the other side allow a or b or both NULL
10964 and sort the NULLs last.
10965 (process_assert_insertions): Sort using compare_assert_loc<false>
10966 instead of compare_assert_loc, later sort using
10967 compare_assert_loc<true> before calling process_assert_insertions_for
10968 in a loop. Use break instead of continue once seen NULL pointer.
10970 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10972 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10973 Cortex-R7 and Cortex-R8 processors.
10975 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10977 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10978 uninitialized while src is not.
10980 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10982 * common/config/arm/arm-common.c: Adjust include path for
10984 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10985 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10986 (arm-cpu-data.h): Likewise.
10987 (arm-cpu-cdata.h): Likewise.
10988 * config/arm/arm-cpu.h: Delete.
10989 * config/arm/arm-cpu-cdata.h: Delete.
10990 * config/arm/arm-cpu-data.h: Delete.
10992 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10994 * config/arm/arm-cpus.in (cortex-a55): New.
10995 (cortex-a75): Likewise.
10996 (cortex-a75.cortex-a55): Likewise.
10997 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10999 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11000 * config/arm/arm-cpu-cdata.h: Regenerate.
11001 * config/arm/arm-cpu-data.h: Regenerate.
11002 * config/arm/arm-cpu.h: Regenerate.
11003 * config/arm/arm-tables.opt: Regenerate.
11004 * config/arm/arm-tune.md: Regenerate.
11006 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11008 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11010 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11012 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11015 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11017 PR tree-optimization/81292
11018 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11019 full_string_p, also call adjust_related_strinfos if the adjustment
11020 is simple, otherwise invalidate related strinfos.
11022 2017-07-04 Martin Liska <mliska@suse.cz>
11025 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11026 newly created variable as DECL_IGNORED_P.
11028 2017-07-04 Martin Liska <mliska@suse.cz>
11031 * ipa-inline.c (inline_small_functions):
11032 Use xstrdup_for_dump.
11034 2017-07-04 Tom de Vries <tom@codesourcery.com>
11036 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11038 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11041 * config/darwin.c (darwin_function_switched_text_sections):
11044 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11046 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11048 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11050 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11052 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11054 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11055 min_profitable_iters, and th as inclusive lower bounds.
11056 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11057 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11058 for min_profitable_iters and min_profitable_estimate.
11059 (vect_transform_loop): Treat th as an inclusive lower bound.
11060 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11062 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11065 * config/darwin.c (darwin_function_switched_text_sections):
11066 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11067 in two pieces, and suppress the use of buf.
11069 2017-07-03 Nathan Sidwell <nathan@acm.org>
11071 * hash-table.h (hash_table_mod1): Fix indentation.
11073 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11075 PR middle-end/81290
11076 * predict.c (force_edge_cold): Be more careful about propagation
11078 * profile-count.h (profile_probability::guessed,
11079 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11081 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
11083 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
11085 * doc/invoke.texi (rcpc architecture extension): Document it.
11087 2017-07-03 Richard Biener <rguenther@suse.de>
11089 PR tree-optimization/60510
11090 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
11091 the scalar reduction PHI and use it.
11092 (vectorizable_reduction): Properly guard the single_defuse_cycle
11093 path for non-SLP reduction chains where we cannot use it.
11094 Rework reduc_def/index and vector type deduction. Rework
11095 vector operand gathering during reduction op code-gen.
11096 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
11097 chains dissolve the chain and leave it to non-SLP reduction
11100 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11102 * tree-data-ref.h (dr_alignment): Declare.
11103 * tree-data-ref.c (dr_alignment): New function.
11104 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11105 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11107 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
11109 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11111 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
11112 and base_misalignment fields.
11113 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
11114 * tree-data-ref.c: Include builtins.h.
11115 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
11116 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
11117 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
11118 * tree-vect-data-refs.c: Include tree-cfg.h.
11119 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
11120 fields instead of calculating an alignment here.
11121 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
11122 innermost_loop_behavior fields.
11124 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11126 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
11128 (DR_STEP_ALIGNMENT): New macro.
11129 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11130 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11131 (create_data_ref): Print it.
11132 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11133 to tell whether the step preserves vector (mis)alignment.
11134 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11135 Move the check for an integer step and generalise to all INTEGER_CST.
11136 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11137 Print the outer step alignment.
11139 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11141 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11142 with offset_alignment.
11143 (DR_ALIGNED_TO): Delete.
11144 (DR_OFFSET_ALIGNMENT): New macro.
11145 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11146 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11147 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11148 (create_data_ref): Likewise.
11149 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11150 (vect_analyze_data_refs): Likewise.
11151 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11152 creating dummy innermost behavior.
11154 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11156 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11157 with a "innermost_loop_behavior *" and refeence tree.
11158 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11159 (create_data_ref): Update call accordingly.
11160 * tree-predcom.c (find_looparound_phi): Likewise.
11162 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11164 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11165 fields with dr_wrt_vec_loop.
11166 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11167 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11168 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11169 (vect_dr_behavior): New function.
11170 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11171 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11172 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11173 track whether the step preserves the misalignment.
11174 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11175 Use vect_dr_behavior.
11176 (vect_setup_realignment): Update call accordingly.
11177 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11178 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11179 call to vect_create_addr_base_for_vector_ref.
11180 (vect_create_cond_for_align_checks): Likewise.
11181 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11182 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11183 (vect_recog_mask_conversion_pattern): Likewise.
11184 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11185 (new_stmt_vec_info): Remove redundant zeroing.
11187 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11189 * common/config/arm/arm-common.c (arm_be8_option): New function.
11190 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11191 (ISA_ARMv6): Add isa_bit_be8.
11192 * config/arm/arm.h (arm_be8_option): Add prototype.
11193 (BE8_SPEC_FUNCTION): New define.
11194 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11195 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11196 (mlittle-endian): Similarly.
11197 (mbe8, mbe32): New options.
11198 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11199 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11201 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11203 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11205 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11207 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11208 (cleanup_tree_cfg_bb): Use it.
11209 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11211 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11213 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11216 * loop-doloop.c (add_test): Update profile.
11218 2017-07-03 Martin Liska <mliska@suse.cz>
11221 * sanopt.c (rewrite_usage_of_param): New function.
11222 (sanitize_rewrite_addressable_params): Likewise.
11223 (pass_sanopt::execute): Call rewrite_usage_of_param.
11225 2017-07-03 Richard Biener <rguenther@suse.de>
11227 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11228 back to using VIEW_CONVERT_EXPR.
11230 2017-07-03 Martin Liska <mliska@suse.cz>
11233 * doc/extend.texi: Document when a resolver function is
11234 generated for target_clones.
11236 2017-07-03 Martin Liska <mliska@suse.cz>
11238 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11239 and shadow memory of auto variables which are subject of
11240 use-after-scope sanitization.
11241 (asan_expand_mark_ifn): Add do set only when is_poison.
11243 2016-07-03 Richard Biener <rguenther@suse.de>
11245 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11247 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11248 (vectorizable_reduction): Perform reduction PHI creation when
11249 visiting a reduction PHI and adjust and simplify code generation
11250 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11251 (vect_transform_loop): Visit reduction PHIs.
11252 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11253 defs into the SLP tree.
11254 (vect_build_slp_tree): Reduction defs terminate the recursion.
11255 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11257 (vect_get_vec_defs_for_stmt_copy): Export.
11258 (vect_get_vec_defs): Likewise.
11259 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11261 (vect_get_vec_defs_for_stmt_copy): Declare.
11262 (vect_get_vec_defs): Likewise.
11264 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11266 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11267 parameter with a "loop" parameter and use it instead of the
11268 loop containing DR_STMT. Don't check simple_iv when doing
11269 BB analysis. Describe the two analysis modes in the comment.
11271 2017-07-03 Tom de Vries <tom@codesourcery.com>
11273 PR tree-optimization/69468
11274 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11275 (find_same_succ_bb): Handle ignore_edge_flags.
11277 2017-07-03 Tom de Vries <tom@codesourcery.com>
11279 PR tree-optimization/81192
11280 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11282 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11284 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11286 2017-07-03 Tom de Vries <tom@codesourcery.com>
11288 PR tree-optimization/81192
11289 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11290 BB_SAME_SUCC (bb) == NULL.
11292 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11294 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11297 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11299 * dumpfile.c: Include profile-count.h
11300 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11302 (insert_cond_bb): Update profile.
11303 * tree-cfg.h (insert_cond_bb): Update prototype.
11304 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11305 * tree-dump.c: Do not include tree-cfg.
11307 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11309 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11311 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11313 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11316 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11318 * tree-complex.c (expand_complex_div_wide): update profile.
11320 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11321 Alan Hayward <alan.hayward@arm.com>
11322 David Sherwood <david.sherwood@arm.com>
11324 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11325 (CORETYPES_H): New define.
11326 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11327 (insn-modes-inline.h, s-modes-inline-h): New rules.
11328 (generated_files): Add insn-modes-inline.h.
11329 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11330 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11331 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11332 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11333 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11334 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11335 (build/gencodes.o, build/genconditions.o): Likewise.
11336 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11337 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11338 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11339 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11340 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11341 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11342 * coretypes.h: Include everything up to real.h for generators.
11343 Include insn-modes.h first. Include wide-int-print.h after
11344 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11345 * machmode.h: Don't include insn-modes.h here.
11346 * function-tests.c: Remove includes of signop.h, machmode.h,
11347 double-int.h and wide-int.h.
11349 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11351 * optc-save-gen.awk: Likewise.
11352 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11353 * godump.c: Remove include of wide-int-print.h.
11354 * pretty-print.h: Likewise.
11355 * wide-int-print.cc: Likewise.
11356 * wide-int.cc: Likewise.
11357 * hash-map-tests.c: Remove include of signop.h.
11358 * hash-set-tests.c: Likewise.
11359 * rtl-tests.c: Likewise.
11360 * mkconfig.sh: Remove include of machmode.h.
11361 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11363 (emit_insn_modes_inline_h): ...this new function. Emit the code
11364 into an insn-modes-inline.h header file, adding appropriate
11365 include guards and end comments.
11366 (emit_insn_modes_c_header): Remove include of machmode.h.
11367 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11369 (main): Handle -i flag and call emit_insn_modes_inline_h when
11372 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11374 * tree-ssa-strlen.c (strinfo): Rename the length field to
11375 nonzero_chars. Add a full_string_p field.
11376 (compare_nonzero_chars, zero_length_string_p): New functions.
11377 (get_addr_stridx): Add an offset_out parameter.
11378 Use compare_nonzero_chars.
11379 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11380 (new_strinfo): Update after above changes to strinfo.
11381 (set_endptr_and_length): Set full_string_p.
11382 (get_string_length): Update after above changes to strinfo.
11383 (unshare_strinfo): Update call to new_strinfo.
11384 (maybe_invalidate): Likewise.
11385 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11386 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11387 as a uhwi instead of an shwi. Update after above changes to
11388 strinfo and new_strinfo.
11389 (zero_length_string): Assert that chainsi contains full strings.
11390 Use zero_length_string_p. Update call to new_strinfo.
11391 (adjust_related_strinfos): Update after above changes to strinfo.
11392 Copy full_string_p from origsi.
11393 (adjust_last_stmt): Use zero_length_string_p.
11394 (handle_builtin_strlen): Update after above changes to strinfo and
11395 new_strinfo. Install the lhs as the string length if the previous
11396 entry didn't describe a full string.
11397 (handle_builtin_strchr): Update after above changes to strinfo
11399 (handle_builtin_strcpy): Likewise.
11400 (handle_builtin_strcat): Likewise.
11401 (handle_builtin_malloc): Likewise.
11402 (handle_pointer_plus): Likewise.
11403 (handle_builtin_memcpy): Likewise. Track nonzero characters
11404 that aren't necessarily followed by a nul terminator.
11405 (handle_char_store): Likewise.
11407 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11409 PR tree-optimization/80769
11410 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11411 for malloc and calloc. Document the new invariant that all related
11412 strinfos have delayed lengths or none do.
11413 (verify_related_strinfos): Move earlier in file.
11414 (set_endptr_and_length): New function, split out from...
11415 (get_string_length): ...here. Also set the lengths of related
11417 (zero_length_string): Assert that chainsi has known (rather than
11419 (adjust_related_strinfos): Likewise.
11421 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11423 PR tree-optimization/81136
11424 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11425 assert that two references with the same misalignment have the same
11426 compile-time misalignment if those compile-time misalignments
11429 2017-07-01 Andi Kleen <ak@linux.intel.com>
11431 * print-tree.c (print_node): Print all attributes.
11433 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11435 * cfg.c (scale_bbs_frequencies): New function.
11436 * cfg.h (scale_bbs_frequencies): Declare it.
11437 * cfgloopanal.c (single_likely_exit): Cleanup.
11438 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11440 (scale_loop_profile): Likewise.
11441 (loop_version): Likewise.
11442 (create_empty_loop_on_edge): Update.
11443 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11444 scale_loop_frequencies, scale_loop_profile, loopify,
11445 loop_version): Update prototypes.
11446 * modulo-sched.c (sms_schedule): Update.
11447 * predict.c (unlikely_executed_edge_p): Also check probability.
11448 (probably_never_executed_edge_p): Fix typo.
11449 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11450 * tree-parloops.c (gen_parallel_loop): Update.
11451 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11452 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11453 * tree-ssa-loop-split.c (split_loop): Update.
11454 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11455 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11456 (vect_loop_versioning): Update.
11457 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11459 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11461 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11463 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11465 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11466 to keep profile consistent.
11468 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11470 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11471 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11472 * profile-count.h (max_safe_multiplier): Make unsigned.
11473 (profile_count::guessed_zero): New.
11475 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11477 * bb-reorder.c (fix_up_crossing_landing_pad,
11478 fix_crossing_conditional_branches): Use make_single_succ_edge
11479 to keep profile consistent.
11481 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11483 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11486 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11489 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11490 the right scopes, make sure cond_jump isn't preserved between multiple
11491 iterations. Search for fallthru edge whenever there are 3+ edges and
11492 use find_fallthru_edge for it.
11494 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11496 Patch by Alexander Monakov <amonakov@ispras.ru>
11497 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11498 probabilities consistently.
11500 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11502 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11503 emit_cmp_and_jump_insns.
11505 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11508 * tree-inline.c (expand_call_inline): Combine profile statuses.
11510 2017-06-30 Andrew Pinski <apinski@cavium.com>
11512 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11513 fold_stmt returned true.
11515 2017-06-30 Nathan Sidwell <nathan@acm.org>
11517 * ggc.h (empty_string): Delete.
11518 * cfgexpand.c (expand_asm_stmt): Use plain "".
11519 * optabs.c (expand_asm_memory_barrier): Likewise.
11520 * stringpool.c (empty_string): Delete.
11521 (digit_vector, digit_string): Delete.
11522 (ggc_alloc_string): Use plain "", don't optimize single digit
11523 strings. Use ggc_alloc_atomic.
11525 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11527 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11528 comparison set and one other set, use the cost of the non-comparison
11531 2017-06-30 Nathan Sidwell <nathan@acm.org>
11533 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11536 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11538 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11539 loops. Remove now unneeded calls to gimple_switch_set_label() that
11540 just set removed labels to NULL_TREE.
11542 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11544 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11545 (set_range_info): ...here. Only call set_range_info_raw if domain
11547 (set_nonzero_bits): Call set_range_info_raw.
11548 * tree-ssanames.h (set_range_info_raw): New.
11550 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11553 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11554 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11555 of nonimmediate_operand and <store_mask_constraint> instead of m
11556 for the input operand. For V8FI iterator, always split if input
11557 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11558 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11559 <store_mask_predicate> instead of register_operand and
11560 <store_mask_constraint> instead of v for the input operand. Make
11561 sure both operands aren't MEMs for if not <mask_applied>.
11563 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11565 * lto-wrapper.c (copy_file) Close both file descriptors before
11568 2017-06-30 Martin Liska <mliska@suse.cz>
11571 * multiple_target.c (create_dispatcher_calls): Make ifunc
11572 also for function that don't have calls or are not referenced.
11574 2017-06-30 Richard Biener <rguenther@suse.de>
11576 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11577 analyze the first scalar stmt. Move vector type computation
11578 for the BB case here from ...
11579 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11580 live operation processing in the SLP case properly.
11582 2017-06-30 Richard Biener <rguenther@suse.de>
11584 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11586 2017-06-30 Martin Liska <mliska@suse.cz>
11589 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11590 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11592 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11594 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11597 2017-06-30 Richard Biener <rguenther@suse.de>
11599 PR tree-optimization/81249
11600 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11601 condition reduction result to original scalar type.
11603 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11605 * profile-count.h (enum profile_quality): Fix typos and whitespace
11608 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11610 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11611 type for branch probabilities.
11613 2017-06-29 Julian Brown <julian@codesourcery.com>
11614 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11616 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11617 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11618 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11619 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11621 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11623 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11624 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11625 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11626 CC usage from generic code to here.
11627 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11628 CC usage into the target macros.
11630 2017-06-29 Maya Rashish <coypu@sdf.org>
11632 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11635 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11637 * arm/arm-builtins.c: Include profile-count.h
11638 * except.c (sjlj_emit_function_enter): Use
11639 profile_probability::unlikely.
11641 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11643 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11644 and tocrel_offset be pointer args rather than implicitly using
11646 (legitimate_constant_pool_address_p, rs6000_emit_move,
11647 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11648 tocrel_offset and use in toc_relative_expr_p call.
11649 (print_operand, print_operand_address): Use static tocrel_base_oac
11650 and tocrel_offset_oac.
11651 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11654 2017-06-29 Maya Rashish <coypu@sdf.org>
11656 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11658 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11660 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11661 objects, take into account only the alignment of 'op0' and 'mode1' if
11664 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11666 * ccmp.c (ccmp_tree_comparison_p): New function.
11667 (ccmp_candidate_p): Update to use above function.
11668 (get_compare_parts): New function.
11669 (expand_ccmp_next): Update to use new functions.
11670 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11672 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11673 take mode as argument.
11674 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11675 * expr.c (expand_expr_real_1): Pass mode as argument.
11677 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11679 * combine.c (combine_instructions): Print insns to dump_file, together
11682 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11684 * asan.c (asan_emit_stack_protection): Update.
11685 (create_cond_insert_point): Update.
11686 * auto-profile.c (afdo_propagate_circuit): Update.
11687 * basic-block.h (struct edge_def): Turn probability to
11688 profile_probability.
11689 (EDGE_FREQUENCY): Update.
11690 * bb-reorder.c (find_traces_1_round): Update.
11691 (better_edge_p): Update.
11692 (sanitize_hot_paths): Update.
11693 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11694 (make_single_succ_edge): Update.
11695 (check_bb_profile): Update.
11696 (dump_edge_info): Update.
11697 (update_bb_profile_for_threading): Update.
11698 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11699 probabilitycount to 0.
11700 * cfgbuild.c (compute_outgoing_frequencies): Update.
11701 * cfgcleanup.c (try_forward_edges): Update.
11702 (outgoing_edges_match): Update.
11703 (try_crossjump_to_edge): Update.
11704 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11705 (expand_gimple_tailcall): Update.
11706 (construct_init_block): Use make_single_succ_edge.
11707 (construct_exit_block): Use make_single_succ_edge.
11708 * cfghooks.c (verify_flow_info): Update.
11709 (redirect_edge_succ_nodup): Update.
11710 (split_edge): Update.
11711 (account_profile_record): Update.
11712 * cfgloopanal.c (single_likely_exit): Update.
11713 * cfgloopmanip.c (scale_loop_profile): Update.
11714 (set_zero_probability): Remove.
11715 (duplicate_loop_to_header_edge): Update.
11716 * cfgloopmanip.h (loop_version): Update prototype.
11717 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11718 (force_nonfallthru_and_redirect): Update.
11719 (update_br_prob_note): Update.
11720 (rtl_verify_edges): Update.
11721 (purge_dead_edges): Update.
11722 (rtl_lv_add_condition_to_bb): Update.
11723 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11724 * cgraphunit.c (init_lowered_empty_function): Update.
11725 (cgraph_node::expand_thunk): Update.
11726 * cilk-common.c: Include profile-count.h
11727 * dojump.c (inv): Remove.
11728 (jumpifnot): Update.
11729 (jumpifnot_1): Update.
11730 (do_jump_1): Update.
11732 (do_jump_by_parts_greater_rtx): Update.
11733 (do_compare_rtx_and_jump): Update.
11734 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11735 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11736 * dwarf2cfi.c: Include profile-count.h
11737 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11738 (sjlj_emit_dispatch_table): Likewise.
11739 * explow.c: Include profile-count.h
11740 * expmed.c (emit_store_flag_force): Update.
11741 (do_cmp_and_jump): Update.
11742 * expr.c (compare_by_pieces_d::generate): Update.
11743 (compare_by_pieces_d::finish_mode): Update.
11744 (emit_block_move_via_loop): Update.
11745 (store_expr_with_bounds): Update.
11746 (store_constructor): Update.
11747 (expand_expr_real_2): Update.
11748 (expand_expr_real_1): Update.
11749 * expr.h (try_casesi, try_tablejump): Update prototypes.
11750 * gimple-pretty-print.c (dump_probability): Update.
11751 (dump_profile): New.
11752 (dump_gimple_label): Update.
11753 (dump_gimple_bb_header): Update.
11754 * graph.c (draw_cfg_node_succ_edges): Update.
11755 * hsa-gen.c (convert_switch_statements): Update.
11756 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11757 (find_if_case_1): Update.
11758 (find_if_case_2): Update.
11759 * internal-fn.c (expand_arith_overflow_result_store): Update.
11760 (expand_addsub_overflow): Update.
11761 (expand_neg_overflow): Update.
11762 (expand_mul_overflow): Update.
11763 (expand_vector_ubsan_overflow): Update.
11764 * ipa-cp.c (good_cloning_opportunity_p): Update.
11765 * ipa-split.c (split_function): Use make_single_succ_edge.
11766 * ipa-utils.c (ipa_merge_profiles): Update.
11767 * loop-doloop.c (add_test): Update.
11768 (doloop_modify): Update.
11769 * loop-unroll.c (compare_and_jump_seq): Update.
11770 (unroll_loop_runtime_iterations): Update.
11771 * lra-constraints.c (lra_inheritance): Update.
11772 * lto-streamer-in.c (input_cfg): Update.
11773 * lto-streamer-out.c (output_cfg): Update.
11774 * mcf.c (adjust_cfg_counts): Update.
11775 * modulo-sched.c (sms_schedule): Update.
11776 * omp-expand.c (expand_omp_for_init_counts): Update.
11777 (extract_omp_for_update_vars): Update.
11778 (expand_omp_ordered_sink): Update.
11779 (expand_omp_for_ordered_loops): Update.
11780 (expand_omp_for_generic): Update.
11781 (expand_omp_for_static_nochunk): Update.
11782 (expand_omp_for_static_chunk): Update.
11783 (expand_cilk_for): Update.
11784 (expand_omp_simd): Update.
11785 (expand_omp_taskloop_for_outer): Update.
11786 (expand_omp_taskloop_for_inner): Update.
11787 * omp-simd-clone.c (simd_clone_adjust): Update.
11788 * optabs.c (expand_doubleword_shift): Update.
11789 (expand_abs): Update.
11790 (emit_cmp_and_jump_insn_1): Update.
11791 (expand_compare_and_swap_loop): Update.
11792 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11793 * predict.c (predictable_edge_p): Update.
11794 (edge_probability_reliable_p): Update.
11795 (set_even_probabilities): Update.
11796 (combine_predictions_for_insn): Update.
11797 (combine_predictions_for_bb): Update.
11798 (propagate_freq): Update.
11799 (estimate_bb_frequencies): Update.
11800 (force_edge_cold): Update.
11801 * profile-count.c (profile_count::dump): Add missing space into dump.
11802 (profile_count::debug): Add newline.
11803 (profile_count::differs_from_p): Explicitly convert to unsigned.
11804 (profile_count::stream_in): Update.
11805 (profile_probability::dump): New member function.
11806 (profile_probability::debug): New member function.
11807 (profile_probability::differs_from_p): New member function.
11808 (profile_probability::differs_lot_from_p): New member function.
11809 (profile_probability::stream_in): New member function.
11810 (profile_probability::stream_out): New member function.
11811 * profile-count.h (profile_count_quality): Rename to ...
11812 (profile_quality): ... this one.
11813 (profile_probability): New.
11814 (profile_count): Update.
11815 * profile.c (compute_branch_probabilities): Update.
11816 * recog.c (peep2_attempt): Update.
11817 * sched-ebb.c (schedule_ebbs): Update.
11818 * sched-rgn.c (find_single_block_region): Update.
11819 (compute_dom_prob_ps): Update.
11820 (schedule_region): Update.
11821 * sel-sched-ir.c (compute_succs_info): Update.
11822 * stmt.c (struct case_node): Update.
11823 (do_jump_if_equal): Update.
11824 (get_outgoing_edge_probs): Update.
11825 (conditional_probability): Update.
11826 (emit_case_dispatch_table): Update.
11827 (expand_case): Update.
11828 (expand_sjlj_dispatch_table): Update.
11829 (emit_case_nodes): Update.
11830 * targhooks.c: Update.
11831 * tracer.c (better_p): Update.
11832 (find_best_successor): Update.
11833 * trans-mem.c (expand_transaction): Update.
11834 * tree-call-cdce.c: Update.
11835 * tree-cfg.c (gimple_split_edge): Upate.
11836 (move_sese_region_to_fn): Upate.
11837 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11838 * tree-eh.c (lower_resx): Upate.
11839 (cleanup_empty_eh_move_lp): Upate.
11840 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11841 * tree-inline.c (copy_edges_for_bb): Update.
11842 (copy_cfg_body): Update.
11843 * tree-parloops.c (gen_parallel_loop): Update.
11844 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11845 (gimple_gen_time_profiler): Update.
11846 * tree-ssa-dce.c (remove_dead_stmt): Update.
11847 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11848 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11849 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11850 (unloop_loops): Update.
11851 (try_peel_loop): Update.
11852 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11853 * tree-ssa-loop-split.c (connect_loops): Update.
11854 (split_loop): Update.
11855 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11856 (hoist_guard): Update.
11857 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11858 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11859 (value_replacement): Update.
11860 * tree-ssa-reassoc.c (branch_fixup): Update.
11861 * tree-ssa-tail-merge.c (replace_block_by): Update.
11862 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11863 (create_edge_and_update_destination_phis): Update.
11864 (compute_path_counts): Update.
11865 (recompute_probabilities): Update.
11866 (update_joiner_offpath_counts): Update.
11867 (freqs_to_counts_path): Update.
11868 (duplicate_thread_path): Update.
11869 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11870 (struct switch_conv_info): Update.
11871 (gen_inbound_check): Update.
11872 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11873 (vect_do_peeling): Update.
11874 (vect_loop_versioning): Update.
11875 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11876 (optimize_mask_stores): Update.
11877 * ubsan.c (ubsan_expand_null_ifn): Update.
11878 * value-prof.c (gimple_divmod_fixed_value): Update.
11879 (gimple_divmod_fixed_value_transform): Update.
11880 (gimple_mod_pow2): Update.
11881 (gimple_mod_pow2_value_transform): Update.
11882 (gimple_mod_subtract): Update.
11883 (gimple_mod_subtract_transform): Update.
11884 (gimple_ic): Update.
11885 (gimple_stringop_fixed_value): Update.
11886 (gimple_stringops_transform): Update.
11887 * value-prof.h: Update.
11889 2017-06-29 Carl Love <cel@us.ibm.com>
11891 * config/rs6000/rs6000-c.c: Add support for built-in functions
11892 vector signed int vec_signed (vector float);
11893 vector signed long long vec_signed (vector double);
11894 vector signed int vec_signed2 (vector double, vector double);
11895 vector signed int vec_signede (vector double);
11896 vector signed int vec_signedo (vector double);
11897 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11898 instruction generator.
11899 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11900 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11901 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11903 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11904 vunsignede_v2df): Add define_expands.
11905 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11906 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11907 VEC_UNSIGNEDO): Add definitions.
11908 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11909 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11910 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11911 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11912 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11913 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11914 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11915 vec_unsignede, vec_unsignedo): Add builtin defines.
11916 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11918 * doc/extend.texi: Update the built-in documentation file for the
11919 new built-in functions.
11921 2017-06-29 Richard Biener <rguenther@suse.de>
11923 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11924 reduction chains to LOOP_VINFO_REDUCTIONS.
11925 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11926 SLP reductions after processing reduction chains.
11928 2017-06-29 Nathan Sidwell <nathan@acm.org>
11930 * builtins.c (fold_builtin_FUNCTION): Use
11931 lang_hooks.decl_printable_name.
11933 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11935 PR middle-end/81194
11936 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11937 with only one label.
11938 * stmt.c (expand_case): Assert NCASES is greater than one.
11940 2017-06-29 Richard Biener <rguenther@suse.de>
11942 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11944 (group_case_labels): Likewise.
11945 (find_taken_edge): Push sanity checking on val to workers...
11946 (find_taken_edge_cond_expr): ... here
11947 (find_taken_edge_switch_expr): ... and here, handle cases
11948 with just a default label.
11949 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11950 (group_case_labels): Likewise.
11951 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11952 group_case_labels does anything cleanup the CFG again.
11954 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11956 PR tree-optimization/81196
11957 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11958 exit condition comparing two IVs.
11960 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11962 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11963 profile to the dummy entry at the end of the list of architectures.
11964 * config/arm/arm-cpu-cdata.h: Regenerated.
11966 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11967 Michael Collison <michael.collison@arm.com>
11970 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11972 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11973 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11974 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11975 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11976 of shift when the shift amount is masked with constant equal to
11977 the size of the mode.
11978 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11981 2017-06-29 Martin Liska <mliska@suse.cz>
11983 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11985 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11988 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11991 (edge_growth_cache_entry::edge_growth_cache_entry): New
11993 (reset_edge_growth_cache): Update to use constructor.
11995 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11997 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11998 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11999 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12001 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12003 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12004 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12006 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12008 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12009 (*-linux-uclibc*): Add t-uclibc tmake_file.
12010 * config/t-musl: New.
12011 * config/t-uclibc: New.
12013 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12015 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12017 (gen_comm_data): Emit architectural setting of arch_prof.
12018 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12020 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12021 (armv8-m.base, armv8-m.main): Likewise.
12022 * arm-protos.h (arm_build_target): Add profile field.
12023 (arch_option): Likewise.
12024 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12026 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12027 arm_active_target.profile.
12029 2017-06-28 Richard Biener <rguenther@suse.de>
12031 PR middle-end/81227
12032 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12033 TYPE_OVERFLOW_WRAPS.
12034 * match.pd (negate_expr_p): Likewise.
12035 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12036 fold_build2, not fold_binary.
12038 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12040 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12041 Convert memory address to Pmode.
12042 (aarch64_print_operand): Assert MEM operands are always Pmode.
12044 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12047 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12048 Remove redundant if.
12049 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12050 * config/arm/aarch-common-protos.h
12051 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12052 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12054 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12057 * multiple_target.c (create_dispatcher_calls): Set the default
12058 clone to be static, not public.
12060 2017-06-28 Richard Biener <rguenther@suse.de>
12062 * tree-vect-loop.c (vectorizable_reduction): Move special
12063 cond reduction IV var creation ...
12064 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12065 parameter. Use STMT_VINFO_VECTYPE.
12066 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12069 2017-06-28 Martin Liska <mliska@suse.cz>
12072 * ipa-visibility.c (non_local_p): Handle visibility.
12074 2017-06-28 Martin Liska <mliska@suse.cz>
12077 * common.opt: Add IntegerRange to various options.
12078 * opt-functions.awk (integer_range_info): New function.
12079 * optc-gen.awk: Add integer_range_info to cl_options struct.
12080 * opts-common.c (decode_cmdline_option): Handle
12081 CL_ERR_INT_RANGE_ARG.
12082 (cmdline_handle_error): Likewise.
12083 * opts.c (print_filtered_help): Show valid interval in
12084 when --help is provided.
12085 * opts.h (struct cl_option): Add range_min and range_max fields.
12086 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
12088 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
12090 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
12091 (x * C EQ/NE y * C): New transformation.
12093 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
12095 * genmultilib (combination_space): Accept '+' in option names.
12097 2017-06-28 Martin Liska <mliska@suse.cz>
12100 * asan.c (instrument_derefs): Bail out inner references
12101 that are hard register variables.
12103 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12106 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12107 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
12109 2017-06-28 Richard Biener <rguenther@suse.de>
12111 * tree-vectorizer.h (vect_get_vec_defs): Remove.
12112 (vect_get_slp_defs): Adjust.
12113 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
12115 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
12117 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
12118 get_initial_defs_for_reduction instead of vect_get_vec_defs.
12119 (vectorizable_reduction): Adjust.
12120 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
12122 (vect_get_slp_defs): Likewise.
12123 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
12124 (vectorizable_bswap): Adjust.
12125 (vectorizable_call): Likewise.
12126 (vectorizable_conversion): Likewise.
12127 (vectorizable_assignment): Likewise.
12128 (vectorizable_shift): Likewise.
12129 (vectorizable_operation): Likewise.
12130 (vectorizable_store): Likewise.
12131 (vectorizable_condition): Likewise.
12132 (vectorizable_comparison): Likewise.
12134 2017-06-28 Michael Collison <michael.collison@arm.com>
12137 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12139 (arm_gen_movmemqi): Removed unused variable 'i'.
12140 Convert 'for' loop into 'while' loop.
12141 (arm_expand_prologue): Remove last unnecessary set of insn.
12142 (thumb_pop): Remove unused variable 'pushed_words'.
12143 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12145 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12147 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12148 * config/s390/s390-protos.h: Add prototype of
12149 s390_rel_address_ok_p.
12150 * config/s390/s390.c (s390_got_symbol): New function.
12151 (s390_rel_address_ok_p): New function.
12152 (legitimize_pic_address): Use s390_rel_address_ok_p.
12153 (s390_load_got): Use s390_got_symbol.
12154 (s390_option_override): Issue error if
12155 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12156 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12158 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12160 2017-06-27 Andrew Pinski <apinski@cavium.com>
12162 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12163 (X * copysign (1.0, X)): New pattern.
12164 (X * copysign (1.0, -X)): New pattern.
12165 (copysign (-1.0, CST)): New pattern.
12167 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12169 * genmultilib (combination_space): Remove variable.
12170 Validate reuse rules against regular expression for any sequence
12171 of multilib options in any order.
12173 2017-06-27 Michael Collison <michael.collison@arm.com>
12175 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12176 call aarch64_split_simd_combine.
12177 * (aarch64_combine_internal<mode>): Delete pattern.
12178 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12179 Allow register and subreg operands.
12181 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12183 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12184 specific need, just fallback on defaults.
12185 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12187 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12188 Olivier Hainque <hainque@adacore.com>
12190 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12192 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12193 targets. Pick a default if no particular attempt applied.
12194 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12197 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12199 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12200 (x86_64-wrs-vxworks7): Likewise.
12202 2017-06-27 Marek Polacek <polacek@redhat.com>
12205 * ubsan.c (instrument_null): Check get_base_address's result for null.
12207 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12209 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12211 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12213 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12214 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12215 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12216 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12217 New function types.
12218 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12219 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12220 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12221 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12222 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12223 BUILT_IN_FEUPDATEENV): New builtins.
12224 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12225 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12226 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12227 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12229 (builtin_structptr_types): Adjust size.
12230 * tree.c (builtin_structptr_types): Add four entries.
12232 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12233 Olivier Hainque <hainque@adacore.com>
12235 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12236 (TLS_SYM): New local macro, forcing reference to __tls__ on
12237 link command lines for VxWorks 7 RTPs, triggering initialization
12239 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12240 OS features TLS support, true for RTPs on VxWorks 7.
12241 * config/vxworks.c (vxworks_override_options): Setup emutls
12244 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12246 * predict.c (test_prediction_value_range): Use -1U instead of -1
12247 to avoid narrowing conversion warning.
12248 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12249 to avoid narrowing conversion warning.
12250 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12252 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12254 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12256 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12257 64bit configurations.
12258 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12259 (SIZE_TYPE): Likewise.
12260 * config/vxworks.c (vxworks_emutls_var_fields): Use
12261 long_unsigned_type_node instead of unsigned_type_node as the offset
12262 field type, which is "pointer" mode in emutls.c.
12264 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12267 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12269 PR middle-end/81207
12270 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12271 gimple_vuse copying separately from gimple_vdef copying.
12273 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12275 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12278 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12279 Olivier Hainque <hainque@adacore.com>
12281 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12282 for all vxworks7 targets.
12283 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12284 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12285 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12286 variations for VX6/VX7 and 32/64bits later on in ...
12287 (VXWORKS_LIB_SPEC): Leverage new macros.
12288 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12289 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12291 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12293 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12294 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12296 2017-06-26 Carl Love <cel@us.ibm.com>
12298 * config/rs6000/rs6000-c.c: Add support for built-in functions
12299 vector bool char vec_reve (vector bool char);
12300 vector signed char vec_reve (vector signed char);
12301 vector unsigned char vec_reve (vector unsigned char);
12302 vector bool int vec_reve (vector bool int);
12303 vector signed int vec_reve (vector signed int);
12304 vector unsigned int vec_reve (vector unsigned int);
12305 vector bool long long vec_reve (vector bool long long);
12306 vector signed long long vec_reve (vector signed long long);
12307 vector unsigned long long vec_reve (vector unsigned long long);
12308 vector bool short vec_reve (vector bool short);
12309 vector signed short vec_reve (vector signed short);
12310 vector double vec_reve (vector double);
12311 vector float vec_reve (vector float);
12312 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12313 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12314 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12315 (altivec_vreve): New pattern.
12316 * config/rs6000/altivec.h (vec_reve): New define.
12317 * doc/extend.texi (vec_rev): Update the built-in documentation file
12318 for the new built-in functions.
12320 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12322 PR tree-optimization/71815
12323 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12325 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12327 (slsr_process_phi): Likewise.
12328 (replace_uncond_cands_and_profitable_phis): Don't replace a
12329 multiply candidate with a stride of 1 (copy or cast).
12330 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12332 (lowest_cost_path): Likewise.
12333 (total_savings): Likewise.
12335 2017-06-26 Richard Biener <rguenther@suse.de>
12338 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12339 Use def_builtin_pure for all gather builtins.
12341 2017-06-26 Richard Biener <rguenther@suse.de>
12343 PR tree-optimization/81203
12344 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12345 non-dominating BBs.
12347 2017-06-26 Marek Polacek <polacek@redhat.com>
12350 * doc/invoke.texi: Document -Wmultistatement-macros.
12352 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12354 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12355 arm_neon_ok_no_float_abi effective target.
12357 2017-06-26 Richard Biener <rguenther@suse.de>
12359 PR tree-optimization/80928
12360 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12361 (copy_bbs): Set BB_DUPLICATED flag early.
12362 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12364 (execute_on_shrinking_pred): Likewise.
12365 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12366 BB_DUPLICATED blocks.
12367 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12368 iterate over all PHIs considering removal of *gsi.
12370 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12372 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12375 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12377 * config/rs6000/rs6000-string.c: (expand_block_clear,
12378 do_load_for_compare, select_block_compare_mode,
12379 compute_current_alignment, expand_block_compare,
12380 expand_strncmp_align_check, expand_strn_compare,
12381 expand_block_move, rs6000_output_load_multiple)
12382 Move functions related to string/block move/compare
12383 to a separate file.
12384 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12385 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12386 for this function which is now used in two files.
12387 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12388 * config.gcc: Add rs6000-string.o to extra_objs for
12389 targets powerpc*-*-* and rs6000*-*-*.
12391 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12394 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12395 32-bit, since indexed is not valid for DImode.
12396 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12397 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12398 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12399 (define_peephole2 for Altivec d-form store): Likewise.
12402 * multiple_target.c (create_dispatcher_calls): Only create the
12403 dispatcher call if the function is the default clone of a
12404 versioned function.
12406 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12408 PR middle-end/80902
12409 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12410 a call, force the call to not be a tail call.
12412 2017-06-23 Jeff Law <law@redhat.com>
12414 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12417 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12419 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12420 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12421 handling for early expansion of vector shifts (sl,sr,sra,rl).
12422 (builtin_function_type): Add vector shift right instructions
12423 to the unsigned argument list.
12425 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12427 rtl-optimizatoin/79286
12428 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12429 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12430 trap. PIC register plus a const unspec without offset can never trap.
12432 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12434 * tree.h (builtin_structptr_type): New type.
12435 (builtin_structptr_types): Declare new array.
12436 * tree.c (builtin_structptr_types): New array.
12437 (free_lang_data, build_common_tree_nodes): Use it.
12439 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12442 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12445 2017-06-22 Matt Turner <mattst88@gmail.com>
12447 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12448 Lake models to skylake case. Assume skylake for unknown
12449 models with clflushopt.
12451 2017-06-22 Jeff Law <law@redhat.com>
12453 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12454 frame sizes that do not satisfy aarch64_uimm12_shift.
12456 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12458 * profile-count.h (apply_probability,
12459 apply_scale, probability_in): Fix checks for zero.
12461 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12463 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12464 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12466 2016-06-22 Richard Biener <rguenther@suse.de>
12468 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12469 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12470 REDUC_MAX_EXPR support.
12471 (vectorizable_reduction): Likewise.
12472 (vect_create_epilog_for_reduction): Likewise.
12474 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12476 * match.pd (A / (1 << B) -> A >> B): New.
12477 * generic-match-head.c: Include optabs-tree.h.
12478 * gimple-match-head.c: Likewise.
12479 * optabs-tree.h (target_supports_op_p): New.
12480 * optabs-tree.c (target_supports_op_p): New.
12482 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12484 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12485 $gcc_cv_ld --help output.
12486 (gcc_cv_ld_demangle): Likewise.
12487 (gcc_cv_ld_eh_frame_hdr): Likewise.
12488 (gcc_cv_ld_pie): Likewise.
12489 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12490 (gcc_cv_ld_buildid): Likewise.
12491 (gcc_cv_ld_sysroot): Likewise.
12492 (ld_bndplt_support): Likewise.
12493 (ld_pushpopstate_support): Likewise.
12494 * configure: Regenerate.
12495 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12497 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12500 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12501 operands indexes to avoid gap between operands and match_dups.
12503 2017-06-21 Andrew Pinski <apinski@cavium.com>
12505 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12506 Increment Arith_shift and Arith_shift_reg by 1.
12507 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12509 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12510 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12511 (aarch64_strip_extend): Add new argument and test for it.
12512 (aarch64_cheap_mult_shift_p): New function.
12513 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12514 add a cost if it is true.
12515 Update calls to aarch64_strip_extend.
12516 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12518 2017-06-21 Andrew Pinski <apinski@cavium.com>
12520 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12522 (thunderxt88): Likewise.
12523 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12524 (thunderx_prefetch_tune): New variable.
12525 (thunderx2t99_prefetch_tune): Update for the correct values.
12526 (thunderxt88_tunings): New variable.
12527 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12528 generic_prefetch_tune.
12529 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12531 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12533 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12534 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12535 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12536 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12537 (aarch64_atomic_cas<mode>, GPI): Likewise.
12539 2017-06-21 Martin Liska <mliska@suse.cz>
12541 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12542 statements on cold and hot labels.
12543 * predict.c (tree_estimate_probability_bb): Remove the
12544 prediction from this place.
12546 2017-06-21 Martin Liska <mliska@suse.cz>
12548 PR tree-optimization/79489
12549 * gimplify.c (maybe_add_early_return_predict_stmt): New
12551 (gimplify_return_expr): Call the function.
12552 * predict.c (tree_estimate_probability_bb): Remove handling
12554 * predict.def: Update comment about early return predictor.
12555 * gimple-predict.h (is_gimple_predict): New function.
12556 * predict.def: Change default value of early return to 66.
12557 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12559 * passes.def: Put pass_strip_predict_hints to the beginning of
12562 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12564 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12565 FUNCTION_DECL declarations.
12566 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12568 (dwaf2out_decl): Likewise.
12569 * godump.c (go_early_global_decl): Skip call to the real debug hook
12570 for FUNCTION_DECL declarations.
12571 * passes.c (rest_of_decl_compilation): Skip call to the
12572 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12573 -fdump-go-spec is passed.
12575 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12577 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12578 Reorder for compactness.
12579 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12580 (def_builtin_pure, def_builtin_pure2): New functions.
12581 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12583 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12585 * match.pd (nop_convert): New predicate.
12586 ((A +- CST1) +- CST2): Allow some NOP conversions.
12588 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12591 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12592 with ctors/dtors if GOVD_SHARED is set.
12594 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12596 * config/aarch64/aarch64.md (movti_aarch64):
12597 Emit mov rather than orr.
12598 (movtf_aarch64): Likewise.
12599 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12600 Emit mov rather than orr.
12602 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12604 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12605 Swap alternatives, make integer dup more expensive.
12607 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12609 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12610 Return true for non-tls symbols.
12612 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12614 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12615 (cortex-a75): Likewise.
12616 (cortex-a75.cortex-a55): Likewise.
12617 * config/aarch64/aarch64-tune.md: Regenerate.
12618 * doc/invoke.texi (-mtune): Document new values for -mtune.
12620 2017-06-21 Tom de Vries <tom@codesourcery.com>
12622 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12623 stack_size feature.
12624 (Effective-Target Keywords, Other attributes): Suggest using
12625 dg-add-options stack_size feature to get stack limit in stack_size
12626 effective target documentation.
12628 2017-06-21 Julian Brown <julian@codesourcery.com>
12629 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12631 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12632 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12633 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12635 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12636 attribute type list for neon_multiply.
12637 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12638 attribute type list for neon_multiply.
12639 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12640 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12641 attribute type list for neon_multiply.
12642 * config/arm/types.md (crypto_pmull): Add.
12643 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12644 attribute type list.
12646 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12648 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12651 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12653 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12654 to make sure not to dereference a NULL cost_classes_ptr pointer.
12656 2017-06-20 Carl Love <cel@us.ibm.com>
12658 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12659 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12660 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12661 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12662 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12663 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12664 VMULOSW): New enum "unspec" values.
12665 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12666 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12667 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12668 altivec_vmulosw): New patterns.
12669 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12670 VMULOSW): Add definitions.
12672 2017-06-20 Julia Koval <julia.koval@intel.com>
12674 * config/i386/i386.c: Fix rounding expand for new pattern.
12675 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12677 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12679 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12680 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12682 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12684 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12687 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12689 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12690 architecture, then by implementer ID.
12691 * config/aarch64/aarch64-tune.md: Regenerate.
12693 2017-06-20 Richard Biener <rguenther@suse.de>
12695 PR middle-end/81097
12696 * fold-const.c (split_tree): Fold to type before negating.
12698 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12700 * diagnostic-show-locus.c
12701 (selftest::test_fixit_deletion_affecting_newline): New function.
12702 (selftest::diagnostic_show_locus_c_tests): Call it.
12704 2017-06-20 Andreas Schwab <schwab@suse.de>
12707 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12710 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12712 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12713 __ARM_FEATURE_COPROC according to support.
12715 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12717 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12718 Rewritten to avoid overflow for > 32-bit pointers.
12721 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12722 by removing enum keyword.
12723 (ubsan_type_descriptor): Likewise. Formatting fix.
12726 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12727 splitter): Require TARGET_SSE2 in the condition.
12729 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12732 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12733 for doing vector set of SFmode on ISA 3.0.
12734 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12735 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12737 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12738 SFmode value into a V4SF variable that was extracted from another
12739 V4SF variable without converting the element to double precision
12740 and back to single precision vector format.
12741 (vsx_insert_extract_v4sf_p9_2): Likewise.
12743 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12745 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12746 in UWHI to avoid undefined overflow.
12749 * ubsan.h (enum ubsan_encode_value_phase): New.
12750 (ubsan_encode_value): Change second argument to
12751 enum ubsan_encode_value_phase with default value of
12752 UBSAN_ENCODE_VALUE_GENERIC.
12753 * ubsan.c (ubsan_encode_value): Change second argument to
12754 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12755 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12756 create_tmp_var_raw instead of create_tmp_var and use a
12758 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12759 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12760 ubsan_encode_value callers.
12763 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12764 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12765 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12767 2017-06-19 Richard Biener <rguenther@suse.de>
12769 PR middle-end/81118
12770 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12771 estimates if we changed anything.
12773 2017-06-19 Richard Biener <rguenther@suse.de>
12775 PR tree-optimization/80887
12776 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12777 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12778 simplified lookups, then reset mprts_hook.
12779 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12781 (try_to_simplify): Likewise.
12783 2017-06-19 Martin Liska <mliska@suse.cz>
12786 * gimplify.c (gimplify_switch_expr):
12787 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12789 2017-06-19 Martin Liska <mliska@suse.cz>
12791 * doc/install.texi: Document that PGO runs in 4 stages.
12793 2017-06-19 Martin Liska <mliska@suse.cz>
12796 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12797 to dispatcher function name.
12798 * multiple_target.c (replace_function_decl): New function.
12799 (create_dispatcher_calls): Redirect both edges and references.
12801 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12803 * profile-count.c (profile_count::dump): Dump quality.
12804 (profile_count::differs_from_p): Update for unsigned val.
12805 * profile-count.h (profile_count_quality): New enum.
12806 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12808 2017-06-19 Richard Biener <rguenther@suse.de>
12810 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12811 struct function as arg.
12812 (estimate_numbers_of_iterations): Export overload with loop arg.
12813 (free_numbers_of_iterations_estimates_loop): Use an overload of
12814 free_numbers_of_iterations_estimates instead.
12815 * tree-cfg.c (remove_bb): Adjust.
12816 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12817 * tree-parloops.c (gen_parallel_loop): Likewise.
12818 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12820 (tree_unroll_loops_completely): Likewise.
12821 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12822 Use an overload instead and export.
12823 (estimated_loop_iterations): Adjust.
12824 (max_loop_iterations): Likewise.
12825 (likely_max_loop_iterations): Likewise.
12826 (estimate_numbers_of_iterations): Take struct function as arg
12828 (loop_exits_before_overflow): Adjust.
12829 (free_numbers_of_iterations_estimates_loop): Use an overload.
12830 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12831 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12833 2017-06-19 Richard Biener <rguenther@suse.de>
12836 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12837 RANGE_EXPR conservatively.
12839 2017-06-16 Carl Love <cel@us.ibm.com>
12841 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12842 definitions for vec_float, vec_float2, vec_floato,
12843 vec_floate built-ins.
12844 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12845 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12847 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12848 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12849 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12850 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12851 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12852 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12853 vec_floato): Add builtin defines.
12854 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12855 Update the built-in documentation file for the new built-in
12858 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12860 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12861 (mthumb): Mark as the negative of -marm.
12863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12865 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12867 (ARM Options, -mtune): Document that this accepts the same
12868 extension options as -mcpu.
12869 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12871 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12873 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12874 permitted extensions.
12876 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12878 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12879 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12880 (armv8-m.main): Add option +nodsp.
12881 * config/arm/arm-cpu-cdata.h: Regenerated.
12883 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12885 * config/arm/t-fuchsia: New file.
12886 * config.gcc (arm*-*-fuchsia*): Use it.
12888 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12890 * config/arm/t-symbian: Rewrite for new option infrastructure.
12892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12894 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12895 (MULTILIB_REQUIRED): Likewise.
12897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12899 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12900 (MULTILIB_RESUE): Likewise.
12901 (MULTILIB_MATCHES): Likewise.
12902 (MULTLIB_REQUIRED): Likewise.
12904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12906 * config/arm/t-rtems: Rewrite for new option framework.
12908 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12910 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12911 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12912 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12913 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12914 * config/arm/t-multilib: ... here.
12915 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12916 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12917 armv7-a and armv8*-a when A-profile libraries have not been built.
12918 * config/arm/t-rmprofile: Rewrite.
12920 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12922 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12923 with a backslash. Remove the backslash after substituting unescaped
12925 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12929 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12930 the driver to pass through the default setting of -mfloat-abi.
12931 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12933 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12934 (all_feat_combs): New rule.
12935 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12937 * config/arm/t-aprofile: Rewrite.
12939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12941 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12942 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12943 fpu is not specified by the user/command-line.
12944 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12945 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12946 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12947 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12948 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12949 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12951 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12953 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12954 * config/arm/t-arm-elf: Rewritten.
12956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12958 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12959 have some floating-point instructions.
12960 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12961 (TARGET_MAYBE_HARD_FLOAT): New macro.
12962 * config/arm/arm-builtins.c (arm_init_builtins): Use
12963 TARGET_MAYBE_HARD_FLOAT.
12964 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12968 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12969 (configargs.h): Include it.
12970 (arm_print_hint_for_fpu_option): New function.
12971 (arm_parse_fpu_option): New function.
12972 (candidate_extension): New class.
12973 (arm_canon_for_multilib): New function.
12974 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12975 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12976 (ARCH_CANONICAL_SPECS): New macro.
12977 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12979 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12981 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12982 are set after handling multilib fragments. Set target_cpu_default2
12985 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12987 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12989 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12990 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12992 (check_arch): Likewise.
12993 * config/arm/arm.c (arm_configure_build_target): Handle
12994 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12995 options in the default.
12997 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12999 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13000 when an option is an alias of another.
13001 * config/arm/parsecpu.awk (optalias): New parser token.
13002 (gen_comm_data): Mark non-alias options as such. Emit entries
13003 for extension aliases.
13004 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13005 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13006 (armv6kz, armv6zk, armv6t2): Likewise.
13007 (armv7): Make vfpv3-d16 an alias.
13008 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13010 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13011 Sort in canonical order.
13012 (armv8-a): Sort in canonical order.
13013 (armv8.1-a, armv8.2-a): Likewise.
13014 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13016 (cortex-a9): Sort in canonical order.
13017 * config/arm/arm.c (selftests.h): Include it.
13018 (arm_test_cpu_arch_data): New function.
13019 (arm_run_self_tests): New function.
13020 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13021 (targetm): Move declaration to the end of the file.
13022 * arm-cpu-cdata.h: Regenerated.
13024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13026 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13027 call to target_mode_check describing the type of option passed.
13028 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13029 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13030 arm_parse_cpu_option_name to match parameters against list of
13032 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13033 arm_arch_core_flags data structure.
13034 * config/arm/arm-cpu_cdata.h: Regenerated.
13036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13038 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13040 (arm_print_hint_for_cpu_option): Likewise.
13041 (arm_print_hint_for_arch_option): Likewise.
13042 (arm_parse_cpu_option_name): Likewise.
13043 (arm_parse_arch_option_name): Likewise.
13044 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13045 of entries in the all_fpus list.
13046 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13047 (arm_parse_cpu_option_name): Declare.
13048 (arm_parse_arch_option_name): Declare.
13049 (arm_parse_option_features): Declare.
13050 (arm_intialize_isa): Declare.
13051 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13053 (gen_comm_data): ... here. Make definitions non-static.
13054 * config/arm/arm-cpu-data.h: Regenerated.
13055 * config/arm/arm-cpu-cdata.h: Regenerated.
13057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13059 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13060 (cpu_arch_extension): New structure.
13061 (cpu_arch_option, arch_option, cpu_option): New structures.
13062 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13063 architecture types.
13064 (gen_data): Generate new format data tables.
13065 * config/arm/arm.c (cpu_tune): New structure.
13066 (cpu_option, processors): Delete.
13067 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13068 (arm_print_hint_for_cpu_option): ... this and ...
13069 (arm_print_hint_for_arch_option): ... this.
13070 (arm_parse_arch_cpu_name): Delete. Replace with ...
13071 (arm_parse_cpu_option_name): ... this and ...
13072 (arm_parse_arch_option_name): ... this.
13073 (arm_unrecognized_feature): Change type of target parameter to
13075 (arm_parse_arch_cpu_features): Delete. Replace with ...
13076 (arm_parse_option_features): ... this.
13077 (arm_configure_build_target): Rework to use new configuration data
13079 (arm_print_tune_info): Rework for new configuration data tables.
13080 * config/arm/arm-cpu-data.h: Regenerated.
13081 * config/arm/arm-cpu.h: Regenerated.
13083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13085 * Makefile.in (OBJS): Move sbitmap.o from here ...
13086 (OBJS-libcommon): ... to here.
13088 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13090 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
13091 (ISA_ALL_CRYPTO): New macro.
13092 (ISA_ALL_SIMD): New macro
13093 (ISA_ALL_FP): New macro.
13094 * config/arm/arm.c (fpu_bitlist): Update initializer.
13095 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
13097 (arm9e): Add fpu. Add option for nofp
13098 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
13099 (arm926ej-s, arm1026ej-s): Likewise.
13100 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
13101 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13102 neon-fp16, neon-vfpv4, nofp and nosimd.
13103 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13104 (cortex-a8): Add fpu. Add option for nofp.
13105 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13106 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13107 (cortex-r4f): Add fpu.
13108 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
13109 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
13111 (cortex-r8): Likewise.
13112 (cortex-m4): Add fpu. Add option for nofp.
13113 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
13114 (cortex-a17.cortex-a7): Likewise.
13115 (cortex-a32): Add fpu. Add options for crypto and nofp.
13116 (cortex-a35, cortex-a53): Likewise.
13117 (cortex-a57): Add fpu. Add option for crypto.
13118 (cortex-a72, cortex-a73): Likewise.
13119 (exynos-m1): Likewise.
13120 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
13121 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
13122 (cortex-m33): Add fpu. Add option for nofp.
13123 * config/arm/arm-cpu-cdata.h: Regenerated
13124 * config/arm/arm-cpu-data.h: Regenerated.
13126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13128 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
13129 (armv5te, armv5tej): Likewise.
13130 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13131 (armv7): Add options fp and vfpv3-d16.
13132 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13133 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13135 (armv7ve): Likewise.
13136 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13137 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13138 (armv8-a): Add nocrypto option.
13139 (armv8.1-a, armv8.2-a): Likewise.
13140 (armv8-m.main): add options fp, fp.dp and nofp.
13142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13144 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13146 (armv8-a+crc): Delete.
13147 (armv8.1-a): Add options simd, crypto and nofp.
13148 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13149 (armv8.2-a+fp16): Delete.
13150 (armv8-m.main): Add option dsp.
13151 (armv8-m.main+dsp): Delete.
13152 (cortex-a8): Add fpu. Add nofp option.
13153 (cortex-a9): Add fpu. Add nofp and nosimd options.
13154 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13155 link to main cpu and architecture data structures.
13156 (gen_comm_data): Only put isa attributes from the main architecture
13158 (option): New statement for architecture and CPU entries.
13159 * arm.c (struct cpu_option): New structure.
13160 (struct processors): Add entry for options.
13161 (arm_unrecognized_feature): New function.
13162 (arm_parse_arch_cpu_name): Ignore any characters after the first
13164 (arm_parse_arch_cpu_feature): New function.
13165 (arm_configure_build_target): Separate out any CPU and architecture
13166 features and parse separately. Don't error out if -mfpu=auto is
13167 used with only an architecture string.
13168 (arm_print_asm_arch_directives): New function.
13169 (arm_file_start): Call it.
13170 * config/arm/arm-cpu-cdata.h: Regenerated.
13171 * config/arm/arm-cpu-data.h: Likewise.
13172 * config/arm/arm-tables.opt: Likewise.
13174 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13176 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13177 assembler when it is not -mfpu=auto.
13179 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13181 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13182 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13183 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13184 (ASM_CPU_SPEC): Rewrite.
13185 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13186 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13187 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13188 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13189 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13190 copied string is NUL-terminated. Also strip any characters prefixed
13192 (arm_rewrite_selected_arch): New function.
13193 (arm_rewrite_march): New function.
13195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13197 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13198 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13199 (march, mcpu, mtune): Convert to string-based options.
13200 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13201 (arm_parse_arch_cpu_name): New function.
13202 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13203 identify selected architecture or CPU.
13204 (arm_option_save): New function.
13205 (TARGET_OPTION_SAVE): Redefine.
13206 (arm_option_restore): Restore string options.
13207 (arm_option_print): Print string options.
13209 2017-06-16 Martin Sebor <msebor@redhat.com>
13211 PR tree-optimization/80933
13212 PR tree-optimization/80934
13213 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13214 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13215 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13216 (gimple_fold_builtin): Call them.
13218 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13220 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13221 as unlikely; update profile.
13223 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13225 * predict.c (force_edge_cold): Handle declaring edges impossible
13228 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13230 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13232 (try_unroll_loop_completely): Fix reporting.
13234 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13236 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13238 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13241 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13242 if given a non-constant argument for an intrinsic which requires a
13245 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13247 * profile.c (compare_freqs): New function.
13248 (branch_prob): Sort edge list.
13249 (find_spanning_tree): Assume that the list is priority sorted.
13251 2017-06-16 Richard Biener <rguenther@suse.de>
13253 PR tree-optimization/81090
13254 * passes.def (pass_record_bounds): Remove.
13255 * tree-pass.h (make_pass_record_bounds): Likewise.
13256 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13257 make_pass_record_bounds): Likewise.
13258 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13259 not free niter estimates at the beginning but at the end.
13260 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13262 2017-06-16 Richard Biener <rguenther@suse.de>
13264 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13265 initializer to workaround ICE in host GCC 4.8.
13267 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13269 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13271 (clone_inlined_nodes): Update.
13273 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13275 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13276 prefetch settings, and enable prefetching by default at -O3.
13278 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13280 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13281 Set flag_prefetch_loop_arrays according to tuning data.
13283 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13285 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13286 New tune structure.
13287 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13288 [Unrelated to main purpose of the patch] Place the pointer field last
13289 to enable type checking errors when tune structure are wrongly merged.
13290 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13291 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13292 (thunderx2t99_prefetch_tune): New tune constants.
13293 (tune_params *_tunings): Update all tunings (no functional change).
13294 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13295 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13296 from tunings structures.
13298 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13301 * ubsan.c (instrument_null): Add T argument, use it instead
13302 of computing it based on IS_LHS.
13303 (instrument_object_size): Likewise.
13304 (pass_ubsan::execute): Adjust instrument_null and
13305 instrument_object_size callers to pass gimple_get_lhs or
13306 gimple_assign_rhs1 result to it. Use instrument_null instead of
13307 calling get_base_address and instrument_mem_ref. Handle
13308 aggregate call arguments for object-size sanitization.
13310 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13312 PR tree-optimization/81089
13313 * tree-vrp.c (is_masked_range_test): Validate operands of
13316 2017-06-15 Martin Sebor <msebor@redhat.com>
13319 * dumpfile.c (dump_register): Avoid calling memset to initialize
13320 a class with a default ctor.
13321 * gcc.c (struct compiler): Remove const qualification.
13322 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13323 * hash-table.h: Ditto.
13324 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13326 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13327 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13329 * params.h (struct param_info): Make struct members non-const.
13330 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13331 with default initialization.
13332 * vec.h (vec_copy_construct, vec_default_construct): New helper
13334 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13335 with vec_copy_construct.
13336 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13337 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13338 * doc/invoke.texi (-Wclass-memaccess): Document.
13340 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13342 * emit-rtl.h (is_leaf): Update comment about local
13343 register allocator.
13345 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13348 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13349 for a variable to have a section before checking if the section has a
13351 Set section to.persistent if persistent attribute is set.
13352 Warn if .persistent attribute is used on an automatic variable.
13354 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13356 PR rtl-optimization/80474
13357 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13359 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13361 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13364 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13366 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13367 (casesi): Emit bounds checking as RTL.
13368 (casesi_internal_mips16_<mode>): Remove bounds checking.
13370 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13372 * config/xtensa/xtensa.c (xtensa_option_override): Append
13373 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13374 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13375 xtensa_doloop_hooks): Define unconditionally.
13376 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13378 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13379 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13380 for it in xtensa_option_override.
13381 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13382 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13384 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13386 * doc/cppopts.texi: Document '-' special value to -MF.
13388 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13390 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13391 (cortex_a53_fconst): Likewise.
13392 (cortex_a53_fpmul): Likewise.
13393 (cortex_a53_f_load_64): Likewise.
13394 (cortex_a53_f_load_many): Likewise.
13395 (cortex_a53_advsimd_alu): Likewise.
13396 (cortex_a53_advsimd_alu_q): Likewise.
13397 (cortex_a53_advsimd_mul): Likewise.
13398 (cortex_a53_advsimd_mul_q): Likewise.
13399 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13400 Add missing fmul, r2f_cvt and fconst cases.
13402 2017-06-14 Richard Biener <rguenther@suse.de>
13404 PR middle-end/81088
13405 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13407 (fold_binary_loc): When associating do not treat pre-existing
13408 TREE_OVERFLOW on literal constants as a reason to allow
13409 TREE_OVERFLOW on associated literal constants.
13411 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13413 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13414 (MASK_FEATURES): New macro.
13415 * config/sparc/sparc.c (sparc_option_override): Remove the special
13416 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13418 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13420 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13421 a division of 0 if non-call exceptions are enabled.
13423 2017-06-14 Andrew Pinski <apinski@cavium.com>
13424 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13427 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13428 Improve vector initialization code gen for only variable case.
13430 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13432 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13434 2017-06-14 Richard Biener <rguenther@suse.de>
13436 PR tree-optimization/81083
13437 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13440 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13442 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13443 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13444 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13445 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13446 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13447 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13449 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13451 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13452 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13454 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13456 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13458 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13460 * config/rs6000/t-rtems: Don't handle SPE.
13462 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13464 * config/rs6000/t-linux: Don't handle SPE.
13466 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13468 * config/rs6000/eabispe.h: Delete file.
13470 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13472 * config/rs6000/t-spe: Delete file.
13474 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13476 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13477 (rs6000_legitimate_offset_address_p): Return false for anything in
13478 V2SImode or V2SFmode.
13480 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13482 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13483 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13484 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13486 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13487 (rs6000_legitimate_offset_address_p): Ditto.
13488 (rs6000_emit_move): Ditto.
13489 (rs6000_init_builtins): Remove V4HI_type_node.
13491 2017-06-13 Martin Liska <mliska@suse.cz>
13494 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13495 (gate_asan): Likewise.
13496 * asan.h (asan_no_sanitize_address_p): Remove the function.
13497 (sanitize_flags_p): New function.
13498 * builtins.def: Fix coding style.
13499 * common.opt: Use renamed enum value.
13500 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13501 * doc/extend.texi: Document no_sanitize attribute.
13502 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13503 to SANITIZE_UNDEFINED_NONDEFAULT.
13504 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13505 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13506 Use sanitize_flags_p.
13507 * gimplify.c (gimplify_function_tree): Likewise.
13508 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13509 * opts.c (parse_no_sanitize_attribute): New function.
13510 (common_handle_option): Use renamed enum value.
13511 * opts.h (parse_no_sanitize_attribute): Declare.
13512 * tree.c (sanitize_flags_p): New function.
13513 * tree.h: Declared here.
13514 * tsan.c: Use sanitize_flags_p.
13515 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13516 (instrument_mem_ref): Likewise.
13517 (instrument_bool_enum_load): Likewise.
13518 (do_ubsan_in_current_function): Remove the function.
13519 (pass_ubsan::execute): Use sanitize_flags_p.
13520 * ubsan.h: Remove do_ubsan_in_current_function
13521 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13522 (dump_function_to_file): Use it here.
13524 2017-06-13 Martin Jambor <mjambor@suse.cz>
13526 PR tree-optimization/80803
13527 PR tree-optimization/81063
13528 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13529 (propagate_subaccesses_across_link): Enqueue subtree whenever
13530 necessary instead of relying on the caller.
13532 2017-06-13 Martin Jambor <mjambor@suse.cz>
13534 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13535 that have a first_link.
13536 (sort_and_splice_var_accesses): Do not check first_link before
13538 (subtree_mark_written_and_enqueue): Likewise.
13539 (propagate_all_subaccesses): Likewise and do not stop at first
13540 parent with a first_link.
13542 2017-06-13 Martin Jambor <mjambor@suse.cz>
13544 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13547 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13549 * match.pd: New pattern.
13551 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13553 * tree-vrp.c (is_masked_range_test): New function.
13554 (register_edge_assert_for): Determine ranges for
13557 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13559 PR tree-optimization/67328
13560 * fold-const.c (maskable_range_p): New function.
13561 (build_range_check): Generate bittests if possible.
13563 2017-06-13 Martin Liska <mliska@suse.cz>
13565 * gimple-pretty-print.c (dump_probability): Add new argument.
13566 (dump_edge_probability): Dump both probability and count.
13567 (dump_gimple_label): Likewise.
13568 (dump_gimple_bb_header): Likewise.
13570 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13573 * config/avr/avr-devices.c: Fix indentation.
13574 * config/avr/gen-avr-mmcu-specs.c: Dito.
13576 2017-06-13 Richard Biener <rguenther@suse.de>
13578 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13579 instead get vector type from stmt_info.
13580 (vectorizable_reduction): Adjust. Remove dead code.
13582 2017-06-13 Richard Biener <rguenther@suse.de>
13584 PR middle-end/81065
13585 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13586 case of C * (x * C2 + C3).
13587 (fold_addr_of_array_ref_difference): Properly fold index difference.
13589 2017-06-12 David S. Miller <davem@davemloft.net>
13592 * config/sparc/sparc.md (return expander): Emit frame blockage if
13593 function uses alloca.
13595 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13597 * combine.c (make_field_assignment): Check len rather than the mode
13598 precision when calling force_to_mode.
13600 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13602 Support multilibs and devices that see flash in RAM address range.
13605 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13606 (avr_mcu_t) <flash_pm_offset>: New field.
13607 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13608 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13609 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13610 (AVR_TINY_PM_OFFSET): Remove macro.
13611 * config/avr/avr.opt (-mshort-calls): New option.
13612 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13613 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13614 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13615 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13616 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13617 instead of avr_arch->have_jmp_call.
13618 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13619 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13620 avr_arch->flash_pm_offset to define.
13621 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13622 new field flash_pm_offset. Add entry for avrxmega3.
13623 (avr_texinfo): Add entry for avrxmega3.
13624 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13625 attiny212, attiny214,
13626 attiny412, attiny414, attiny416, attiny417,
13627 attiny814, attiny816, attiny817,
13628 attiny1614, attiny1616, attiny1617,
13629 attiny3214, attiny3216, attiny3217.
13630 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13631 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13632 (avr_print_operand_address) [AVR_TINY]: Same.
13633 (avr_asm_init_sections) <readonly_data_section>: Only patch
13634 callback if avr_arch->flash_pm_offset = 0.
13635 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13636 for rodata if avr_arch->flash_pm_offset != 0.
13637 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13638 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13639 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13640 (m_options): Append opt_rcall.
13641 (m_dirnames): Append dir_rcall.
13642 * config/avr/t-multilib: Regenerate.
13644 * configure.ac [target=avr]: Check whether avrxmega3 default
13645 linker description file works as needed.
13646 * configure: Regenerate.
13647 * doc/avr-mmcu.texi: Regenerate.
13648 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13649 <__AVR_ARCH__>: Document avrxmega3 and 103.
13650 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13651 <__AVR_SHORT_CALLS__>: Document it.
13652 <__AVR_PM_BASE_ADDRESS__>: Document it.
13653 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13654 (AVR Variable Attributes) <progmem>: Document this is
13655 not needed for avrxmega3.
13656 (AVR Named Address Spaces) <__flash>: Dito.
13658 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13660 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13662 2017-06-12 Doug Rupp <rupp@adacore.com>
13664 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13665 Append vxworks-stdint.h to the tm_file list.
13666 * config/vxworks-stdint.h: New file.
13668 2017-06-12 Martin Liska <mliska@suse.cz>
13670 PR tree-optimization/81041
13671 * tree-profile.c (gimple_gen_ic_func_profiler):
13672 Create an extra BB in profile-generate
13673 (gimple_gen_time_profiler): Likewise.
13675 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13677 PR tree-optimization/81003
13678 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13679 (update_range_test): Use it instead of force_gimple_operand_gsi.
13681 2017-06-12 Richard Biener <rguenther@suse.de>
13683 PR tree-optimization/81053
13684 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13685 with backedge value not defined in loop. Simplify def stmt
13688 2017-06-11 Tom de Vries <tom@codesourcery.com>
13691 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13693 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13694 nvptx_cannot_force_const_mem.
13696 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13698 * opts.c (finish_options): Move test for flag_split_stack after
13699 it has been initialized.
13701 2017-06-11 Jason Merrill <jason@redhat.com>
13703 * tree.h (id_equal): New.
13704 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13705 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13706 instead of strcmp of IDENTIFIER_POINTER.
13708 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13710 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13711 (mark_all_inlined_calls_cdtor): Fix formating.
13712 (inline_transform): Rescale profile before inlining.
13714 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13716 * cgraph.h (cgraph_edge::clone): Update prototype.
13717 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13718 (cgraph_node::create_clone): Update.
13719 (cgraph_node::create_version_clone): Update.
13720 * tree-inline.c (copy_bb): Update.
13721 (expand_call_inline): Update.
13723 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13725 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13726 factored out from ...
13727 (rs6000_emit_prologue): ... here.
13729 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13731 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13732 factored out from ...
13733 (rs6000_emit_prologue): ... here.
13735 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13737 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13739 (handle_missing_profiles): Fix computation of tp_first_run.
13740 (counts_to_freqs): Do not touch freqs when count is 0.
13742 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13744 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13747 2017-06-10 Tom de Vries <tom@codesourcery.com>
13749 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13750 attributes): Document signal effective target.
13752 2017-06-10 Tom de Vries <tom@codesourcery.com>
13754 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13755 Document effective target stack_size.
13757 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13759 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13760 to the edit_context if they can be auto-applied.
13762 2017-06-9 Ian Lance Taylor <iant@golang.org>
13764 * opts.c (finish_options): If -fsplit-stack, disable implicit
13765 -forder-blocks-and-partition.
13766 * doc/invoke.texi (Optimize Options): Document that when using
13767 -fsplit-stack -forder-blocks-and-partition is not implicitly
13770 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13772 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13773 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13774 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13775 * builtins.def (abort, trap, unreachable): Declare cold.
13776 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13777 * tree-core.h (ECF_COLD): New.
13778 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13779 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13781 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13783 * predict.c (unlikely_executed_stmt_p): Cleanup.
13785 2017-06-09 Richard Biener <rguenther@suse.de>
13787 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13788 model if the ref is always written to.
13790 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13792 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13794 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13796 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13799 2017-06-09 Tom de Vries <tom@codesourcery.com>
13802 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13803 "target cannot support label values" when encountering LABEL_REF.
13805 2017-06-09 Martin Liska <mliska@suse.cz>
13807 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13808 (gimple_gen_ic_func_profiler): Emit direct comparison
13809 of __gcov_indirect_call_callee with NULL.
13810 (gimple_gen_time_profiler): Change probability from
13811 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13813 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13815 * profile.c (edge_gcov_counts): Turn to pointer.
13816 (compute_branch_probabilities, compute_branch_probabilities): Update.
13817 (branch_prob): Do not clear edge_gcov_count.
13818 * profile.h (edge_gcov_counts): Turn to pointer.
13819 (edge_gcov_count): Update.
13821 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13823 * gimple.h (gimple_check_failed): Mark cold.
13825 2017-06-09 Richard Biener <rguenther@suse.de>
13827 PR tree-optimization/66623
13828 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13829 refactor check_reduction into two parts, properly computing
13830 whether we have to check reduction validity for outer loop
13833 2017-06-09 Richard Biener <rguenther@suse.de>
13835 PR tree-optimization/79483
13836 * graphite-scop-detection.c (order): New global.
13837 (get_order): Compute bb to order mapping that satisfies code
13838 generation constraints.
13839 (cmp_pbbs): New helper.
13840 (build_scops): Start domwalk at entry block, sort generated
13843 2017-06-09 Richard Biener <rguenther@suse.de>
13845 PR middle-end/81007
13846 * ipa-polymorphic-call.c
13847 (ipa_polymorphic_call_context::restrict_to_inner_class):
13848 Skip FIELD_DECLs with error_mark_node type.
13849 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13852 2017-06-09 Martin Liska <mliska@suse.cz>
13854 * predict.c (struct branch_predictor): New struct.
13855 (test_prediction_value_range): New test.
13856 (predict_c_tests): New function.
13857 * selftest-run-tests.c (selftest::run_tests): Run the function.
13858 * selftest.h: Declare new tests.
13860 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13863 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13864 gen_add3_insn did not fail.
13865 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13866 r0, construct that number in a temporary reg and add that reg to r0.
13867 If asked to put the result in r0 as well, fail.
13869 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13872 for early expansion of vec_eqv.
13874 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13876 PR middle-end/81005
13877 * ubsan.c (instrument_null): Avoid pointless code temporary.
13878 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13880 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13884 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13887 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13890 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13892 * predict.c (unlikely_executed_edge_p): Move ahead.
13893 (probably_never_executed_edge_p): Use it.
13895 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13897 PR middle-end/79988
13898 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13899 gimple_call_builtin_p call.
13901 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13903 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13904 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13905 rtl_check_failed_type2, rtl_check_failed_code1,
13906 rtl_check_failed_code2, rtl_check_failed_code_mode,
13907 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13908 rtvec_check_failed_bounds, rtl_check_failed_flag,
13909 _fatal_insn_not_found, _fatal_insn): Likewise.
13910 * tree.h (tree_contains_struct_check_failed,
13911 tree_check_failed, tree_not_check_failed,
13912 tree_class_check_failed, tree_range_check_failed,
13913 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13914 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13915 tree_operand_check_failed, omp_clause_check_failed,
13916 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13919 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13921 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13922 flag_branch_probabilities.
13923 * ipa-inline.c (edge_badness): Likewise.
13924 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13925 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13926 * predict.c (maybe_hot_frequency_p): Likewise.
13927 (probably_never_executed): Likewise.
13928 * sched-ebb.c (schedule_ebbs): Likewise.
13929 * sched-rgn.c (find_single_block_region): Likewise.
13930 * tracer.c (tail_duplicate): Likewise.
13932 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13934 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13935 longer requires x_flag_profile_use.
13937 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13939 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13940 instead of flag_reorder_blocks_and_partition.
13941 * dbxout.c (dbxout_function_end): Likewise.
13942 * dwarf2out.c (gen_subprogram_die): Likewise.
13943 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13944 * hw-doloop.c (reorg_loops): Likewise.
13945 * varasm.c (assemble_start_function,
13946 assemble_end_function): Likewise.
13947 (decide_function_section): Do not check for
13948 flag_reorder_blocks_and_partition.
13950 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13952 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13954 (chkp_get_hard_register_fake_addr_expr): Ditto.
13955 (chkp_build_addr_expr): Add check for hard reg case.
13956 (chkp_parse_array_and_component_ref): Ditto.
13957 (chkp_find_bounds_1): Ditto.
13958 (chkp_process_stmt): Don't generate bounds store for
13961 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13963 * predict.c (maybe_hot_bb_p): Do not check profile status.
13964 (maybe_hot_edge_p): Likewise.
13965 (probably_never_executed): Check for zero counts even if profile
13967 (unlikely_executed_edge_p): New function.
13968 (unlikely_executed_stmt_p): New function.
13969 (unlikely_executed_bb_p): New function.
13970 (set_even_probabilities): Use unlikely predicates.
13971 (combine_predictions_for_bb): Likewise.
13972 (predict_paths_for_bb): Likewise.
13973 (predict_paths_leading_to_edge): Likewise.
13974 (determine_unlikely_bbs): New function.
13975 (estimate_bb_frequencies): Use it.
13976 (compute_function_frequency): Use zero counts even if profile is
13978 * profile-count.h: Fix typo.
13980 2017-08-08 Julia Koval <julia.koval@intel.com>
13982 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13983 _mm512_mask_cvtsepi16_storeu_epi8,
13984 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13985 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13986 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13987 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13988 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13989 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13990 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13991 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13992 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13993 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13994 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13995 __builtin_ia32_pmovuswb256mem_mask,
13996 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13997 __builtin_ia32_pmovwb512mem_mask): New builtins.
13999 2017-08-08 Julia Koval <julia.koval@intel.com>
14001 PR target/73350,80862
14002 * config/i386/subst.md (round): Fix round pattern.
14003 * config/i386/i386.c (ix86_erase_embedded_rounding):
14004 Fix erasing rounding for the fixed pattern.
14006 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14008 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14010 2017-06-08 Martin Liska <mliska@suse.cz>
14012 PR gcov-profile/80911
14013 * gcov.c (block_info::block_info): New constructor.
14015 2017-06-07 Carl Love <cel@us.ibm.com>
14017 * config/rs6000/rs6000-c: The return type of the following
14018 built-in functions was implemented as int not long long. Fix sign
14019 of return value for the unsigned version of vec_mulo and vec_mule.
14020 vector unsigned long long vec_bperm (vector unsigned long long,
14021 vector unsigned char)
14022 vector signed long long vec_mule (vector signed int,
14024 vector unsigned long long vec_mule (vector unsigned int,
14025 vector unsigned int)
14026 vector signed long long vec_mulo (vector signed int,
14028 vector unsigned long long vec_mulo (vector unsigned int,
14029 vector unsigned int)
14030 * doc/extend.texi: Fix the documentation for the built-in
14033 2017-06-07 Carl Love <cel@us.ibm.com>
14036 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14039 2017-06-07 Carl Love <cel@us.ibm.com>
14041 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14042 support, Generate doublehv for signed int/float for BE case only.
14044 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14046 * doc/invoke.texi (mcx16): Rewrite.
14048 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14050 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14051 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14052 *mov<mode>_softfloat, and an anonymous splitter): Use
14053 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14055 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14057 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14059 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14060 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14061 (rs6000_debug_reg_global): Adjust.
14062 (rs6000_init_hard_regno_mode_ok): Adjust.
14063 (rs6000_dbx_register_number): Adjust.
14064 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14065 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14066 Remove SPE_ACC and SPEFSCR.
14067 (REG_ALLOC_ORDER): Ditto.
14068 (FRAME_POINTER_REGNUM): Change to 111.
14069 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14070 (REG_CLASS_NAMES): Ditto.
14071 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14072 (REGISTER_NAMES): Ditto.
14073 (ADDITIONAL_REG_NAMES): Ditto.
14074 (rs6000_reg_names): Ditto.
14075 * config/rs6000/rs6000.md: Renumber some register number
14078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14080 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
14082 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14083 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
14085 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
14086 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14087 Delete the SPE high registers.
14088 (REG_ALLOC_ORDER): Ditto.
14089 (enum reg_class): Remove SPE_HIGH_REGS.
14090 (REG_CLASS_NAMES): Ditto.
14091 (REG_CLASS_CONTENTS): Delete the SPE high registers.
14092 (REGISTER_NAMES): Ditto.
14093 (rs6000_reg_names): Ditto.
14094 * doc/tm.texi.in: Remove SPE as example.
14095 * doc/tm.texi: Regenerate.
14097 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14099 * config/rs6000/8540.md (ppc8540_brinc): Delete.
14100 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
14101 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14102 * config/rs6000/rs6000.md (type): Remove "brinc".
14104 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14106 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14107 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
14108 * config/rs6000/linuxspe.h: Delete file.
14109 * config/rs6000/rs6000.md: Don't include spe.md.
14110 * config/rs6000/spe.h: Delete file.
14111 * config/rs6000/spe.md: Delete file.
14112 * config/rs6000/t-rs6000: Remove spe.md.
14114 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14116 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
14117 (reg_or_none500mem_operand): Delete.
14118 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
14119 instead of reg_or_none500mem_operand.
14121 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14123 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14124 handling of SPE flags.
14125 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
14127 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14129 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14131 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14132 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14133 paired_divv2sf3): Similar.
14134 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14135 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14136 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14138 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14139 Rename the paired_* instruction patterns.
14140 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14142 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14143 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14144 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14145 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14146 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14147 PAIRED_VECTOR_MODE.
14148 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14149 (spe_func_has_64bit_regs_p): Delete.
14150 (spe_expand_predicate_builtin): Delete.
14151 (spe_expand_evsel_builtin): Delete.
14152 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14153 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14154 (invalid_e500_subreg): Delete.
14155 (rs6000_legitimize_address): Always force_reg op2 as well, for
14156 paired single memory accesses.
14157 (rs6000_member_type_forces_blk): Delete.
14158 (rs6000_spe_function_arg): Delete.
14159 (rs6000_expand_unop_builtin): Delete SPE handling.
14160 (rs6000_expand_binop_builtin): Ditto.
14161 (spe_expand_stv_builtin): Delete.
14162 (bdesc_2arg_spe): Delete.
14163 (spe_expand_builtin): Delete.
14164 (spe_expand_predicate_builtin): Delete.
14165 (spe_expand_evsel_builtin): Delete.
14166 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14167 (spe_init_builtins): Delete.
14168 (spe_func_has_64bit_regs_p): Delete.
14169 (savres_routine_name): Delete "info" parameter. Adjust callers.
14170 (rs6000_emit_stack_reset): Ditto.
14171 (rs6000_dwarf_register_span): Delete.
14172 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14173 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14174 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14176 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14178 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14179 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14180 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14181 mulv2sf3, divv2sf3): Delete expanders.
14183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14185 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14187 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14189 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14190 * config/rs6000/rs6000.c: Ditto.
14192 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14194 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14195 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14196 comparison_operator.
14198 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14200 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14201 * config/rs6000/rs6000.opt: Ditto.
14202 * config/rs6000/t-rtems: Ditto.
14204 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14206 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14207 TARGET_E500_SINGLE by 0, simplify.
14208 * config/rs6000/rs6000.c: Ditto.
14209 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14210 (spe_build_register_parallel): Delete.
14211 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14212 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14213 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14214 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14215 (E500_CONVERT): Delete.
14216 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14218 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14220 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14221 * config/rs6000/dfp.md: Ditto.
14222 (negdd2, *negdd2_fpr): Merge.
14223 (absdd2, *absdd2_fpr): Merge.
14224 (negtd2, *negtd2_fpr): Merge.
14225 (abstd2, *abstd2_fpr): Merge.
14226 * config/rs6000/e500.h: Delete file.
14227 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14228 TARGET_FPRS by 1 and simplify.
14229 * config/rs6000/rs6000-c.c: Ditto.
14230 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14231 TARGET_DF_SPE by 0.
14232 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14234 * config/rs6000/rs6000.md: Ditto.
14235 (floatdidf2, *floatdidf2_fpr): Merge.
14236 (move_from_CR_gt_bit): Delete.
14237 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14238 (E500_CR_IOR_COMPARE): Delete.
14239 (All patterns that require !TARGET_FPRS): Delete.
14240 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14242 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14244 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14246 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14248 * graphds.c (add_edge): Intitialize edge's attached data.
14249 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14250 pointer parameter. Call pointed function on each edge during
14251 graph traversing. Skip traversing the edge when the function
14253 (graphds_dfs, graphds_scc): Ditto.
14254 (for_each_edge): New parameter. Pass the new parameter to callback
14256 * graphds.h (skip_edge_callback): New function pointer type.
14257 (graphds_dfs, graphds_scc): New function pointer parameter.
14258 (graphds_edge_callback, for_each_edge): New parameter.
14260 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14262 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14263 out code checking if runtime alias check is possible to below ...
14264 Call the new function.
14265 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14266 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14268 2017-06-07 Marek Polacek <polacek@redhat.com>
14271 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14272 TYPE_OVERFLOW_WRAPS check.
14274 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14276 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14277 if versioning is required.
14278 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14279 peeling with the check for versioning.
14281 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14283 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14284 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14285 Set true to new parameter if new ssa variable is defined.
14286 (vect_gen_vector_loop_niters): Refactor. Set range information
14287 for the new vector loop bound variable.
14288 (vect_do_peeling): Ditto.
14290 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14292 * tree-affine.c (ssa.h): Include header file.
14293 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14294 has wrapping overflow behavior.
14296 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14298 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14300 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14302 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14303 (tree_to_aff_combination): ... here.
14305 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14307 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14308 reg_pressure model function.
14309 (ivopts_global_cost_for_size): Delete.
14310 (determine_set_costs, iv_ca_recount_cost): Call new model function
14311 ivopts_estimate_reg_pressure.
14313 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14315 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14316 expensive than udiv. Remove floating point cases from mod.
14318 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14320 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14321 Increase idiv cost.
14323 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14325 * config/aarch64/aarch64.md
14326 (copysignsf3): Fix mask generation.
14328 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14330 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14332 (class dump_manager): Add register_dumps method.
14333 * dumpfile.c: Include langhooks.h.
14334 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14335 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14336 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14337 (dump_manager::dump_register): Start with 512 entries instead of 32.
14338 (dump_manager::register_dumps): New method.
14339 * toplev.c (general_init): Instead of invoking register_dumps
14340 langhook, invoke register_dumps method on the dump manager.
14341 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14344 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14346 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14347 Say that # requires an associated define_split to exist, and that
14348 the define_split must be suitable for use after register allocation.
14350 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14352 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14353 (compute_outgoing_frequencies): Also initialize zero counts.
14354 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14355 around loops; preserve more of profile when nothing changes.
14357 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14359 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14361 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14362 * config/arm/arm-cpu-cdata.h: Regenerate.
14363 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14364 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14365 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14366 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14367 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14369 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14371 * config/arm/t-rmprofile: Likewise.
14372 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14374 2017-06-06 David S. Miller <davem@davemloft.net>
14377 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14378 blockage if function uses alloca.
14380 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14382 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14383 New "uid" fields to hold pretty-print IDs of group and ref.
14384 Memory references are now identified as <group_id>:<ref_id>
14385 instead of using [random] addresses.
14386 (dump_mem_details): Simplify, no functional change.
14387 (dump_mem_ref): Simplify and make output more concise.
14388 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14389 (find_or_create_group): Initialize group uid.
14390 (record_ref): Initialize ref uid. Improve debug output.
14391 (prune_group_by_reuse, should_issue_prefetch_p,)
14392 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14393 (mark_nontemporal_store, determine_loop_nest_reuse):
14394 Improve debug output.
14396 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14398 * dbgcnt.def (prefetch): New debug counter.
14399 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14400 (schedule_prefetches): Stop issueing prefetches if debug counter
14403 2017-06-06 Tom de Vries <tom@codesourcery.com>
14405 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14406 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14408 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14410 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14411 Use aarch64_reg_or_zero predicate for operand 4.
14412 (aarch64_compare_and_swap<mode> define_insn_and_split):
14413 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14414 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14416 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14418 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14419 (arm_compute_save_core_reg_mask): This.
14420 (thumb1_compute_save_reg_mask): Rename into ...
14421 (thumb1_compute_save_core_reg_mask): This.
14422 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14423 (arm_compute_frame_layout): Likewise.
14425 2017-06-06 Richard Biener <rguenther@suse.de>
14427 PR tree-optimization/80974
14428 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14429 keep or clear leaders SSA info.
14431 2017-06-06 Tom de Vries <tom@codesourcery.com>
14433 * config/nvptx/nvptx.c (split_mode_p): New function.
14434 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14436 2017-06-06 Tom de Vries <tom@codesourcery.com>
14438 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14440 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14443 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14446 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14448 * shrink-wrap.c (handle_simple_exit): Update profile.
14449 (try_shrink_wrapping): Upate profile.
14451 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14453 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14454 (tree_guess_outgoing_edge_probabilities): New.
14455 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14456 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14458 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14460 * ipa-split.c (split_function): Initialize return bb profile.
14462 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14464 * profile.c (compute_branch_probabilities): Also initialize
14465 EXIT_BLOCK profile.
14467 2017-06-06 Richard Biener <rguenther@suse.de>
14469 PR tree-optimization/80928
14470 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14471 (vect_analyze_loop_operations): Properly guard analysis for
14473 (vect_transform_loop): Likewise.
14474 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14475 (vect_model_induction_cost): Do not cost for pure SLP.
14476 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14477 of induction in inner loop vectorization.
14478 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14479 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14480 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14482 (vect_analyze_slp_cost_1): Cost induction.
14483 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14484 (vect_get_slp_vect_defs): Likewise.
14485 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14486 (vect_transform_stmt): Handle SLP reductions.
14487 * tree-vectorizer.h (vectorizable_induction): Adjust.
14489 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14491 * config/rs6000/rs6000.c (make_resolver_func): Update
14492 init_lowered_empty_function call.
14494 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14496 * doc/invoke.texi: Document the -fprofile-abs-path option.
14497 * common.opt (fprofile-abs-path): New option.
14498 * gcov-io.h (gcov_write_filename): Declare.
14499 * gcov-io.c (gcov_write_filename): New function.
14500 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14501 * profile.c (output_location): Likewise.
14503 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14505 * shring-wrap.c: Revert accidental commit.
14507 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14509 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14511 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14513 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14515 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14516 profile in callgraph edge.
14517 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14518 (apply_scale): Likewise.
14519 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14520 Also scale profile when inlining function with zero profile.
14521 (initialize_cfun): Update exit block profile even when it is zero.
14522 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14523 when profile is read.
14525 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14527 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14528 (CLONE_*): New constants to define the processors we can generate
14529 code for with the target_clone attribute.
14530 (rs6000_clone_map): New array to identify which clone processors
14531 the current program is running on.
14532 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14533 target_clone attribute.
14534 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14535 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14536 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14537 (cpu_expand_builtin): Add support for target_clone attribute.
14538 (rs6000_valid_attribute_p): Allow "default" attribute.
14539 (get_decl_name): New debug function to simplify printing the
14540 current function name in debugging statements.
14541 (rs6000_clone_priority): New functions to support the target_clone
14542 attribute, and be able to generate code to switch between ISA 2.05
14543 through ISA 3.0 (power6 through power9).
14544 (rs6000_compare_version_priority): Likewise.
14545 (rs6000_get_function_versions_dispatcher): Likewise.
14546 (make_resolver_func): Likewise.
14547 (add_condition_to_bb): Likewise.
14548 (dispatch_function_versions): Likewise.
14549 (rs6000_generate_version_dispatcher_body): Likewise.
14550 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14551 (fusion_gpr_load_p): Fix a spacing issue.
14552 * doc/extend.texi (Common Function Attributes): Document that the
14553 PowerPC supports the target_clone attribute.
14555 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14557 * config/arm/arm.h: explain F symbol found in description of ARM
14558 register allocation in its legend.
14560 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14562 * config/mips/frame-header-opt.c: Include profile-count.h.
14563 * config/riscv/riscv.c: Include profile-count.h
14565 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14567 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14569 (sm_set_flag_if_changed): Add bbs field.
14570 (execute_sm_if_changed_flag_set): Pass BBS.
14571 (execute_sm): Update.
14573 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14575 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14578 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14580 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14581 (peephole2): New peephole2 to emit the above.
14582 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14584 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14586 * config/aarch64/aarch64.c (define_peephole2 above
14587 *sub_<shift>_<mode>): New peephole.
14589 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14591 * config/i386/i386.c (make_resolver_func): Update.
14592 * Makefile.in: Add profile-count.h and profile-count.o
14593 * auto-profile.c (afdo_indirect_call): Update to new API.
14594 (afdo_set_bb_count): Update.
14595 (afdo_propagate_edge): Update.
14596 (afdo_propagate_circuit): Update.
14597 (afdo_calculate_branch_prob): Update.
14598 (afdo_annotate_cfg): Update.
14599 * basic-block.h: Include profile-count.h
14600 (struct edge_def): Turn count to profile_count.
14601 (struct basic_block_def): Likewie.
14602 (REG_BR_PROB_BASE): Move to profile-count.h
14603 (RDIV): Move to profile-count.h
14604 * bb-reorder.c (max_entry_count): Turn to profile_count.
14605 (find_traces): Update.
14606 (rotate_loop):Update.
14607 (connect_traces):Update.
14608 (sanitize_hot_paths):Update.
14609 * bt-load.c (migrate_btr_defs): Update.
14610 * cfg.c (RDIV): Remove.
14611 (init_flow): Use alloc_block.
14612 (alloc_block): Uninitialize count.
14613 (unchecked_make_edge): Uninitialize count.
14614 (check_bb_profile): Update.
14615 (dump_edge_info): Update.
14616 (dump_bb_info): Update.
14617 (update_bb_profile_for_threading): Update.
14618 (scale_bbs_frequencies_int): Update.
14619 (scale_bbs_frequencies_gcov_type): Update.
14620 (scale_bbs_frequencies_profile_count): New.
14621 * cfg.h (update_bb_profile_for_threading): Update.
14622 (scale_bbs_frequencies_profile_count): Declare.
14623 * cfgbuild.c (compute_outgoing_frequencies): Update.
14624 (find_many_sub_basic_blocks): Update.
14625 * cfgcleanup.c (try_forward_edges): Update.
14626 (try_crossjump_to_edge): Update.
14627 * cfgexpand.c (expand_gimple_tailcall): Update.
14628 (construct_exit_block): Update.
14629 * cfghooks.c (verify_flow_info): Update.
14630 (dump_bb_for_graph): Update.
14631 (split_edge): Update.
14632 (make_forwarder_block): Update.
14633 (duplicate_block): Update.
14634 (account_profile_record): Update.
14635 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14636 (get_estimated_loop_iterations): Update.
14637 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14638 (single_likely_exit): Update.
14639 * cfgloopmanip.c (scale_loop_profile): Update.
14641 (set_zero_probability): Update.
14642 (lv_adjust_loop_entry_edge): Update.
14643 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14644 (purge_dead_edges): Update.
14645 (rtl_account_profile_record): Update.
14646 * cgraph.c (cgraph_node::create): Uninitialize count.
14647 (symbol_table::create_edge): Uninitialize count.
14648 (cgraph_update_edges_for_call_stmt_node): Update.
14649 (cgraph_edge::dump_edge_flags): Update.
14650 (cgraph_node::dump): Update.
14651 (cgraph_edge::maybe_hot_p): Update.
14652 * cgraph.h: Include profile-count.h
14653 (create_clone), create_edge, create_indirect_edge): Update.
14654 (cgraph_node): Turn count to profile_count.
14655 (cgraph_edge0: Likewise.
14656 (make_speculative, clone): Update.
14657 (create_edge): Update.
14658 (init_lowered_empty_function): Update.
14659 * cgraphclones.c (cgraph_edge::clone): Update.
14660 (duplicate_thunk_for_node): Update.
14661 (cgraph_node::create_clone): Update.
14662 * cgraphunit.c (cgraph_node::analyze): Update.
14663 (cgraph_node::expand_thunk): Update.
14664 * final.c (dump_basic_block_info): Update.
14665 * gimple-streamer-in.c (input_bb): Update.
14666 * gimple-streamer-out.c (output_bb): Update.
14667 * graphite.c (print_global_statistics): Update.
14668 (print_graphite_scop_statistics): Update.
14669 * hsa-brig.c: Include basic-block.h.
14670 * hsa-dump.c: Include basic-block.h.
14671 * hsa-gen.c (T sum_slice): Update.
14672 (convert_switch_statements):Update.
14673 * hsa-regalloc.c: Include basic-block.h.
14674 * ipa-chkp.c (chkp_produce_thunks): Update.
14675 * ipa-cp.c (struct caller_statistics): Update.
14676 (init_caller_stats): Update.
14677 (gather_caller_stats): Update.
14678 (ipcp_cloning_candidate_p): Update.
14679 (good_cloning_opportunity_p): Update.
14680 (get_info_about_necessary_edges): Update.
14681 (dump_profile_updates): Update.
14682 (update_profiling_info): Update.
14683 (update_specialized_profile): Update.
14684 (perhaps_add_new_callers): Update.
14685 (decide_about_value): Update.
14686 (ipa_cp_c_finalize): Update.
14687 * ipa-devirt.c (struct odr_type_warn_count): Update.
14688 (struct decl_warn_count): Update.
14689 (struct final_warning_record): Update.
14690 (possible_polymorphic_call_targets): Update.
14691 (ipa_devirt): Update.
14692 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14693 * ipa-icf.c (sem_function::merge): Update.
14694 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14695 * ipa-inline.c (compute_uninlined_call_time): Update.
14696 (compute_inlined_call_time): Update.
14697 (want_inline_small_function_p): Update.
14698 (want_inline_self_recursive_call_p): Update.
14699 (edge_badness): Update.
14700 (lookup_recursive_calls): Update.
14701 (recursive_inlining): Update.
14702 (inline_small_functions): Update.
14703 (dump_overall_stats): Update.
14704 (dump_inline_stats): Update.
14705 * ipa-profile.c (ipa_profile_generate_summary): Update.
14706 (ipa_propagate_frequency): Update.
14707 (ipa_profile): Update.
14708 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14709 * ipa-utils.c (ipa_merge_profiles): Update.
14710 * loop-doloop.c (doloop_modify): Update.
14711 * loop-unroll.c (report_unroll): Update.
14712 (unroll_loop_runtime_iterations): Update.
14713 * lto-cgraph.c (lto_output_edge): Update.
14714 (lto_output_node): Update.
14715 (input_node): Update.
14716 (input_edge): Update.
14717 (merge_profile_summaries): Update.
14718 * lto-streamer-in.c (input_cfg): Update.
14719 * lto-streamer-out.c (output_cfg): Update.
14720 * mcf.c (create_fixup_graph): Update.
14721 (adjust_cfg_counts): Update.
14722 (sum_edge_counts): Update.
14723 * modulo-sched.c (sms_schedule): Update.
14724 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14725 * predict.c (maybe_hot_count_p): Update.
14726 (probably_never_executed): Update.
14727 (dump_prediction): Update.
14728 (combine_predictions_for_bb): Update.
14729 (propagate_freq): Update.
14730 (handle_missing_profiles): Update.
14731 (counts_to_freqs): Update.
14732 (rebuild_frequencies): Update.
14733 (force_edge_cold): Update.
14734 * predict.h: Include profile-count.h
14735 (maybe_hot_count_p, counts_to_freqs): UPdate.
14736 * print-rtl-function.c: Do not include cfg.h
14737 * print-rtl.c: Include basic-block.h
14738 * profile-count.c: New file.
14739 * profile-count.h: New file.
14740 * profile.c (is_edge_inconsistent): Update.
14741 (correct_negative_edge_counts): Update.
14742 (is_inconsistent): Update.
14743 (set_bb_counts): Update.
14744 (read_profile_edge_counts): Update.
14745 (compute_frequency_overlap): Update.
14746 (compute_branch_probabilities): Update; Initialize and deinitialize
14748 (branch_prob): Update.
14749 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14750 (edge_gcov_count): New.
14751 (bb_gcov_count): New.
14752 * shrink-wrap.c (try_shrink_wrapping): Update.
14753 * tracer.c (better_p): Update.
14754 * trans-mem.c (expand_transaction): Update.
14755 (ipa_tm_insert_irr_call): Update.
14756 (ipa_tm_insert_gettmclone_call): Update.
14757 * tree-call-cdce.c: Update.
14758 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14759 (gimple_duplicate_sese_tail): Update.
14760 (gimple_account_profile_record): Update.
14761 (execute_fixup_cfg): Update.
14762 * tree-inline.c (copy_bb): Update.
14763 (copy_edges_for_bb): Update.
14764 (initialize_cfun): Update.
14765 (freqs_to_counts): Update.
14766 (copy_cfg_body): Update.
14767 (expand_call_inline): Update.
14768 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14769 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14770 (try_unroll_loop_completely): Update.
14771 (try_peel_loop): Update.
14772 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14773 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14774 * tree-ssa-loop-split.c (connect_loops): Update.
14775 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14776 * tree-ssa-reassoc.c (branch_fixup): Update.
14777 * tree-ssa-tail-merge.c (replace_block_by): Update.
14778 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14779 (compute_path_counts): Update.
14780 (update_profile): Update.
14781 (recompute_probabilities): Update.
14782 (update_joiner_offpath_counts): Update.
14783 (estimated_freqs_path): Update.
14784 (freqs_to_counts_path): Update.
14785 (clear_counts_path): Update.
14786 (ssa_fix_duplicate_block_edges): Update.
14787 (duplicate_thread_path): Update.
14788 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14789 (struct switch_conv_info): Update.
14790 * tree-tailcall.c (decrease_profile): Update.
14791 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14792 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14793 * value-prof.c (check_counter): Update.
14794 (gimple_divmod_fixed_value): Update.
14795 (gimple_mod_pow2): Update.
14796 (gimple_mod_subtract): Update.
14797 (gimple_ic_transform): Update.
14798 (gimple_stringop_fixed_value): Update.
14799 * value-prof.h (gimple_ic): Update.
14801 2017-06-02 Carl Love <cel@us.ibm.com>
14803 * config/rs6000/rs6000-c: Add support for built-in functions
14804 vector double vec_doublee (vector signed int);
14805 vector double vec_doublee (vector unsigned int);
14806 vector double vec_doublee (vector float);
14807 vector double vec_doubleh (vector signed int);
14808 vector double vec_doubleh (vector unsigned int);
14809 vector double vec_doubleh (vector float);
14810 vector double vec_doublel (vector signed int);
14811 vector double vec_doublel (vector unsigned int);
14812 vector double vec_doublel (vector float);
14813 vector double vec_doubleo (vector signed int);
14814 vector double vec_doubleo (vector unsigned int);
14815 vector double vec_doubleo (vector float);.
14816 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14817 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14819 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14820 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14821 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14823 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14824 vec_doublel, vec_doubleh.
14825 * doc/extend.texi: Update the built-in documentation file for the
14826 new built-in functions.
14828 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14831 * ipa-inline-analysis.c (free_growth_caches): Set
14832 edge_removal_hook_holder to NULL after removing it.
14834 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14836 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14837 comparision with zero.
14839 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14840 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14841 for early expansion of vec_min and vec_max builtins.
14842 (builtin_function_type): Add min/max unsigned variants to those
14843 identified as having unsigned arguments.
14845 2017-06-02 Olivier Hainque <hainque@adacore.com>
14847 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14849 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14851 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14852 Use VALL_F16 iterator rather than VALL.
14854 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14856 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14857 Emit CBNZ inside loop when doing a strong exchange and comparing
14858 against zero. Generate the CC flags after the loop.
14860 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14862 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14863 (dl_section_ref): New.
14864 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14865 On AIX, append an expression to subtract the size of the
14866 section length to dl_section_ref.
14868 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14870 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14871 for early expansion of vector absolute builtins.
14873 2017-06-02 Richard Biener <rguenther@suse.de>
14875 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14876 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14878 2017-06-02 Richard Biener <rguenther@suse.de>
14880 PR tree-optimization/80948
14881 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14882 stmt order as well.
14884 2017-06-02 Richard Biener <rguenther@suse.de>
14886 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14888 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14889 for inductions as relevant.
14891 2017-06-02 Richard Biener <rguenther@suse.de>
14893 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14894 (vectorizable_induction): ... this. Remove dead code.
14896 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14898 * builtins. (expand_builtin_alloca): Remove second parameter and
14899 infer its value from the first parameter instead.
14900 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14902 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14904 PR rtl-optimization/80903
14905 * loop-doloop.c (add_test): Unshare sequence.
14907 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14909 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14911 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14913 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14915 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14916 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14917 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14918 (xlogue_layout::get_stub_rtx): Make static.
14919 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14920 (xlogue_layout::compute_stub_managed_regs): Rename to...
14921 (xlogue_layout::count_stub_managed_regs): ...this.
14922 (xlogue_layout::is_stub_managed_reg): New function.
14923 (xlogue_layout::m_stub_names): Rename to...
14924 (xlogue_layout::s_stub_names): ...this, make static.
14925 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14926 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14927 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14928 xlogue_layout::s_stub_names): Instantiate statics.
14929 (stub_managed_regs): Remove.
14930 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14931 (disable_call_ms2sysv_xlogues): Rename to...
14932 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14933 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14935 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14936 change after reload_completed.
14937 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14939 (ix86_expand_prologue): Likewise.
14940 (ix86_expand_epilogue): Likewise.
14941 (ix86_expand_split_stack_prologue): Likewise.
14942 (ix86_compute_frame_layout): Remove frame parameter ...
14943 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14944 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14946 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14947 (ix86_frame): Move from here ...
14948 * config/i386/i386.h (ix86_frame): ... to here.
14949 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14950 complete ix86_frame data structure instead. Remove some_ld_name.
14952 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14954 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14955 symbols that hold a DECL_VALUE_EXPR.
14957 2017-06-01 Martin Jambor <mjambor@suse.cz>
14959 PR tree-optimization/80898
14960 * tree-sra.c (process_subtree_disqualification): Removed.
14961 (disqualify_candidate): Do not acll
14962 process_subtree_disqualification.
14963 (subtree_mark_written_and_enqueue): New function.
14964 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14965 RHS has been disqualified and re-queue LHS if necessary. Apart
14966 from that, ignore disqualified RHS.
14968 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14970 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14971 address fetch for z10 or later.
14973 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14975 * config/arc/arc.md (tst_movb): Add guard when splitting.
14977 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14979 * config/arc/arc.c (arc_can_eliminate): Test against
14980 arc_frame_pointer_needed.
14982 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14984 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14985 to prevent store reordering.
14986 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14987 (type): Add block type.
14988 (stack_tie): Define special instruction to be used in
14991 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14993 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14994 constraint. It is not valid for the pattern.
14995 (noncommutative_binary_comparison): Likewise.
14997 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14999 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15002 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15004 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15005 be used by the reg-alloc.
15007 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15009 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15010 reg-alloc when having mul64 or mul32x16 instructions.
15011 (mulsidi3): Likewise.
15012 (umulsidi3): Likewise.
15013 (mulsi32x16): New pattern.
15014 (mulsi64): Likewise.
15015 (mulsidi64): Likewise.
15016 (umulsidi64): Likewise.
15017 (MUL32x16_REG): Define.
15018 (mul64_600): Use MUL32x16_REG.
15019 (mac64_600): Likewise.
15020 (umul64_600): Likewise.
15021 (umac64_600): Likewise.
15023 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15025 * config/arc/arc.md (mulsi3_700): Make it commutative.
15027 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15029 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15031 (*sign_extendsidi2_insn): Likewise for movstosw.
15033 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15035 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15036 the type of the input discriminant value. Convert the
15037 discriminant value of signedness vary.
15039 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15041 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15042 Add to -Wall section.
15044 2017-06-01 Richard Biener <rguenther@suse.de>
15046 PR middle-end/66313
15047 * fold-const.c (fold_plusminus_mult_expr): If the factored
15048 factor may be zero use a wrapping type for the inner operation.
15049 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15050 and handle moved defs.
15051 (process_assignment): Properly guard the unary op case. Return a
15052 tri-state indicating that moving the stmt before the call may allow
15053 to continue. Pass through to_move.
15054 (find_tail_calls): Handle moving unrelated defs before
15057 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15060 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15061 splitter result in the canonical way.
15063 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15065 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15066 also for 32bit target. Update insn attributes.
15067 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15069 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15071 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15072 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15073 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15074 (_mm_maskz_min_ss): New intrinsics.
15076 2017-05-31 Martin Liska <mliska@suse.cz>
15078 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15079 Change comment style to one we normally use.
15080 (vectorizable_reduction): Likewise.
15081 (vectorizable_induction): Likewise.
15082 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15083 (vectorizable_call): Likewise.
15084 (vectorizable_simd_clone_call): Likewise.
15085 (vectorizable_conversion): Likewise.
15086 (vectorizable_assignment): Likewise.
15087 (vectorizable_shift): Likewise.
15088 (vectorizable_operation): Likewise.
15089 (vectorizable_store): Likewise.
15090 (vectorizable_load): Likewise.
15091 * tree-vectorizer.h: Likewise.
15093 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
15095 * passes.c (emergency_dump_function): New.
15096 * tree-pass.h (emergency_dump_function): Declare.
15097 * plugin.c (plugins_internal_error_function): Remove.
15098 * plugin.h (plugins_internal_error_function): Remove declaration.
15099 * toplev.c (internal_error_function): New static function. Use it...
15100 (general_init): ...here.
15102 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15104 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15105 (arc_rtx_costs): Add costs for new patterns.
15106 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15107 * config/arc/predicates.md: Add _1_2_3_operand predicate.
15109 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15111 * tree-ssa-strlen.c (get_next_strinfo): New function.
15112 (get_stridx_plus_constant): Use it.
15113 (zero_length_string): Likewise.
15114 (adjust_related_strinfos): Likewise.
15115 (adjust_last_stmt): Likewise.
15117 2017-05-31 Richard Biener <rguenther@suse.de>
15120 * config/i386/i386.c (ix86_expand_builtin): Remove assert
15121 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
15123 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15125 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
15126 loop_vinfo argument and use of dependence distance vectors.
15127 Check instead whether the two references differ only in their
15128 initial value and assume that they have the same alignment if the
15129 difference is a multiple of the vector alignment.
15130 (vect_analyze_data_refs_alignment): Update call accordingly.
15132 2017-05-31 Martin Liska <mliska@suse.cz>
15135 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15137 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15139 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15140 (create_intersect_range_checks): Move from ...
15141 * tree-data-ref.c (create_intersect_range_checks_index)
15142 (create_intersect_range_checks): ... to here.
15143 (create_runtime_alias_checks): New function factored from ...
15144 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15145 here. Call above function.
15146 * tree-data-ref.h (create_runtime_alias_checks): New function.
15148 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15150 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15151 segment length for dr_b and compute it in wide_int.
15153 2017-05-31 Richard Biener <rguenther@suse.de>
15155 PR tree-optimization/80906
15156 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15157 and pass through iv_map.
15158 (copy_bb_and_scalar_dependences): Adjust.
15159 (translate_pending_phi_nodes): Likewise.
15160 (copy_loop_close_phi_args): Handle code-generating IVs instead
15163 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15165 * diagnostic-color.c (color_dict): Add "type-diff".
15166 (parse_gcc_colors): Update comment.
15167 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15168 -fdiagnostics-show-template-tree and -fno-elide-type.
15169 (GCC_COLORS): Add type-diff to example.
15171 (-fdiagnostics-show-template-tree): New.
15172 (-fno-elide-type): New.
15173 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15174 the pp_format_decoder callback. Call any m_format_postprocessor's
15176 (pretty_printer::pretty_printer): Initialize
15177 m_format_postprocessor.
15178 (pretty_printer::~pretty_printer): Delete any
15179 m_format_postprocessor.
15180 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15181 (class format_postprocessor): New class.
15182 (struct pretty_printer::format_decoder): Document the new parameters.
15183 (struct pretty_printer::m_format_postprocessor): New field.
15184 * tree-diagnostic.c (default_tree_printer): Update for new
15185 bool and const char ** params.
15186 * tree-diagnostic.h (default_tree_printer): Likewise.
15188 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15190 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15191 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15192 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15193 rs6000_gen_cell_microcode code.
15194 (rs6000_final_prescan_insn): Delete.
15195 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15196 "warn-cell-microcode" entries.
15197 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15198 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15199 throughout. Change cc_reg_not_micro_cr0_operand to
15200 cc_reg_not_cr0_operand throughout.
15201 (*extendhi<mode>2_noload): Delete.
15202 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15203 (mwarn-cell-microcode): Delete.
15204 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15205 -mgen-cell-microcode and -mwarn-cell-microcode.
15207 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15210 * config/i386/constraints.md (Yd): New constraint.
15212 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15213 and (?Yd, r) alternatives. Update insn attributes.
15214 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15215 and (?*Yd, r) alternatives. Update insn attributes.
15216 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15218 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15220 * gimplify.c (gimplify_modify_expr): Don't create a
15221 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15224 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15226 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15228 2017-05-30 Richard Biener <rguenther@suse.de>
15230 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15231 and reduc_def fields.
15232 (STMT_VINFO_REDUC_TYPE): New define.
15233 (STMT_VINFO_REDUC_DEF): Likewise.
15234 (vect_force_simple_reduction): Adjust prototype.
15235 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15236 (vect_is_simple_reduction): Remove check_reduction argument.
15237 (vect_force_simple_reduction): Adjust and set
15238 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15239 (vectorizable_reduction): Do not re-do reduction analysis
15240 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15241 * tree-parloops.c (gather_scalar_reductions): Adjust.
15243 2017-05-30 Richard Biener <rguenther@suse.de>
15245 PR middle-end/80901
15246 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15249 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15251 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15252 Introduce unknown_misalignment parameter and remove vf.
15253 (vect_peeling_hash_get_lowest_cost):
15254 Pass unknown_misalignment parameter.
15255 (vect_enhance_data_refs_alignment):
15256 Fix unsupportable data ref treatment.
15258 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15260 * tree-vect-data-refs.c (vect_get_data_access_cost):
15261 Workaround for SLP handling.
15262 (vect_enhance_data_refs_alignment):
15263 Compute costs for doing no peeling at all, compare to the best
15264 peeling costs so far and avoid peeling if cheaper.
15266 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15268 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15269 Return peeling info and set costs to zero for unlimited cost
15271 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15272 with unknown misalignment. Compute and costs for unknown
15273 misalignment, compare them to the costs for known misalignment
15274 and choose the cheapest for peeling.
15276 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15278 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15279 (vect_get_peeling_costs_all_drs): Create function.
15280 (vect_peeling_hash_get_lowest_cost):
15281 Use vect_get_peeling_costs_all_drs.
15282 (vect_peeling_supportable): Create function.
15283 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15285 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15287 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15288 DR_HAS_NEGATIVE_STEP.
15289 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15290 (vect_enhance_data_refs_alignment): Use.
15291 (vect_duplicate_ssa_name_ptr_info): Use.
15292 * tree-vectorizer.h (dr_misalignment): Use.
15293 (known_alignment_for_access_p): Use.
15295 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15298 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15300 (has_section_name): New function.
15302 2017-05-30 Martin Liska <mliska@suse.cz>
15305 * auto-profile.c (get_function_decl_from_block): Fix
15308 2017-05-30 Richard Biener <rguenther@suse.de>
15310 PR middle-end/80876
15311 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15313 2017-05-30 Martin Liska <mliska@suse.cz>
15315 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15316 * dumpfile.h (struct dump_file_info): Remove ctors.
15318 2017-05-30 Martin Liska <mliska@suse.cz>
15320 * predict.def: Fix GNU coding style.
15322 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15324 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15325 Mark 'to' argument with ATTRIBUTE_UNUSED.
15327 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15329 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15330 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15331 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15334 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15336 * doc/install.texi (Options specification): Restore entry of
15337 --enable-sjlj-exceptions.
15339 2017-05-27 Michael Eager <eager@eagercon.com>
15342 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15344 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15346 * config/microblaze/microblaze.h
15347 (FIXED_REGISTERS): Update in macro.
15348 (CALL_USED_REGISTERS): Update in macro.
15350 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15352 * doc/install.texi: Add links to macOS binary distributions.
15354 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15358 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15360 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15362 2017-05-26 Martin Liska <mliska@suse.cz>
15364 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15366 2017-05-26 Martin Liska <mliska@suse.cz>
15368 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15369 always leading ';; '.
15370 (dump_bb_info): Likewise.
15371 (brief_dump_cfg): Likewise.
15372 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15373 * dumpfile.c: Remove usage of TDF_VERBOSE.
15374 * dumpfile.h (enum dump_kind): Likewise.
15375 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15376 * print-tree.c (debug_verbose): Remove.
15377 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15378 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15379 * tree-diagnostic.c (default_tree_printer): Replace
15380 TDF_DIAGNOSTIC with TDF_SLIM.
15382 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15384 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15385 in parameter loop, rather than loop_vinfo.
15386 (create_intersect_range_checks): Ditto.
15387 (vect_create_cond_for_alias_checks): Update call to above functions.
15389 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15391 PR tree-optimization/80815
15392 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15393 for merging runtime alias checks. Handle negative DR_STEPs.
15395 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15397 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15399 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15400 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15401 out code pruning runtime alias checks.
15402 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15403 factored out from above.
15404 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15406 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15408 (prune_runtime_alias_test_list): New decalaration.
15410 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15412 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15413 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15414 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15415 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15416 (operator==, comp_dr_with_seg_len_pair): Ditto.
15417 (vect_prune_runtime_alias_test_list): Ditto.
15419 2017-05-26 Martin Liska <mliska@suse.cz>
15422 * params.def: Bound partial-inlining-entry-probability param.
15424 2017-05-26 Marek Polacek <polacek@redhat.com>
15427 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15430 2017-05-26 Richard Biener <rguenther@suse.de>
15432 PR tree-optimization/80842
15433 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15436 2017-05-26 Richard Biener <rguenther@suse.de>
15438 PR tree-optimization/80844
15439 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15441 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15443 * doc/md.texi (Machine Constraints): Update x86 family
15444 machine constraints section to match 'config/i386/constraints.md'.
15446 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15448 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15450 2017-05-25 Nathan Sidwell <nathan@acm.org>
15452 * doc/invoke.texi (--enable-languages): Update documentation.
15454 2017-05-25 Martin Liska <mliska@suse.cz>
15456 * dumpfile.c: Add TDF_FOLDING.
15457 * dumpfile.h (enum dump_kind): Likewise.
15458 * genmatch.c (dt_simplify::gen_1): Use it.
15460 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15462 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15464 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15466 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15467 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15469 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15471 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15472 * match.pd (X == C): Rewrite it here.
15473 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15474 with_certain_nonzero_bits2): New predicates.
15475 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15477 2017-05-24 Nathan Sidwell <nathan@acm.org>
15479 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15482 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15485 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15487 * config/powerpcspe: New port. Files are copied from the rs6000
15488 port, with "rs6000" in filenames replaced by "powerpcspe".
15490 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15492 PR rtl-optimization/80754
15493 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15495 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15497 * config/sparc/sparc.md (length): Return the correct value for -mflat
15498 sibcalls to match output_sibcall.
15500 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15504 * config/rs6000/rs6000.c (struct machine_function): Add new field
15506 (rs6000_get_separate_components): Init that field, use it.
15507 (rs6000_components_for_bb): Use the field.
15509 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15511 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15513 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15515 PR middle-end/80823
15516 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15518 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15521 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15522 address against address_operand predicate.
15523 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15525 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15527 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15528 parameters passed indirectly.
15530 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15532 * config/i386/i386.md (*movdi_internal): Remove SSE4
15533 alternative 18 (?r, *v). Update insn attributes.
15534 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15535 Update insn attributes.
15536 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15537 Update insn attributes.
15538 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15539 alternative 1 (r, v). Remove isa attribute.
15540 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15541 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15542 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15544 2017-05-23 Tom de Vries <tom@codesourcery.com>
15546 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15549 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15551 * cgraphunit.c (symbol_table::process_new_functions): Update.
15552 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15553 (inline_generate_summary): Rename to ...
15554 (ipa_fn_summary_generate): ... this one.
15555 (inline_read_summary): Rename to ...
15556 (ipa_fn_summary_read): ... this one.
15557 (inline_write_summary): Rename to ...
15558 (ipa_fn_summary_write): ... this one.
15559 (inline_free_summary): Rename to ...
15560 (ipa_free_fn_summary): ... this one.
15561 (pass_data_local_fn_summary, pass_local_fn_summary,
15562 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15563 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15564 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15565 make_pass_ipa_fn_summary): New.
15566 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15567 inline_write_summary, inline_free_summary): Remove.
15568 (ipa_free_fn_summary) : New.
15569 * ipa-inline.c (ipa_inline): Update.
15570 (pass_ipa_inline): Do not generate summaries.
15571 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15573 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15574 and add pass_ipa_fn_summary.
15575 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15577 (make_pass_inline_parameters): Remove.
15579 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15581 * omp-low.c (struct omp_context): Remove "default_kind" member.
15584 * omp-offload.c (execute_oacc_device_lower): Remove the
15585 parallelism dimensions function attributes for unparallelized
15586 OpenACC kernels constructs.
15588 2017-05-23 Martin Liska <mliska@suse.cz>
15590 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15592 (cgraph_edge::make_speculative): Likewise.
15593 (cgraph_edge::resolve_speculation): Likewise.
15594 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15595 (cgraph_node::dump): Likewise.
15596 * cgraph.h: Likewise.
15597 * cgraphunit.c (analyze_functions): Likewise.
15598 (symbol_table::compile): Likewise.
15599 * ipa-cp.c (print_all_lattices): Likewise.
15600 (determine_versionability): Likewise.
15601 (initialize_node_lattices): Likewise.
15602 (ipcp_verify_propagated_values): Likewise.
15603 (estimate_local_effects): Likewise.
15604 (update_profiling_info): Likewise.
15605 (create_specialized_node): Likewise.
15606 (perhaps_add_new_callers): Likewise.
15607 (decide_about_value): Likewise.
15608 (decide_whether_version_node): Likewise.
15609 (identify_dead_nodes): Likewise.
15610 (ipcp_store_bits_results): Likewise.
15611 * ipa-devirt.c (dump_targets): Likewise.
15612 (ipa_devirt): Likewise.
15613 * ipa-icf.c (sem_item::dump): Likewise.
15614 (sem_function::equals): Likewise.
15615 (sem_variable::equals): Likewise.
15616 (sem_item_optimizer::read_section): Likewise.
15617 (sem_item_optimizer::execute): Likewise.
15618 (congruence_class::dump): Likewise.
15619 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15620 (dump_inline_summary): Likewise.
15621 (estimate_node_size_and_time): Likewise.
15622 (inline_analyze_function): Likewise.
15623 * ipa-inline-transform.c (inline_call): Likewise.
15624 * ipa-inline.c (report_inline_failed_reason): Likewise.
15625 (want_early_inline_function_p): Likewise.
15626 (edge_badness): Likewise.
15627 (update_edge_key): Likewise.
15628 (inline_small_functions): Likewise.
15629 * ipa-profile.c (ipa_profile): Likewise.
15630 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15631 (ipa_make_edge_direct_to_target): Likewise.
15632 (remove_described_reference): Likewise.
15633 (ipa_impossible_devirt_target): Likewise.
15634 (propagate_controlled_uses): Likewise.
15635 (ipa_print_node_params): Likewise.
15636 (ipcp_transform_function): Likewise.
15637 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15638 (propagate_pure_const): Likewise.
15639 * ipa-reference.c (generate_summary): Likewise.
15640 (read_write_all_from_decl): Likewise.
15641 (propagate): Likewise.
15642 (ipa_reference_read_optimization_summary): Likewise.
15643 * ipa-utils.c (ipa_merge_profiles): Likewise.
15644 * ipa.c (walk_polymorphic_call_targets): Likewise.
15645 (symbol_table::remove_unreachable_nodes): Likewise.
15646 (ipa_single_use): Likewise.
15647 * passes.c (execute_todo): Likewise.
15648 * predict.c (drop_profile): Likewise.
15649 * symtab.c (symtab_node::get_dump_name): New function.
15650 (symtab_node::dump_name): Likewise.
15651 (symtab_node::dump_asm_name): Likewise.
15652 (symtab_node::dump_references): Likewise.
15653 (symtab_node::dump_referring): Likewise.
15654 (symtab_node::dump_base): Likewise.
15655 (symtab_node::debug_symtab): Likewise.
15656 * tree-sra.c (convert_callers_for_node): Likewise.
15657 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15658 * value-prof.c (init_node_map): Likewise.
15660 2017-05-23 Martin Liska <mliska@suse.cz>
15662 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15663 and symtab_node::debug_symtab to symbol_table::debug.
15664 * cgraphunit.c (analyze_functions): Use the renamed function.
15665 (symbol_table::compile): Likewise.
15666 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15667 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15668 * passes.c (execute_todo): Likewise.
15669 * symtab.c (symbol_table::dump): New function.
15670 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15672 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15674 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15675 that nonconst implies exec.
15677 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15679 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15680 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15681 (inline_edge_summary_vec): Turn into ...
15682 (ipa_call_summaries): ... this one.
15683 (redirect_to_unreachable, edge_set_predicate,
15684 evaluate_properties_for_edge, inline_summary_alloc,
15685 reset_ipa_call_summary, reset_inline_summary,
15686 inline_summary_t::duplicate): Update.
15687 (inline_edge_duplication_hook): Turn to ...
15688 (ipa_call_summary_t::duplicate): ... this one.
15689 (inline_edge_removal_hook): Turn to ...
15690 (ipa_call_summary_t::remove): ... this one.
15691 (dump_inline_edge_summary): Turn to ...
15692 (dump_ipa_call_summary): ... this one.
15693 (estimate_function_body_sizes): Update.
15694 (inline_update_callee_summaries): Update.
15695 (remap_edge_change_prob): Update.
15696 (remap_edge_summaries): Update.
15697 (inline_merge_summary): Update.
15698 (do_estimate_edge_time): Update.
15699 (inline_generate_summary): Update.
15700 (inline_read_section): Update.
15701 (inline_read_summary): Update.
15702 (inline_free_summary): Update.
15703 * ipa-inline.c (can_inline_edge_p): Update.
15704 (compute_inlined_call_time): Update.
15705 (want_inline_small_function_p): Update.
15706 (edge_badness): Update.
15707 (early_inliner): Update.
15708 * ipa-inline.h (inline_edge_summary): Turn to ...
15709 (ipa_call_summary): ... this one.
15710 (ipa_call_summary_t): New class.
15711 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15712 (ipa_call_summaries): New.
15713 (inline_edge_summary): Remove.
15714 (estimate_edge_growth): Update.
15715 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15716 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15717 * ipa-split.c (execute_split_functions): Update.
15718 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15720 2017-05-23 Tom de Vries <tom@codesourcery.com>
15722 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15723 attributes): Document rdrand effective target.
15725 2017-05-23 Tom de Vries <tom@codesourcery.com>
15727 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15728 attributes): Sort alphabetically.
15730 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15732 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15734 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15737 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15738 V2DF/V2DI splat into two separate patterns, one that handles
15739 registers, and the other that only handles memory. Drop support
15740 for splatting from a GPR on ISA 2.07 and then splitting the
15741 splat into direct move and splat.
15742 (vsx_splat_<mode>_reg): Likewise.
15743 (vsx_splat_<mode>_mem): Likewise.
15745 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15747 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15749 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15751 PR middle-end/80809
15752 * omp-low.c (finish_taskreg_remap): New function.
15753 (finish_taskreg_scan): If unit size of ctx->record_type
15754 is non-constant, unshare the size expression and replace
15755 decls in it with possible outer var refs.
15757 PR middle-end/80809
15758 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15759 GOVD_SHARED rather than GOVD_PRIVATE with it.
15760 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15761 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15763 PR middle-end/80853
15764 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15765 as last argument to build_outer_var_ref for pointer bases of array
15766 section reductions.
15768 2017-05-19 Martin Sebor <msebor@redhat.com>
15770 * print-tree.c (print_node): Print DECL_READ_P flag.
15772 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15774 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15775 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15776 * cgraph.c: Likewise.
15777 * cgraphunit.c: Likewise.
15778 * gengtype.c: Likewise.
15779 * ipa-cp.c: Likewise.
15780 * ipa-devirt.c: Likewise.
15781 * ipa-icf.c: Likewise.
15782 * ipa-predicate.c: Likewise.
15783 * ipa-profile.c: Likewise.
15784 * ipa-prop.c: Likewise.
15785 * ipa-split.c: Likewise.
15787 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15788 edge_predicate_pool, dump_inline_hints,
15789 inline_summary::account_size_time, redirect_to_unreachable,
15790 edge_set_predicate, set_hint_predicate,
15791 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15792 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15793 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15794 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15795 ipa_call_summary_t::remove, initialize_growth_caches,
15796 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15797 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15798 mark_modified, unmodified_parm_1, unmodified_parm,
15799 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15800 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15801 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15802 will_be_nonconstant_predicate, record_modified_bb_info,
15803 get_minimal_bb, record_modified, param_change_prob,
15804 phi_result_unknown_predicate, predicate_for_phi_result,
15805 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15806 estimate_function_body_sizes, compute_inline_parameters,
15807 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15808 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15809 inline_update_callee_summaries, remap_edge_change_prob,
15810 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15811 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15812 inline_analyze_function, inline_summary_t::insert,
15813 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15814 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15815 inline_free_summary): Move to ipa-fnsummary.h
15816 (predicate_t): Remove.
15817 * ipa-fnsummary.c: New file.
15818 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15819 (enum inline_hints_vals, inline_hints, agg_position_info,
15820 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15821 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15822 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15823 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15824 inline_read_summary, inline_write_summary, inline_free_summary,
15825 inline_analyze_function, initialize_inline_failed,
15826 inline_merge_summary, inline_update_overall_summary,
15827 compute_inline_parameters): Move to ipa-fnsummary.h
15828 * ipa-fnsummary.h: New file.
15829 * ipa-inline-transform.h: Include ipa-inline.h.
15830 * ipa-inline.c: LIkewise.
15832 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15834 * ipa-inline.c (edge_badness): Use inlined_time instead of
15835 inline_summaries->get.
15837 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15839 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15841 2017-05-22 Nathan Sidwell <nathan@acm.org>
15843 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15844 (fdump-lang): Document 'raw' option.
15845 * dumpfile.h (TDI_tu): Delete.
15846 * dumpfile.c (dump_files): Remove translation-unit.
15847 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15849 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15851 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15852 command option from $(AWK) call.
15853 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15854 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15855 [FORMAT]: Remove handling of variable.
15856 * config/avr/t-multilib: Regenerate.
15858 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15860 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15862 (dump_inline_summary): Do not print self_time.
15863 (estimate_function_body_sizes): Do not set self_time.
15864 (compute_inline_parameters): Likewise.
15865 (inline_read_section, inline_write_summary): Do not stream self_time.
15866 * ipa-inline.h (inline_summary): Drop self_time.
15868 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15870 * ipa-inline-analysis.c (account_size_time): Rename to ...
15871 (inline_summary::account_size_time): ... this one.
15872 (reset_ipa_call_summary): Turn to ...
15873 (ipa_call_summary::reset): ... this one.
15874 (reset_inline_summary): Turn to ...
15875 (inline_summary::reset): ... this one.
15876 (inline_summary_t::remove): Update.
15877 (inline_summary_t::duplicate): Update.
15878 (ipa_call_summary_t::remove): Update.
15879 (dump_inline_summary): Update.
15880 (estimate_function_body_sizes): Update.
15881 (compute_inline_parameters): Update.
15882 (estimate_node_size_and_time): Update.
15883 (inline_merge_summary): Update.
15884 (inline_update_overall_summary): Update.
15885 (inline_read_section): Update.
15886 (inline_write_summary): Update.
15887 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15888 add account_size_time and reset member functions.
15889 (ipa_call_summary): Add reset function.
15890 * ipa-predicate.h (predicate::operator &): Constify.
15892 2017-05-22 Richard Biener <rguenther@suse.de>
15894 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15896 2017-05-19 Jason Merrill <jason@redhat.com>
15898 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15900 2017-05-19 Marek Polacek <polacek@redhat.com>
15903 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15904 TYPE_OVERFLOW_WRAPS checks.
15906 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15908 * tree-core.h (enum omp_clause_default_kind): Add
15909 "OMP_CLAUSE_DEFAULT_PRESENT".
15910 * tree-pretty-print.c (dump_omp_clause): Handle it.
15911 * gimplify.c (enum gimplify_omp_var_data): Add
15912 "GOVD_MAP_FORCE_PRESENT".
15913 (gimplify_adjust_omp_clauses_1): Map it to
15914 "GOMP_MAP_FORCE_PRESENT".
15915 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15917 * gimplify.c (oacc_default_clause): Clarify.
15919 2017-05-19 Nathan Sidwell <nathan@acm.org>
15921 LANG_HOOK_REGISTER_DUMPS
15922 * toplev.c (general_init): Call register dump lang hook.
15923 * doc/invoke.texi: Document -fdump-lang option family.
15924 * dumpfile.c (dump_files): Remove class dump here.
15925 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15926 * dumpfile.h (tree_dump_index): Remove TDI_class.
15927 * langhooks-def.h (lhd_register_dumps): Declare.
15928 (LANG_HOOKS_REGISTER_DUMPS): Define.
15929 (LANG_HOOKS_INITIALIZER): Add it.
15930 * langhooks.c (lhd_register_dumps): Define.
15931 * langhooks.h (struct lang_hooks): Add register_dumps.
15933 2017-05-19 Nathan Sidwell <nathan@acm.org>
15935 * context.h (context::set_passes): New.
15936 * context.c (context::context): Do not create pass manager.
15937 * toplev.c (general_init): Create pass manager here.
15939 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15941 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15942 use this splitter if two add or or instructions would also work for
15943 the constant we want to generate.
15945 2017-05-19 Richard Biener <rguenther@suse.de>
15948 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15949 predicate evaluation.
15951 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15953 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15955 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15958 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15960 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15961 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15962 (inline_edge_summary_vec): Turn into ...
15963 (ipa_call_summaries): ... this one.
15964 (redirect_to_unreachable, edge_set_predicate,
15965 evaluate_properties_for_edge, inline_summary_alloc,
15966 reset_ipa_call_summary, reset_inline_summary,
15967 inline_summary_t::duplicate): Update.
15968 (inline_edge_duplication_hook): Turn to ...
15969 (ipa_call_summary_t::duplicate): ... this one.
15970 (inline_edge_removal_hook): Turn to ...
15971 (ipa_call_summary_t::remove): ... this one.
15972 (dump_inline_edge_summary): Turn to ...
15973 (dump_ipa_call_summary): ... this one.
15974 (estimate_function_body_sizes): Update.
15975 (inline_update_callee_summaries): Update.
15976 (remap_edge_change_prob): Update.
15977 (remap_edge_summaries): Update.
15978 (inline_merge_summary): Update.
15979 (do_estimate_edge_time): Update.
15980 (inline_generate_summary): Update.
15981 (inline_read_section): Update.
15982 (inline_read_summary): Update.
15983 (inline_free_summary): Update.
15984 * ipa-inline.c (can_inline_edge_p): Update.
15985 (compute_inlined_call_time): Update.
15986 (want_inline_small_function_p): Update.
15987 (edge_badness): Update.
15988 (early_inliner): Update.
15989 * ipa-inline.h (inline_edge_summary): Turn to ...
15990 (ipa_call_summary): ... this one.
15991 (ipa_call_summary_t): New class.
15992 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15993 (ipa_call_summaries): New.
15994 (inline_edge_summary): Remove.
15995 (estimate_edge_growth): Update.
15996 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15997 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15998 * ipa-split.c (execute_split_functions): Update.
15999 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16001 2017-05-19 Richard Biener <rguenther@suse.de>
16003 PR middle-end/80764
16004 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16006 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16008 * config/rs6000/rs6000.c (struct machine_function): Add field
16009 fpr_is_wrapped_separately.
16010 (rs6000_get_separate_components): Use 64 components. Handle the
16011 new FPR components.
16012 (rs6000_components_for_bb): Handle the FPR components.
16013 (rs6000_emit_prologue_components): Handle the FPR components.
16014 (rs6000_emit_epilogue_components): Handle the FPR components.
16015 (rs6000_set_handled_components): Handle the FPR components.
16016 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16017 that are already separately shrink-wrapped.
16018 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16019 that are already separately shrink-wrapped.
16021 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16024 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16027 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16028 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16029 cases where the register allocator uses a move and an offsettable
16030 memory operation to/from a FPR register on ISA 2.06/2.07.
16031 (define_peephole2 for Altivec d-form store): Likewise.
16033 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16036 * config/i386/mmx.md (*mov<mode>_internal): Enable
16037 alternatives 11, 12, 13 and 14 also for 32bit targets.
16038 Remove alternatives 15, 16, 17 and 18.
16039 * config/i386/sse.md (vec_concatv2di): Change
16040 alternative (!x, *y) to (x, ?!*Yn).
16042 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16044 * dumpfile.h (enum dump_kind): Remove stray comma.
16046 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16048 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16049 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16050 predicate::num_conditions
16051 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16052 (CHANGED): turn into predicate::changed.
16053 (agg_position_info): Move to ipa-predicate.h
16054 (add_condition, predicate::add_clause, predicate::operator &=,
16055 predicate::or_with, predicate::evaluate, predicate::probability,
16056 dump_condition, dump_clause, predicate::dump,
16057 predicate::remap_after_duplication, predicate::remap_after_inlining,
16058 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16059 (evaluate_conditions_for_known_args): Update.
16060 (set_cond_stmt_execution_predicate): Update.
16061 * ipa-inline.h: Include ipa-predicate.h
16062 (condition, inline_param_summary, conditions, agg_position_info,
16063 predicate): Move to ipa-predicate.h
16064 * ipa-predicate.c: New file.
16065 * ipa-predicate.h: New file.
16067 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16069 * final.c (leaf_function_p): Check we are not in a sequence.
16071 2017-05-18 Martin Liska <mliska@suse.cz>
16073 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16074 * dumpfile.c (dump_register): Use new enum dump_kind.
16075 (get_dump_file_name): Likewise.
16076 (dump_enable_all): Likewise.
16077 (dump_switch_p_1): Likewise.
16078 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16079 * dumpfile.h (enum dump_kind): New enum type.
16080 (struct dump_file_info): Create constructor and
16081 format fields and comments.
16082 * passes.c (pass_manager::register_one_dump_file):
16084 * statistics.c (statistics_early_init): Likewise.
16085 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
16086 TDF_TREE with TDF_SLIM.
16087 (gather_memory_references_ref): Likewise.
16089 2017-05-18 Martin Liska <mliska@suse.cz>
16091 * vec.h (struct vnull): Use it.
16093 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16095 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
16096 (true_predicate, false_predicate, true_predicate_p,
16097 false_predicate_p): Remove.
16098 (single_cond_predicate, not_inlined_predicate): Turn to member function
16100 (add_condition): Update.
16101 (add_clause): Turn to...
16102 (predicate::add_clause): ... this one; update; allow passing NULL
16104 (and_predicates): Turn to ...
16105 (predicate::operator &=): ... this one.
16106 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16107 (or_predicates): Turn to ...
16108 (predicate::or_with): ... this one.
16109 (evaluate_predicate): Turn to ...
16110 (predicate::evaluate): ... this one.
16111 (predicate_probability): Turn to ...
16112 (predicate::probability): ... this one.
16113 (dump_condition): Update.
16114 (dump_predicate): Turn to ...
16115 (predicate::dump): ... this one.
16116 (account_size_time): Update.
16117 (edge_set_predicate): Update.
16118 (set_hint_predicate): UPdate.
16119 (evaluate_conditions_for_known_args): Update.
16120 (evaluate_properties_for_edge): Update.
16121 (remap_predicate_after_duplication): Turn to...
16122 (predicate::remap_after_duplication): ... this one.
16123 (remap_hint_predicate_after_duplication): Update.
16124 (inline_summary_t::duplicate): UPdate.
16125 (dump_inline_edge_summary): Update.
16126 (dump_inline_summary): Update.
16127 (set_cond_stmt_execution_predicate): Update.
16128 (set_switch_stmt_execution_predicate): Update.
16129 (compute_bb_predicates): Update.
16130 (will_be_nonconstant_expr_predicate): Update.
16131 (will_be_nonconstant_predicate): Update.
16132 (phi_result_unknown_predicate): Update.
16133 (predicate_for_phi_result): Update.
16134 (array_index_predicate): Update.
16135 (estimate_function_body_sizes): Update.
16136 (estimate_node_size_and_time): Update.
16137 (estimate_ipcp_clone_size_and_time): Update.
16138 (remap_predicate): Rename to ...
16139 (predicate::remap_after_inlining): ... this one.
16140 (remap_hint_predicate): Update.
16141 (inline_merge_summary): Update.
16142 (inline_update_overall_summary): Update.
16143 (estimate_size_after_inlining): Update.
16144 (read_predicate): Rename to ...
16145 (predicate::stream_in): ... this one.
16146 (read_inline_edge_summary): Update.
16147 (write_predicate): Rename to ...
16148 (predicate::stream_out): ... this one.
16149 (write_inline_edge_summary): Update.
16150 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16151 (clause_t): Turn to uint32_t
16152 (predicate): Turn to class; implement constructor and operators
16154 (size_time_entry): Update.
16155 (inline_summary): Update.
16156 (inline_edge_summary): Update.
16158 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16160 * fold-const.c (fold_binary_loc): Move transformation...
16161 * match.pd (C - X CMP X): ... here.
16163 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16165 * config/sparc/sparc.c (sparc_option_override): Set function
16166 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16167 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16169 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16171 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16173 2017-05-18 Marek Polacek <polacek@redhat.com>
16176 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16177 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16178 gimple_assign_load_p.
16180 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16182 PR middle-end/80692
16183 * real.c (do_compare): Give decimal_do_compare preference over
16184 comparing just the signs.
16186 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16188 * doc/md.texi (Canonicalization of Instructions): Describe the
16189 canonical form of instructions that inherently set a condition
16192 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16194 PR middle-end/80775
16195 * tree-cfg.c: Move deletion of unreachable case statements to after
16196 the merging of consecutive case labels.
16198 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16200 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16201 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16202 restoring of callee-saved registers.
16204 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16206 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16207 * config/visium/visium.c (single_set_and_flags): Likewise.
16208 * config/visium/visium.md (Substitutions): Likewise.
16210 2017-05-17 Martin Liska <mliska@suse.cz>
16212 * cfg.c: Introduce dump_flags_t type and
16213 use it instead of int type.
16215 * cfghooks.c: Likewise.
16216 * cfghooks.h (struct cfg_hooks): Likewise.
16217 * cfgrtl.c: Likewise.
16218 * cfgrtl.h: Likewise.
16219 * cgraph.c (cgraph_node::get_body): Likewise.
16220 * coretypes.h: Likewise.
16221 * domwalk.c: Likewise.
16222 * domwalk.h: Likewise.
16223 * dumpfile.c (struct dump_option_value_info): Likewise.
16224 (dump_enable_all): Likewise.
16225 (dump_switch_p_1): Likewise.
16226 (opt_info_switch_p): Likewise.
16227 * dumpfile.h (enum tree_dump_index): Likewise.
16228 (struct dump_file_info): Likewise.
16229 * genemit.c: Likewise.
16230 * generic-match-head.c: Likewise.
16231 * gengtype.c (open_base_files): Likewise.
16232 * gimple-pretty-print.c: Likewise.
16233 * gimple-pretty-print.h: Likewise.
16234 * graph.c (print_graph_cfg): Likewise.
16235 * graphite-scop-detection.c (dot_all_sese): Likewise.
16236 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16237 * loop-unroll.c (report_unroll): Likewise.
16238 * passes.c (pass_manager::register_one_dump_file): Likewise.
16239 * print-tree.c: Likewise.
16240 * statistics.c: Likewise.
16241 * tree-cfg.c: Likewise.
16242 * tree-cfg.h: Likewise.
16243 * tree-dfa.c: Likewise.
16244 * tree-dfa.h: Likewise.
16245 * tree-dump.c (dump_function): Likewise.
16246 * tree-dump.h (struct dump_info): Likewise.
16247 * tree-pretty-print.c: Likewise.
16248 * tree-pretty-print.h: Likewise.
16249 * tree-ssa-live.c: Likewise.
16250 * tree-ssa-live.h: Likewise.
16251 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16252 * tree-vect-loop.c: Likewise.
16253 * tree-vect-slp.c: Likewise.
16255 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16256 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16258 PR tree-optimization/80457
16259 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16260 of all arguments to a statement as scalar_to_vec operations.
16261 (vectorizable_call): Adjust call to vect_model_simple_cost for
16263 (vectorizable_conversion): Likewise.
16264 (vectorizable_assignment): Likewise.
16265 (vectorizable_shift): Likewise.
16266 (vectorizable_operation): Likewise.
16267 (vectorizable_comparison): Likewise.
16268 (vect_is_simple_cond): Record the def types for operands.
16269 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16270 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16271 for statement argument count.
16273 2017-05-16 Carl Love <cel@us.ibm.com>
16275 * config/rs6000/rs6000-c: Add support for built-in functions
16276 vector unsigned long long vec_bperm (vector unsigned long long,
16277 vector unsigned char)
16278 vector signed long long vec_mule (vector signed int,
16280 vector unsigned long long vec_mule (vector unsigned int,
16281 vector unsigned int)
16282 vector signed long long vec_mulo (vector signed int,
16284 vector unsigned long long vec_mulo (vector unsigned int,
16285 vector unsigned int)
16286 vector signed char vec_sldw (vector signed char,
16287 vector signed char,
16289 vector unsigned char vec_sldw (vector unsigned char,
16290 vector unsigned char,
16292 vector signed short vec_sldw (vector signed short,
16293 vector signed short,
16295 vector unsigned short vec_sldw (vector unsigned short,
16296 vector unsigned short,
16298 vector signed int vec_sldw (vector signed int,
16301 vector unsigned int vec_sldw (vector unsigned int,
16302 vector unsigned int,
16304 vector signed long long vec_sldw (vector signed long long,
16305 vector signed long long,
16307 vector unsigned long long vec_sldw (vector unsigned long long,
16308 vector unsigned long long,
16310 * config/rs6000/rs6000-c: Add support for built-in functions
16311 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16312 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16313 * doc/extend.texi: Update the built-in documentation for the
16314 new built-in functions.
16316 2017-05-16 Marek Polacek <polacek@redhat.com>
16320 * tree.c (save_expr): Don't fold the expression.
16322 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16324 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16325 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16326 and (?*y,m). Update insn attributes.
16328 2017-05-16 Martin Liska <mliska@suse.cz>
16330 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16331 flags argument of print_gimple_stmt, print_gimple_expr,
16332 print_generic_stmt and print_generic_expr.
16333 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16334 * coretypes.h: Likewise.
16335 * except.c (dump_eh_tree): Likewise.
16336 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16337 * gimple-pretty-print.h: Likewise.
16338 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16339 (backprop::push_to_worklist): Likewise.
16340 (backprop::pop_from_worklist): Likewise.
16341 (backprop::process_use): Likewise.
16342 (backprop::intersect_uses): Likewise.
16343 (note_replacement): Likewise.
16344 * gimple-ssa-store-merging.c
16345 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16346 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16347 (pass_store_merging::execute): Likewise.
16348 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16349 (ssa_base_cand_dump_callback): Likewise.
16350 (dump_incr_vec): Likewise.
16351 (replace_refs): Likewise.
16352 (replace_mult_candidate): Likewise.
16353 (create_add_on_incoming_edge): Likewise.
16354 (create_phi_basis): Likewise.
16355 (insert_initializers): Likewise.
16356 (all_phi_incrs_profitable): Likewise.
16357 (introduce_cast_before_cand): Likewise.
16358 (replace_one_candidate): Likewise.
16359 * gimplify.c (gimplify_expr): Likewise.
16360 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16361 (set_rename): Likewise.
16362 (rename_uses): Likewise.
16363 (copy_loop_phi_nodes): Likewise.
16364 (add_close_phis_to_merge_points): Likewise.
16365 (copy_loop_close_phi_args): Likewise.
16366 (copy_cond_phi_args): Likewise.
16367 (graphite_copy_stmts_from_block): Likewise.
16368 (translate_pending_phi_nodes): Likewise.
16369 * graphite-poly.c (print_pdr): Likewise.
16370 (dump_gbb_cases): Likewise.
16371 (dump_gbb_conditions): Likewise.
16372 (print_scop_params): Likewise.
16373 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16374 (build_cross_bb_scalars_use): Likewise.
16375 (gather_bbs::before_dom_children): Likewise.
16376 * hsa-dump.c (dump_hsa_immed): Likewise.
16377 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16378 (get_replacement_map): Likewise.
16379 * ipa-inline-analysis.c (dump_condition): Likewise.
16380 (estimate_function_body_sizes): Likewise.
16381 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16382 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16383 * ipa-prop.c (ipa_dump_param): Likewise.
16384 (ipa_print_node_jump_functions_for_edge): Likewise.
16385 (ipa_modify_call_arguments): Likewise.
16386 (ipa_modify_expr): Likewise.
16387 (ipa_dump_param_adjustments): Likewise.
16388 (ipa_dump_agg_replacement_values): Likewise.
16389 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16390 * ipa-pure-const.c (check_stmt): Likewise.
16391 (pass_nothrow::execute): Likewise.
16392 * ipa-split.c (execute_split_functions): Likewise.
16393 * omp-offload.c (dump_oacc_loop_part): Likewise.
16394 (dump_oacc_loop): Likewise.
16395 * trans-mem.c (tm_log_emit): Likewise.
16396 (tm_memopt_accumulate_memops): Likewise.
16397 (dump_tm_memopt_set): Likewise.
16398 (dump_tm_memopt_transform): Likewise.
16399 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16400 (print_loop): Likewise.
16401 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16402 (chkp_gather_checks_info): Likewise.
16403 (chkp_get_check_result): Likewise.
16404 (chkp_remove_check_if_pass): Likewise.
16405 (chkp_use_outer_bounds_if_possible): Likewise.
16406 (chkp_reduce_bounds_lifetime): Likewise.
16407 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16408 (chkp_mark_completed_bounds): Likewise.
16409 (chkp_register_incomplete_bounds): Likewise.
16410 (chkp_mark_invalid_bounds): Likewise.
16411 (chkp_maybe_copy_and_register_bounds): Likewise.
16412 (chkp_build_returned_bound): Likewise.
16413 (chkp_get_bound_for_parm): Likewise.
16414 (chkp_build_bndldx): Likewise.
16415 (chkp_get_bounds_by_definition): Likewise.
16416 (chkp_generate_extern_var_bounds): Likewise.
16417 (chkp_get_bounds_for_decl_addr): Likewise.
16418 * tree-chrec.c (chrec_apply): Likewise.
16419 * tree-data-ref.c (dump_data_reference): Likewise.
16420 (dump_subscript): Likewise.
16421 (dump_data_dependence_relation): Likewise.
16422 (analyze_overlapping_iterations): Likewise.
16423 * tree-inline.c (expand_call_inline): Likewise.
16424 (tree_function_versioning): Likewise.
16425 * tree-into-ssa.c (dump_defs_stack): Likewise.
16426 (dump_currdefs): Likewise.
16427 (dump_names_replaced_by): Likewise.
16428 (dump_update_ssa): Likewise.
16429 (update_ssa): Likewise.
16430 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16431 * tree-parloops.c (build_new_reduction): Likewise.
16432 (try_create_reduction_list): Likewise.
16433 (ref_conflicts_with_region): Likewise.
16434 (oacc_entry_exit_ok_1): Likewise.
16435 (oacc_entry_exit_single_gang): Likewise.
16436 * tree-pretty-print.h: Likewise.
16437 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16438 (get_scalar_evolution): Likewise.
16439 (add_to_evolution): Likewise.
16440 (get_loop_exit_condition): Likewise.
16441 (analyze_evolution_in_loop): Likewise.
16442 (analyze_initial_condition): Likewise.
16443 (analyze_scalar_evolution): Likewise.
16444 (instantiate_scev): Likewise.
16445 (number_of_latch_executions): Likewise.
16446 (gather_chrec_stats): Likewise.
16447 (final_value_replacement_loop): Likewise.
16448 (scev_const_prop): Likewise.
16449 * tree-sra.c (dump_access): Likewise.
16450 (disqualify_candidate): Likewise.
16451 (create_access): Likewise.
16452 (reject): Likewise.
16453 (maybe_add_sra_candidate): Likewise.
16454 (create_access_replacement): Likewise.
16455 (analyze_access_subtree): Likewise.
16456 (analyze_all_variable_accesses): Likewise.
16457 (sra_modify_assign): Likewise.
16458 (initialize_constant_pool_replacements): Likewise.
16459 (find_param_candidates): Likewise.
16460 (decide_one_param_reduction): Likewise.
16461 (replace_removed_params_ssa_names): Likewise.
16462 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16463 * tree-ssa-copy.c (dump_copy_of): Likewise.
16464 (copy_prop_visit_cond_stmt): Likewise.
16465 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16466 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16467 (record_equivalences_from_stmt): Likewise.
16468 * tree-ssa-dse.c (compute_trims): Likewise.
16469 (delete_dead_call): Likewise.
16470 (delete_dead_assignment): Likewise.
16471 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16472 (forward_propagate_into_cond): Likewise.
16473 (pass_forwprop::execute): Likewise.
16474 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16475 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16477 (move_computations_worker): Likewise.
16478 (execute_sm): Likewise.
16479 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16480 (remove_exits_and_undefined_stmts): Likewise.
16481 (remove_redundant_iv_tests): Likewise.
16482 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16483 (adjust_iv_update_pos): Likewise.
16484 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16485 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16486 (value_replacement): Likewise.
16487 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16488 * tree-ssa-pre.c (print_pre_expr): Likewise.
16489 (get_representative_for): Likewise.
16490 (create_expression_by_pieces): Likewise.
16491 (insert_into_preds_of_block): Likewise.
16492 (eliminate_insert): Likewise.
16493 (eliminate_dom_walker::before_dom_children): Likewise.
16494 (eliminate): Likewise.
16495 (remove_dead_inserted_code): Likewise.
16496 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16497 * tree-ssa-reassoc.c (get_rank): Likewise.
16498 (eliminate_duplicate_pair): Likewise.
16499 (eliminate_plus_minus_pair): Likewise.
16500 (eliminate_not_pairs): Likewise.
16501 (undistribute_ops_list): Likewise.
16502 (eliminate_redundant_comparison): Likewise.
16503 (update_range_test): Likewise.
16504 (optimize_range_tests_var_bound): Likewise.
16505 (optimize_vec_cond_expr): Likewise.
16506 (rewrite_expr_tree): Likewise.
16507 (rewrite_expr_tree_parallel): Likewise.
16508 (linearize_expr): Likewise.
16509 (break_up_subtract): Likewise.
16510 (linearize_expr_tree): Likewise.
16511 (attempt_builtin_powi): Likewise.
16512 (attempt_builtin_copysign): Likewise.
16513 (transform_stmt_to_copy): Likewise.
16514 (transform_stmt_to_multiply): Likewise.
16515 (dump_ops_vector): Likewise.
16516 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16517 (print_scc): Likewise.
16518 (set_ssa_val_to): Likewise.
16519 (visit_reference_op_store): Likewise.
16520 (visit_use): Likewise.
16521 (sccvn_dom_walker::before_dom_children): Likewise.
16522 (run_scc_vn): Likewise.
16523 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16525 (expr_hash_elt::print): Likewise.
16526 (const_and_copies::pop_to_marker): Likewise.
16527 (const_and_copies::record_const_or_copy_raw): Likewise.
16528 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16529 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16530 (dump_predicates): Likewise.
16531 (find_uninit_use): Likewise.
16532 (warn_uninitialized_phi): Likewise.
16533 (pass_late_warn_uninitialized::execute): Likewise.
16534 * tree-ssa.c (verify_vssa): Likewise.
16535 (verify_ssa): Likewise.
16536 (maybe_optimize_var): Likewise.
16537 * tree-vrp.c (dump_value_range): Likewise.
16538 (dump_all_value_ranges): Likewise.
16539 (dump_asserts_for): Likewise.
16540 (register_edge_assert_for_2): Likewise.
16541 (vrp_visit_cond_stmt): Likewise.
16542 (vrp_visit_switch_stmt): Likewise.
16543 (vrp_visit_stmt): Likewise.
16544 (vrp_visit_phi_node): Likewise.
16545 (simplify_cond_using_ranges_1): Likewise.
16546 (fold_predicate_in): Likewise.
16547 (evrp_dom_walker::before_dom_children): Likewise.
16548 (evrp_dom_walker::push_value_range): Likewise.
16549 (evrp_dom_walker::pop_value_range): Likewise.
16550 (execute_early_vrp): Likewise.
16552 2017-05-16 Richard Biener <rguenther@suse.de>
16554 * dwarf2out.c (loc_list_from_tree_1): Do not create
16555 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16557 2017-05-16 Richard Biener <rguenther@suse.de>
16559 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16561 (note_variable_value_in_expr): If we resolved the decl ref
16562 do not push to the stack.
16564 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16566 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16567 operations in fast-math mode.
16568 (vaddq_f16): Likewise.
16569 (vmul_f16): Likewise.
16570 (vmulq_f16): Likewise.
16571 (vsub_f16): Likewise.
16572 (vsubq_f16): Likewise.
16573 * config/arm/neon.md (add<mode>3): New.
16575 (fma:<VH:mode>3): New. Also remove outdated comment.
16578 2017-05-16 Martin Liska <mliska@suse.cz>
16582 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16583 (odr_types_equivalent_p): Likewise.
16585 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16587 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16589 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16592 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16593 non-interunit SSE move alternatives with '?'.
16594 (zero-extendsidi peephole2): New peephole to skip intermediate
16595 general register in SSE zero-extend sequence.
16597 2017-05-15 Jeff Law <law@redhat.com>
16599 * reorg.c (relax_delay_slots): Create a new variable to hold
16600 the temporary target rather than clobbering TARGET_LABEL.
16602 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16603 missing argument to extract_bit_field call.
16604 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16606 2017-05-15 Martin Liska <mliska@suse.cz>
16609 * gcc.c (process_command): Do not allow empty argument of -o option.
16611 2017-05-15 Renlin Li <renlin.li@arm.com>
16613 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16614 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16615 * config/aarch64/constraints.md (Usf): Add long call check.
16616 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16617 (call_value): Likewise.
16618 (sibcall): Likewise.
16619 (sibcall_value): Likewise.
16621 (call_value_insn): New.
16622 (sibcall_insn): Update rtx pattern.
16623 (sibcall_value_insn): Likewise.
16624 (call_internal): Remove.
16625 (call_value_internal): Likewise.
16626 (sibcall_internal): Likewise.
16627 (sibcall_value_internal): Likewise.
16628 (call_reg): Likewise.
16629 (call_symbol): Likewise.
16630 (call_value_reg): Likewise.
16631 (call_value_symbol): Likewise.
16633 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16636 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16638 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16640 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16641 compatible with CCGOCmode and with CCZmode.
16643 2017-05-14 Martin Sebor <msebor@redhat.com>
16645 PR middle-end/77671
16646 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16647 (gimple_fold_builtin_snprintf): Same.
16648 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16649 (gimple_fold_builtin_snprintf): Same.
16650 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16651 of character types.
16652 (is_call_safe): New function.
16653 (try_substitute_return_value): Call it.
16654 (try_simplify_call): New function.
16655 (pass_sprintf_length::handle_gimple_call): Call it.
16657 2017-05-14 Martin Sebor <msebor@redhat.com>
16659 PR middle-end/80669
16660 * builtins.c (expand_builtin_stpncpy): Simplify.
16662 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16664 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16665 * config/i386/i386.h
16666 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16667 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16668 (struct machine_function): Add new members call_ms2sysv,
16669 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16670 (struct machine_frame_state): New fields sp_realigned and
16671 sp_realigned_offset.
16672 * config/i386/i386.c
16673 (enum xlogue_stub): New enum.
16674 (enum xlogue_stub_sets): New enum.
16675 (class xlogue_layout): New class.
16676 (struct ix86_frame): New fields stack_realign_allocate_offset,
16677 stack_realign_offset and outlined_save_offset. Modify comments to
16678 detail stack layout when using out-of-line stubs.
16679 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16680 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16681 -mcall-ms2sysv-xlogues.
16682 (stub_managed_regs): New static variable.
16683 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16684 registers managed by out-of-line stub.
16685 (disable_call_ms2sysv_xlogues): New function.
16686 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16687 m->call_ms2sysv when appropriate and compute frame layout for
16689 (sp_valid_at, fp_valid_at): New inline functions.
16690 (choose_basereg): New function.
16691 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16693 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16694 Use align parameter of choose_baseaddr to generated aligned SSE movs
16696 (pro_epilogue_adjust_stack): Modify to track
16697 machine_frame_state::sp_realigned.
16698 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16699 (ix86_nsaved_sseregs): Likewise.
16700 (ix86_emit_save_regs): Likewise.
16701 (ix86_emit_save_regs_using_mov): Likewise.
16702 (ix86_emit_save_sse_regs_using_mov): Likewise.
16703 (get_scratch_register_on_entry): Likewise.
16704 (gen_frame_set): New function.
16705 (gen_frame_load): Likewise.
16706 (gen_frame_store): Likewise.
16707 (emit_outlined_ms2sysv_save): Likewise.
16708 (emit_outlined_ms2sysv_restore): Likewise.
16709 (ix86_expand_prologue): Modify stack re-alignment code and call
16710 emit_outlined_ms2sysv_save when appropriate.
16711 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16712 parameter rtx_insn *insn, which allows the function to be used to only
16713 generate the notes.
16714 (ix86_expand_epilogue): Modify validity checks of frame and stack
16715 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16716 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16717 * config/i386/predicates.md
16718 (save_multiple): New predicate.
16719 (restore_multiple): Likewise.
16720 * config/i386/sse.md
16721 (save_multiple<mode>): New pattern.
16722 (save_multiple_realign<mode>): Likewise.
16723 (restore_multiple<mode>): Likewise.
16724 (restore_multiple_and_return<mode>): Likewise.
16725 (restore_multiple_leave_return<mode>): Likewise.
16726 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16728 2017-05-14 Julia Koval <julia.koval@intel.com>
16730 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16731 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16732 (__builtin_ia32_xsetbv): New builtins.
16733 * config/i386/i386.c (ix86_expand_special_args_builtin):
16735 (ix86_expand_builtin): Special expand for new intrinsics.
16736 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16737 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16738 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16740 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16742 * cfganal.c (inverted_post_order_compute): Change argument type
16744 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16745 * df-core.c (rest_of_handle_df_initialize): Adjust.
16746 (rest_of_handle_df_finish): Likewise.
16747 (df_analyze_1): Likewise.
16748 (df_analyze): Likewise.
16749 (loop_inverted_post_order_compute): Change argument to be a vec *.
16750 (df_analyze_loop): Adjust.
16751 (df_get_n_blocks): Likewise.
16752 (df_get_postorder): Likewise.
16753 * df.h (struct df_d): Change field to be a vec.
16754 * lcm.c (compute_laterin): Adjust.
16755 (compute_available): Likewise.
16756 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16757 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16758 * tree-ssa-pre.c (compute_antic): Likewise.
16760 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16762 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16763 (depth_first_search::depth_first_search): Change structure init
16764 function to this constructor.
16765 (depth_first_search::add_bb): Rename function to this member.
16766 (depth_first_search::execute): Likewise.
16767 (flow_dfs_compute_reverse_finish): Adjust.
16769 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16771 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16772 (longest_simple_path): Likewise.
16773 * shrink-wrap.c (spread_components): Likewise.
16774 (disqualify_problematic_components): Likewise.
16775 (emit_common_heads_for_components): Likewise.
16776 (emit_common_tails_for_components): Likewise.
16777 (insert_prologue_epilogue_for_components): Likewise.
16779 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16781 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16784 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16786 * df-core.c (df_set_blocks): Start using auto_bitmap.
16787 (df_compact_blocks): Likewise.
16788 * df-problems.c (df_rd_confluence_n): Likewise.
16789 * df-scan.c (df_insn_rescan_all): Likewise.
16790 (df_process_deferred_rescans): Likewise.
16791 (df_update_entry_block_defs): Likewise.
16792 (df_update_exit_block_uses): Likewise.
16793 (df_entry_block_bitmap_verify): Likewise.
16794 (df_exit_block_bitmap_verify): Likewise.
16795 (df_scan_verify): Likewise.
16796 * lra-constraints.c (lra_constraints): Likewise.
16797 (undo_optional_reloads): Likewise.
16798 (lra_undo_inheritance): Likewise.
16799 * lra-remat.c (calculate_gen_cands): Likewise.
16800 (do_remat): Likewise.
16801 * lra-spills.c (assign_spill_hard_regs): Likewise.
16802 (spill_pseudos): Likewise.
16803 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16804 (bitmap_set_subtract_values): Likewise.
16806 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16808 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16809 management with auto_bitmap.
16810 (fix_inter_tick): Likewise.
16811 (fix_recovery_deps): Likewise.
16812 * ira.c (add_store_equivs): Likewise.
16813 (find_moveable_pseudos): Likewise.
16814 (split_live_ranges_for_shrink_wrap): Likewise.
16815 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16816 (rtx_reuse_manager::seen_def_p): Likewise.
16817 (rtx_reuse_manager::set_seen_def): Likewise.
16818 * print-rtl.h (class rtx_reuse_manager): Likewise.
16820 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16822 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16824 (migrate_btr_def): Likewise.
16825 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16826 * df-core.c (loop_post_order_compute): Likewise.
16827 (loop_inverted_post_order_compute): Likewise.
16828 * hsa-common.h: Likewise.
16829 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16830 * init-regs.c (initialize_uninitialized_regs): Likewise.
16831 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16832 (inline_small_functions): Likewise.
16833 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16834 * ira.c (combine_and_move_insns): Likewise.
16835 (build_insn_chain): Likewise.
16836 * loop-invariant.c (find_invariants): Likewise.
16837 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16838 * predict.c (tree_predict_by_opcode): Likewise.
16839 (predict_paths_leading_to): Likewise.
16840 (predict_paths_leading_to_edge): Likewise.
16841 (estimate_loops_at_level): Likewise.
16842 (estimate_loops): Likewise.
16843 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16844 (spread_components): Likewise.
16845 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16846 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16847 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16848 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16849 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16850 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16851 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16852 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16853 (mark_threaded_blocks): Likewise.
16854 (thread_through_all_blocks): Likewise.
16855 * tree-ssa.c (verify_ssa): Likewise.
16856 (execute_update_addresses_taken): Likewise.
16857 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16859 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16861 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16863 (post_order_compute): Likewise.
16864 (inverted_post_order_compute): Likewise.
16865 (pre_and_rev_post_order_compute_fn): Likewise.
16867 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16869 * genrecog.c (int_set::int_set): Explicitly construct our
16870 auto_vec base class.
16871 * vec.h (auto_vec::auto_vec): New constructor.
16873 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16875 * bitmap.h (class auto_bitmap): New constructor taking
16876 bitmap_obstack * argument.
16878 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16880 * bitmap.h (class auto_bitmap): Change type of m_bits to
16881 bitmap_head, and adjust ctor / dtor and member operators.
16883 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16885 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16886 when returned register mode doesn't match original mode.
16888 2017-05-12 Jeff Law <law@redhat.com>
16889 Jakub Jelinek <jakub@redhat.com>
16891 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16892 we look for cc setter after the compare-elim changes.
16893 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16894 within the vector to match what compare-elim now expects.
16895 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16896 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16898 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16899 after the compare-elim changes.
16900 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16901 the vector to match what compare-elim now expects.
16902 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16903 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16904 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16905 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16906 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16908 * config/visium/visium.c (single_set_and_flags): Fix where
16909 we look for cc setter after the compare-elim changes.
16910 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16911 with the vector to match what compare-elim now expects.
16912 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16913 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16914 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16915 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16916 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16917 (neg<mode>2_insn_set_overflow): Likewise.
16919 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16921 PR middle-end/79794
16922 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16923 maybe_expand_insn call, set ops[0].target. If still set after call,
16924 set alt_rtl. Add extra arg to recursive calls.
16925 (extract_bit_field): Add alt_rtl argument. Pass to
16927 * expmed.h (extract_bit_field): Fix prototype.
16928 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16929 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16930 to extract_bit_field_calls.
16931 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16932 Pass alt_rtl to extract_bit_field calls.
16933 * calls.c (store_unaligned_arguments_into_psuedos)
16934 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16935 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16937 * optabs.h (struct expand_operand): Add target bitfield.
16939 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16941 * compare-elim.c (try_eliminate_compare): Canonicalize
16942 operation with embedded compare to
16943 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16944 (set (reg) (operation)].
16946 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16948 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16951 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16952 cost of adding a carry flag for ADC instruction.
16953 [case MINUS]: Ignore the cost of subtracting a carry flag
16954 for SBB instruction.
16956 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16958 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16960 * config/rs6000/bmiintrin.h: New file.
16961 * config/rs6000/bmi2intrin.h: New file.
16962 * config/rs6000/x86intrin.h: New file.
16964 2017-05-12 Jeff Law <law@redhat.com>
16966 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16969 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16971 PR middle-end/80707
16972 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16974 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16976 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16977 early expansion of vector divide builtins.
16978 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16979 builtins identified as having unsigned arguments.
16981 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16983 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16984 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16985 expansion of vector logical operations (and, andc, or, xor,
16988 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16990 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16991 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16993 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16995 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16996 early GIMPLE expansion of vector multiplies.
16998 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17000 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17001 TARGET_HAVE_MOVT conditional.
17002 (movt splitter): Likewise.
17004 2017-05-12 Richard Biener <rguenther@suse.de>
17006 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17007 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17008 Fold all stmts not inplace.
17010 2017-05-12 Richard Biener <rguenther@suse.de>
17012 PR tree-optimization/80713
17013 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17014 inserted_exprs bit for not removed stmts.
17016 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17018 PR middle-end/69921
17019 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17020 parallelized" attribute for parallelized OpenACC kernels.
17021 * omp-offload.c (execute_oacc_device_lower): Use it.
17023 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17024 Set "oacc kernels" attribute.
17025 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17026 parameter. Adjust all users.
17027 (oacc_fn_attrib_kernels_p): Remove function.
17028 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17029 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17030 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17031 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17032 assert "oacc kernels" attribute is set.
17034 2017-05-11 Carl Love <cel@us.ibm.com>
17036 * config/rs6000/rs6000-c: Add support for built-in functions
17037 vector unsigned char vec_popcnt (vector signed char)
17038 vector unsigned char vec_popcnt (vector unsigned char)
17039 vector unsigned short vec_popcnt (vector signed short)
17040 vector unsigned short vec_popcnt (vector unsigned short)
17041 vector unsigned int vec_popcnt (vector signed int)
17042 vector unsigned int vec_popcnt (vector unsigned int)
17043 vector unsigned long long vec_popcnt (vector signed long long)
17044 vector unsigned long long vec_popcnt (vector unsigned long long)
17045 vector signed long long vec_slo (vector signed long long,
17046 vector signed char)
17047 vector signed long long vec_slo (vector signed long long,
17048 vector unsigned char)
17049 vector unsigned long long vec_slo (vector unsigned long long,
17050 vector signed char)
17051 vector unsigned long long vec_slo (vector unsigned long long,
17052 vector unsigned char)
17053 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17054 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17055 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17056 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17057 * doc/extend.texi: Update the built-in documentation file for the
17058 new built-in functions.
17060 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17062 * attribs.h (sorted_attr_string): Move machine independent
17063 functions for target clone support from the i386 port to common
17064 code. Rename ix86_function_versions to common_function_versions.
17065 Rename make_name to make_unique_name.
17066 (common_function_versions): Likewise.
17067 (make_unique_name): Likewise.
17068 (make_dispatcher_decl): Likewise.
17069 (is_function_default_version): Likewise.
17070 * attribs.c (attr_strcmp): Likewise.
17071 (sorted_attr_string): Likewise.
17072 (common_function_versions): Likewise.
17073 (make_unique_name): Likewise.
17074 (make_dispatcher_decl): Likewise.
17075 (is_function_default_version): Likewise.
17076 * config/i386/i386.c (attr_strcmp): Likewise.
17077 (sorted_attr_string): Likewise.
17078 (ix86_function_versions): Likewise.
17079 (make_name): Likewise.
17080 (make_dispatcher_decl): Likewise.
17081 (is_function_default_version): Likewise.
17082 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
17084 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17087 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17088 Account for direct move costs for vec_construct of integer
17091 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
17094 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
17095 (UNSPEC_STX_ATOMIC): Ditto.
17096 (loaddi_via_sse): New insn.
17097 (storedi_via_sse): Ditto.
17098 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
17099 Update corresponding peephole2 patterns.
17100 (atomic_storedi_fpu): Ditto.
17102 2017-05-11 Julia Koval <julia.koval@intel.com>
17104 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17105 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17107 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
17108 (__builtin_ia32_rsqrt14ss_mask): New builtins.
17109 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
17111 2017-05-11 Nathan Sidwell <nathan@acm.org>
17113 * graphite-poly.c: Include dumpfile.h.
17115 * dumpfle.h (dump_function): Declare here ...
17116 * tree-dump.h (dump_function): ... not here.
17117 * dumpfile.c: #include tree-cfg.h.
17118 (dump_function): Move here from ...
17119 * tree-dump.c (dump_function): ... here.
17120 * gimplify.c: #include splay-tree.h, not tree-dump.h.
17121 * graphite-poly.c: Don't include tree-dump.h.
17122 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
17123 * print-tree.c: Likewise.
17124 * stor-layout.c: Likewise.
17125 * tree-nested.c: Likewise.
17127 * dumpfile.c (dump_start): Use TDF_FLAGS.
17128 (dump_enable_all): Fix TDF_KIND check thinko.
17130 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17132 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17133 array entries to represent two legal parameterizations of the
17134 overloaded __builtin_cmpb function, as represented by the
17135 P6_OV_BUILTIN_CMPB constant.
17136 (altivec_resolve_overloaded_builtin): Add special case handling
17137 for the __builtin_cmpb function, as represented by the
17138 P6_OV_BUILTIN_CMPB constant.
17139 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17140 (BU_P6_64BIT_2): New macro.
17141 (BU_P6_OVERLOAD_2): New macro
17142 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17143 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17144 (CMPB): Add overload support to represent both 32-bit and 64-bit
17145 compare-bytes function.
17146 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17147 support for TARGET_CMPB.
17148 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17149 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17150 documentation of the __builtin_cmpb overloaded built-in function.
17152 2017-05-11 Richard Biener <rguenther@suse.de>
17154 PR tree-optimization/80705
17155 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17156 bases are not vectorizable.
17158 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17160 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17161 when counting register pressure.
17163 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17165 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17166 (iv_ca_more_deps): Renamed to ...
17167 (iv_ca_compare_deps): ... this.
17168 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17170 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17172 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17174 (determine_group_iv_costs): ... here.
17175 (find_inv_vars_cb): Record inv var if it's not recorded before.
17177 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17179 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17180 (get_shiftadd_cost): Ditto.
17182 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17184 * tree-ssa-address.c: Include header file.
17185 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17187 (add_to_parts): Refactor.
17188 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17189 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17192 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17194 PR tree-optimization/53090
17195 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17197 (extract_cond_operands): Detect condition with IV on both sides
17198 and return COMP_IV_EXPR_2.
17199 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17200 (rewrite_use_compare): Simplify by removing call to function
17201 extract_cond_operands.
17203 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17205 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17206 (extract_cond_operands): Detect condition comparing against non-
17207 invariant bound and return appropriate enum value.
17208 (find_interesting_uses_cond): Update use of extract_cond_operands.
17209 Handle its return value accordingly.
17210 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17214 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17215 nonlinear iv_use computation in loop invariant sensitive way.
17217 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17219 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17220 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17222 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17224 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17225 (dump_cand): Support iv_cand.inv_exprs.
17226 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17228 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17231 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17233 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17235 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17236 as local function. Include necessary header files.
17237 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17239 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17241 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17245 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17246 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17247 RSHIFT_EXPR and BIT_NOT_EXPR.
17249 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17251 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17252 (adjust_setup_cost): New parameter supporting round up adjustment.
17253 (struct address_cost_data): Delete.
17254 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17255 (split_address_cost, ptr_difference_cost): Delete.
17256 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17257 (struct ainc_cost_data): New struct.
17258 (get_address_cost_ainc): New function.
17259 (get_address_cost, get_computation_cost): Reimplement.
17260 (determine_group_iv_cost_address): Record inv_expr for all uses of
17262 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17263 (iv_ca_has_deps): Reimplemented to ...
17264 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17266 (iv_ca_extend): Call iv_ca_more_deps.
17268 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17270 * tree-ssa-address.c (struct mem_address): Move to header file.
17271 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17272 * tree-ssa-address.h (struct mem_address): Move from C file.
17273 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17275 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17277 * tree-affine.h (aff_combination_type): New interface.
17278 (aff_combination_zero_p): Remove static.
17279 (aff_combination_const_p): New interface.
17280 (aff_combination_singleton_var_p): New interfaces.
17282 2017-05-11 Richard Biener <rguenther@suse.de>
17284 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17285 Skip unreachable blocks and destinations.
17286 (eliminate): Move stmt removal and fixup ...
17287 (fini_eliminate): ... here. Skip inserted exprs.
17288 (pass_pre::execute): Move fini_pre after fini_eliminate.
17289 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17290 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17291 PRE to get rid of dead code that has invalid SSA form and
17292 split critical edges again.
17294 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17296 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17298 2017-05-11 Richard Biener <rguenther@suse.de>
17300 * passes.c (execute_function_todo): Verify loops if they are
17301 said to be up-to-date.
17302 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17303 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17305 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17308 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17309 handle calling assemble_external ourself.
17312 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17313 modes with zero size. Enhance comment.
17315 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17317 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17318 built-ins for vec_xl and vec_xst with short and char pointer
17321 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17323 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17324 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17325 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17326 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17327 (_mm_maskz_min_round_ss): New intrinsics.
17328 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17329 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17330 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17331 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17332 (__builtin_ia32_minss_mask_round): New builtins.
17333 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17334 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17335 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17337 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17338 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17340 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17343 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17345 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17346 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17347 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17348 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17349 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17350 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17351 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17352 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17353 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17354 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17355 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17356 (__builtin_ia32_mulss_mask_round): New builtins.
17357 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17358 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17359 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17361 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17362 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17364 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17367 2017-05-10 Julia Koval <julia.koval@intel.com>
17369 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17370 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17371 (_mm256_setr_m128i): New intrinsics.
17373 2017-05-10 Julia Koval <julia.koval@intel.com>
17375 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17376 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17377 (_mm_maskz_rcp14_ss): New intrinsics.
17378 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17379 (__builtin_ia32_rcp14ss_mask): New builtins.
17380 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17382 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17384 PR tree-optimization/51513
17385 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17386 (assert_unreachable_fallthru_edge_p): Use it.
17387 (group_case_labels_stmt): Likewise.
17388 * tree-cfg.h: Prototype it.
17389 * stmt.c: Include cfghooks.h and tree-cfg.h.
17390 (emit_case_dispatch_table) <gap_label>: New local variable.
17391 Use it to fill dispatch table gaps.
17392 Test for default_label before updating probabilities.
17393 (expand_case) <default_label>: Remove unneeded initialization.
17394 Test for unreachable default case statement and remove its edge.
17395 Set default_label accordingly.
17396 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17398 2017-05-10 Carl Love <cel@us.ibm.com>
17400 * config/rs6000/rs6000-c: Add support for built-in functions
17401 vector signed char vec_neg (vector signed char)
17402 vector signed short int vec_neg (vector short int)
17403 vector signed int vec_neg (vector signed int)
17404 vector signed long long vec_neg (vector signed long long)
17405 vector float vec_neg (vector float)
17406 vector double vec_neg (vector double)
17407 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17409 * config/rs6000/altivec.h: Add define for vec_neg
17410 * doc/extend.texi: Update the built-in documentation for the
17411 new built-in functions.
17413 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17415 PR tree-optimization/77644
17416 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17418 2017-05-10 Nathan Sidwell <nathan@acm.org>
17420 * dumpfile.h (TDI_lang_all): New.
17421 (TDF_KIND): New. Renumber others
17422 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17424 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17426 (get_dump_file_name): Adjust suffix generation.
17427 (dump_enable_all): Use TDF_KIND.
17428 * doc/invoke.texi (-fdump-lang-all): Document.
17430 * dumpfile.h: Tabify.
17432 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17435 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17436 Move member access before delete.
17438 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17440 * tree-inline.c (expand_call_inline): Split block at stmt
17443 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17446 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17447 are now unused after splitting mov{sf,sd}_hardfloat.
17448 (f32_lr2): Likewise.
17449 (f32_lm): Likewise.
17450 (f32_lm2): Likewise.
17451 (f32_li): Likewise.
17452 (f32_li2): Likewise.
17453 (f32_lv): Likewise.
17454 (f32_sr): Likewise.
17455 (f32_sr2): Likewise.
17456 (f32_sm): Likewise.
17457 (f32_sm2): Likewise.
17458 (f32_si): Likewise.
17459 (f32_si2): Likewise.
17460 (f32_sv): Likewise.
17461 (f32_dm): Likewise.
17462 (f32_vsx): Likewise.
17463 (f32_av): Likewise.
17464 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17465 For movsf, order stores so the VSX stores occur before the GPR
17466 store which encourages the register allocator to use a traditional
17467 FPR instead of a GPR. For movsd, order the stores so that the GPR
17468 store comes before the VSX stores to allow the power6 to work.
17469 This is due to the power6 not having a 32-bit integer store
17470 instruction from a FPR.
17471 (movsf_hardfloat): Likewise.
17472 (movsd_hardfloat): Likewise.
17474 2017-05-09 Martin Sebor <msebor@redhat.com>
17476 PR translation/80280
17477 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17480 PR translation/80280
17481 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17482 data member added in r247778.
17483 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17485 2017-05-09 Nathan Sidwell <nathan@acm.org>
17487 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17489 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17492 2017-05-09 Marek Polacek <polacek@redhat.com>
17494 * doc/invoke.texi: Fix typo.
17496 2017-05-09 Richard Biener <rguenther@suse.de>
17498 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17499 (vrp_val_is_min): Likewise.
17500 (set_value_range_to_value): Likewise.
17501 (set_value_range_to_nonnegative): Likewise.
17502 (gimple_assign_nonzero_p): Likewise.
17503 (gimple_stmt_nonzero_p): Likewise.
17504 (vrp_int_const_binop): Likewise. Remove unreachable case.
17505 (adjust_range_with_scev): Adjust comments.
17506 (compare_range_with_value): Likewise.
17507 (extract_range_from_phi_node): Likewise.
17508 (test_for_singularity): Likewise.
17510 2017-05-09 Richard Biener <rguenther@suse.de>
17512 * tree-vrp.c (get_single_symbol): Add assert that we don't
17513 get overflowed constants as invariant part.
17514 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17515 checks. Use wi::cmp instead of recursing for integer constants.
17516 (compare_values): Just ignore whether we assumed undefined
17517 overflow instead of failing the compare.
17518 (extract_range_for_var_from_comparison_expr): Add comment before the
17519 TREE_NO_WARNING sets.
17520 (test_for_singularity): Likewise.
17521 (extract_range_from_comparison): Do not disable optimization
17522 when we assumed undefined overflow.
17523 (extract_range_basic): Remove init of unused var.
17525 2017-05-09 Richard Biener <rguenther@suse.de>
17527 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17528 (extract_range_from_multiplicative_op_1): Adjust.
17529 (extract_range_from_binary_expr_1): Use int_const_binop.
17531 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17534 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17535 rs6000_store_data_bypass_p in seven define_bypass directives and
17536 in several comments.
17537 * config/rs6000/rs6000-protos.h: Add prototype for
17538 rs6000_store_data_bypass_p function.
17539 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17540 function implements slightly different (rs6000-specific) semantics
17541 than store_data_bypass_p, returning false rather than aborting
17542 with assertion error when arguments do not satisfy the
17543 requirements of store data bypass.
17544 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17545 rs6000_store_data_bypass_p.
17547 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17549 * config/xtensa/xtensa-protos.h
17550 (xtensa_initial_elimination_offset): New declaration.
17551 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17552 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17553 macro definition, add case for FRAME_POINTER_REGNUM when
17554 FRAME_GROWS_DOWNWARD.
17555 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17556 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17557 xtensa_initial_elimination_offset.
17559 2017-05-08 Nathan Sidwell <nathan@acm.org>
17561 * doc/invoke.texi: Alphabetize -fdump options.
17563 2017-05-08 Martin Sebor <msebor@redhat.com>
17565 PR translation/80280
17566 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17568 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17570 * target.def (compute_frame_layout): New optional target hook.
17571 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17572 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17573 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17575 * reload1.c (verify_initial_elim_offsets): Likewise.
17576 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17577 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17578 (arm_get_frame_offsets): Split up into this ...
17579 (arm_compute_frame_layout): ... and this function.
17581 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17583 * config/aarch64/constraints.md (Usa): New constraint.
17584 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17586 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17588 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17589 with_multilib_list after it has been checked.
17591 2017-05-08 Richard Biener <rguenther@suse.de>
17593 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17594 (bitmap_set_subtract_values): Likewise.
17596 2017-05-08 Richard Biener <rguenther@suse.de>
17598 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17599 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17601 (gimple_stmt_nonzero_warnv_p): Rename to ...
17602 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17604 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17605 (extract_range_basic): Adjust, do not disable propagation on
17606 strict overflow sensitive simplification.
17607 (vrp_visit_cond_stmt): Likewise.
17609 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17611 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17612 body size unconditionally.
17614 2017-05-07 Jeff Law <law@redhat.com>
17617 2017-05-06 Jeff Law <law@redhat.com>
17618 PR tree-optimization/78496
17619 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17622 PR tree-optimization/78496
17623 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17624 (simplify_stmt_using_ranges): Call it.
17625 (vrp_dom_walker::before_dom_children): Extract equivalences
17626 from an ASSERT_EXPR with an equality comparison against a
17629 2017-05-06 Jeff Law <law@redhat.com>
17631 PR tree-optimization/78496
17632 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17635 PR tree-optimization/78496
17636 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17637 (simplify_stmt_using_ranges): Call it.
17638 (vrp_dom_walker::before_dom_children): Extract equivalences
17639 from an ASSERT_EXPR with an equality comparison against a
17642 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17644 * lra-constraints.c (lra_copy_reg_equiv): New function.
17645 (split_reg): Use it to copy equivalence information from the
17646 original register to the spill register.
17648 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17650 PR rtl-optimization/75964
17651 * simplify-rtx.c (simplify_const_relational_operation): Remove
17652 invalid handling of comparisons of integer ABS.
17654 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17656 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17657 initialize to zero.
17658 (init_regs): Remove declaration.
17659 (function_arg_advance_32): Initialize error_p as boolean variable.
17661 2017-05-05 Nathan Sidwell <nathan@acm.org>
17663 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17664 lines. Use for (;;).
17666 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17668 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17669 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17670 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17671 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17672 VF=2 that require versioning.
17674 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17676 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17679 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17681 * diagnostic.h (diagnostic_override_option_index): Convert from
17682 macro to inline function.
17684 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17686 * diagnostic.c (last_module_changed_p): New function.
17687 (set_last_module): New function.
17688 (diagnostic_report_current_module): Convert macro usage to
17689 the above functions.
17690 * diagnostic.h (diagnostic_context::last_module): Strengthen
17691 from const line_map * to const line_map_ordinary *.
17692 (diagnostic_last_module_changed): Delete macro.
17693 (diagnostic_set_last_module): Delete macro.
17695 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17697 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17698 with diagnostic_report_diagnostic.
17699 (diagnostic_n_impl_richloc): Likewise.
17700 * diagnostic.h (report_diagnostic): Delete macro.
17701 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17702 with diagnostic_report_diagnostic.
17703 * substring-locations.c (format_warning_va): Likewise.
17705 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17707 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17708 save/restor of format_spec. Move option-printing code to...
17709 (print_option_information): ...this new function, and
17710 reimplement by simply printing to the pretty_printer,
17711 rather than appending to the format string.
17713 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17715 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17716 handling logic into...
17717 (update_effective_level_from_pragmas): ...this new function.
17719 2017-05-04 Andrew Waterman <andrew@sifive.com>
17721 * config/riscv/riscv.opt (mstrict-align): New option.
17722 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17723 (SLOW_UNALIGNED_ACCESS): Define.
17724 (riscv_slow_unaligned_access): Declare.
17725 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17727 (riscv_slow_unaligned_access): New variable.
17728 (rocket_tune_info): Set slow_unaligned_access to true.
17729 (optimize_size_tune_info): Set slow_unaligned_access to false.
17730 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17731 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17732 (riscv_option_override): Set riscv_slow_unaligned_access.
17733 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17735 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17737 * config/riscv/riscv.md: Unify indentation.
17739 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17744 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17746 (extendsi<mode>2): Add support for doing sign extension via
17747 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17748 don't have ISA 3.0 instructions.
17749 (extendsi<mode>2 splitter): Likewise.
17750 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17751 generate the normal insns since SImode can now go in vector
17752 registers. Disallow the special UNSPECs needed for previous
17753 machines to hide SImode being used. Add new insns
17754 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17755 (fix_trunc<mode>si2_stfiwx): Likewise.
17756 (fix_trunc<mode>si2_internal): Likewise.
17757 (fixuns_trunc<mode>si2): Likewise.
17758 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17759 (fctiw<u>z_<mode>_smallint): Likewise.
17760 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17761 of floating point to 32-bit integer from doing a direct move to
17762 the GPR registers to do a store.
17763 (fctiwz_<mode>): Break long line.
17765 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17767 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17768 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17769 (addr_list, addr_offset_valid_p): New.
17770 (split_address_groups): Check offset validity with above function.
17771 (gt-tree-ssa-loop-ivopts.h): Include header file.
17773 2017-05-05 Nathan Sidwell <nathan@acm.org>
17775 * config.gcc (arm*-*-*): Add missing 'fi'.
17777 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17779 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17780 included in -fopt-info does not matter.
17781 * doc/optinfo.texi (-fopt-info): Fix description of default
17782 behavour. Explicitly say order of options included in -fopt-info
17785 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17787 * config.gcc: Allow combinations of aprofile and rmprofile values for
17788 --with-multilib-list.
17789 * config/arm/t-multilib: New file.
17790 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17791 variables. Remove setting of ISA and floating-point ABI in
17792 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17793 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17794 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17795 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17797 * config/arm/t-rmprofile: Likewise except for the matches changes.
17798 * doc/install.texi (--with-multilib-list): Document the combination of
17799 aprofile and rmprofile values and warn about pitfalls in doing that.
17801 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17803 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17804 (movdi_aarch64): Likewise.
17806 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17808 PR tree-optimization/80632
17809 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17811 (build_arrays): Initialize it for virtual phis.
17812 (fix_phi_nodes): Use it for virtual phis.
17814 PR tree-optimization/80558
17815 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17816 [x, y] op z into [x op, y op z] for op & or | if conditions
17819 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17820 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17823 * config/arm/arm.md (use_literal_pool): Remove.
17824 (64-bit immediate split): No longer takes cost into consideration
17825 if arm_disable_literal_pool is enabled.
17826 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17827 used when arm_disable_literal_pool is enabled.
17828 (arm_max_const_double_inline_cost): Remove use of
17829 arm_disable_literal_pool.
17830 (push_minipool_fix): Add assert.
17831 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17832 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17833 (no_literal_pool_sf_immediate): New.
17835 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17837 PR tree-optimization/80613
17838 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17839 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17841 2017-05-05 Richard Biener <rguenther@suse.de>
17843 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17845 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17847 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17848 of this flag from insn conditions due to removal from r247495.
17850 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17852 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17854 (arm_early_store_addr_dep_ptr): Likewise.
17855 * config/arm/aarch-common-protos.h
17856 (arm_early_load_addr_dep_ptr): Add prototype.
17857 (arm_early_store_addr_dep_ptr): Likewise.
17858 * config/arm/cortex-a53.md: Add new bypasses.
17860 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17862 * tree.c (next_type_uid): Change type to unsigned.
17863 (type_hash_canon): Decrement back next_type_uid if
17864 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17865 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17868 2017-05-04 Martin Sebor <msebor@redhat.com>
17870 * builtins.c: Fix a trivial typo in a comment.
17872 PR middle-end/79234
17873 * builtins.c (check_sizes): Adjust to handle reading past the end.
17874 Avoid printing excessive upper bound of ranges. Use %E to print
17875 tree nodes instead of converting them to %wu.
17876 (expand_builtin_memchr): New function.
17877 (compute_dest_size): Rename...
17878 (compute_objsize): ...to this.
17879 (expand_builtin_memcpy): Adjust.
17880 (expand_builtin_mempcpy): Adjust.
17881 (expand_builtin_strcat): Adjust.
17882 (expand_builtin_strcpy): Adjust.
17883 (check_strncat_sizes): Adjust.
17884 (expand_builtin_strncat): Adjust.
17885 (expand_builtin_strncpy): Adjust and simplify.
17886 (expand_builtin_memset): Adjust.
17887 (expand_builtin_bzero): Adjust.
17888 (expand_builtin_memcmp): Adjust.
17889 (expand_builtin): Handle memcmp.
17890 (maybe_emit_chk_warning): Check strncat just once.
17892 2017-05-04 Martin Sebor <msebor@redhat.com>
17894 PR preprocessor/79214
17895 PR middle-end/79222
17896 PR middle-end/79223
17897 * builtins.c (check_sizes): Add inlining context and issue
17898 warnings even when -Wno-system-headers is set.
17899 (check_strncat_sizes): Same.
17900 (expand_builtin_strncat): Same.
17901 (expand_builtin_memmove): New function.
17902 (expand_builtin_stpncpy): Same.
17903 (expand_builtin): Handle memmove and stpncpy.
17905 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17907 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17908 which is not used any more.
17910 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17912 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17914 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17916 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17917 (cortexa53_tunings): Likewise.
17918 (cortexa57_tunings): Likewise.
17919 (cortexa72_tunings): Likewise.
17920 (cortexa73_tunings): Likewise.
17922 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17924 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17925 Set loop alignment to 8.
17927 2017-05-04 Martin Sebor <msebor@redhat.com>
17929 PR translation/80280
17930 * builtins.c (expand_builtin_object_size): Add missing quoting to
17931 %D and like directives.
17932 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17933 (hsa_type_for_tree_type): Same.
17934 (verify_function_arguments): Same.
17935 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17936 * varasm.c (get_section): Same.
17939 2017-05-04 Martin Sebor <msebor@redhat.com>
17941 PR translation/80280
17942 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17944 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17946 * config/aarch64/aarch64.c (generic_addrcost_table):
17947 Change HI/TI mode setting.
17949 2017-05-04 Martin Jambor <mjambor@suse.cz>
17951 PR tree-optimization/80622
17952 * tree-sra.c (comes_initialized_p): New function.
17953 (build_accesses_from_assign): Only set write lazily when
17954 comes_initialized_p is false.
17955 (analyze_access_subtree): Use comes_initialized_p.
17956 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17957 instead of testing for PARM_DECL.
17959 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17961 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17962 constraint on operand 0 to allow more general addressing modes.
17963 Adjust output template.
17964 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17966 * config/aarch64/aarch64-protos.h
17967 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17968 * config/aarch64/constraints.md (Dp): New address constraint.
17969 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17972 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17974 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17975 update use of estimate_ipcp_clone_size_and_time.
17976 (estimate_local_effects): Update use of
17977 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17978 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17979 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17980 Return nonspecialized time.
17982 2017-05-04 Richard Biener <rguenther@suse.de>
17984 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17985 for the last VUSE which def dominates the PHI. Directly call
17987 (get_continuation_for_phi_1): Remove.
17989 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17991 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17992 to explain the use of truncating division. Cap the number of
17993 iterations to the maximum given by nb_iterations_upper_bound,
17996 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17998 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17999 * configure: Regenerate.
18000 * config.in: Regenerate.
18001 * config/i386/driver-mingw32.c: new file.
18002 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18003 * config.host: Link driver-mingw32.o on MinGW host.
18004 * doc/install.texi: Document new --enable-mingw-wildcard configure
18007 2017-05-04 Marek Polacek <polacek@redhat.com>
18009 PR tree-optimization/80612
18010 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18012 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18013 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18015 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18016 (movt splitter): Likewise.
18017 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18018 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18019 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18020 block for Thumb-1 with MOVT.
18021 (thumb2_legitimate_address_p): Move code block ...
18022 (can_avoid_literal_pool_for_label_p): ... into this new function.
18023 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18025 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18026 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18027 "M-profile targets with the MOVT instruction".
18029 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18031 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18032 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18033 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18035 2017-05-04 Martin Liska <mliska@suse.cz>
18037 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18038 variable cond_code.
18040 2017-05-04 Richard Biener <rguenther@suse.de>
18042 * tree.c (array_at_struct_end_p): Handle arrays at struct
18043 end with flexarrays more conservatively. Refactor and treat
18044 arrays of arrays or aggregates more strict. Fix
18045 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18046 * tree.c (array_at_struct_end_p): Adjust prototype.
18047 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18048 * gimple-fold.c (get_range_strlen): Likewise.
18049 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18051 2017-05-04 Richard Biener <rguenther@suse.de>
18053 PR tree-optimization/31130
18054 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18056 (supports_overflow_infinity): Likewise.
18057 (is_negative_overflow_infinity): Likewise.
18058 (is_positive_overflow_infinity): Likewise.
18059 (is_overflow_infinity): Likewise.
18060 (stmt_overflow_infinity): Likewise.
18061 (overflow_infinity_range_p): Likewise.
18062 (usable_range_p): Remove as always returning true.
18063 (make_overflow_infinity): Remove.
18064 (negative_overflow_infinity): Likewise.
18065 (positive_overflow_infinity): Likewise.
18066 (avoid_overflow_infinity): Likewise.
18067 (set_value_range): Adjust accordingly.
18068 (set_value_range_to_nonnegative): Likewise, remove now unused
18069 overflow_infinity arg.
18070 (vrp_operand_equal_p): Adjust.
18071 (update_value_range): Likewise.
18072 (range_int_cst_singleton_p): Likewise.
18073 (operand_less_p): Likewise.
18074 (compare_values_warnv): Likewise.
18075 (extract_range_for_var_from_comparison_expr): Likewise.
18076 (vrp_int_const_binop): Likewise.
18077 (zero_nonzero_bits_from_vr): Likewise.
18078 (extract_range_from_multiplicative_op_1): Likewise.
18079 (extract_range_from_binary_expr_1): Likewise.
18080 (extract_range_from_unary_expr): Likewise.
18081 (extract_range_from_comparison): Likewise.
18082 (extract_range_basic): Likewise.
18083 (adjust_range_with_scev): Likewise.
18084 (compare_ranges): Likewise.
18085 (compare_range_with_value): Likewise.
18086 (dump_value_range): Likewise.
18087 (test_for_singularity): Likewise, remove strict_overflow_p parameter
18089 (simplify_cond_using_ranges): Adjust.
18091 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18093 * brig-builtins.def: Added a builtin for class_f64.
18094 * builtin-types.def: Added a builtin type needed by class_f64.
18096 2017-05-03 Jason Merrill <jason@redhat.com>
18098 * timevar.def: Add TV_CONSTEXPR.
18100 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18102 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18104 2017-05-03 Martin Jambor <mjambor@suse.cz>
18106 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18107 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
18108 * ipa-cp.c (ipcp_propagate_stage): Do not call
18109 ipa_update_after_lto_read.
18110 * ipa-inline.c (ipa_inline): Likewise.
18112 2017-05-03 Martin Jambor <mjambor@suse.cz>
18114 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
18115 tag. Added a default constructor and a destructor.
18116 (ipa_edge_args_sum_t): New class;
18117 (ipa_edge_args_sum): Declare.
18118 (ipa_edge_args_vector): Remove declaration.
18119 (IPA_EDGE_REF): Use ipa_edge_args_sum.
18120 (ipa_free_edge_args_substructures): Remove declaration.
18121 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
18122 (ipa_edge_args_info_available_for_edge_p): Likewise.
18123 * ipa-prop.c (ipa_edge_args_vector): Removed.
18124 (edge_removal_hook_holder): Likewise.
18125 (edge_duplication_hook_holder): Likewise.
18126 (ipa_edge_args_sum): New variable.
18127 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
18128 ipa_edge_args_vector.
18129 (ipa_free_edge_args_substructures): Likewise.
18130 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18131 ipa_edge_args_vector.
18132 (ipa_edge_removal_hook): Turned into method
18133 ipa_edge_args_sum_t::remove.
18134 (ipa_edge_duplication_hook): Turned into method
18135 ipa_edge_args_sum_t::duplicate.
18136 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18137 registering edge hooks.
18138 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18139 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18140 ipa_edge_args_sum instead of ipa_edge_args_vector.
18141 * ipa-profile.c (ipa_profile): Likewise.
18143 2017-05-03 Martin Jambor <mjambor@suse.cz>
18145 * symbol-summary.h (function_summary): New method exists.
18146 (function_summary::symtab_removal): Deallocate through release.
18147 (call_summary): New class.
18148 (gt_ggc_mx): New overload.
18149 (gt_pch_nx): Likewise.
18150 (gt_pch_nx): Likewise.
18152 2017-05-03 Jeff Law <law@redhat.com>
18154 PR tree-optimization/78496
18155 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18156 from simplify_cond_using_ranges. Split off code to walk
18157 backwards through casts into ...
18158 (simplify_cond_using_ranges_2): New function.
18159 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18160 (execute_vrp): After identifying jump threads, call
18161 simplify_cond_using_ranges_2.
18163 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18166 * ipa-inline.h (inline_summary): Add ctor.
18167 (create_ggc): Do not use ggc_cleared_alloc.
18169 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18170 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18172 * gcc.c (handle_braces): Support escaping in switch matching
18174 * doc/invoke.texi (Spec Files): Document it.
18175 Remove superfluous @code markup in items.
18177 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18179 * diagnostic-show-locus.c (struct column_range): New struct.
18180 (get_affected_columns): New function.
18181 (get_printed_columns): New function.
18182 (struct correction): New struct.
18183 (correction::ensure_capacity): New function.
18184 (correction::ensure_terminated): New function.
18185 (struct line_corrections): New struct.
18186 (line_corrections::~line_corrections): New dtor.
18187 (line_corrections::add_hint): New function.
18188 (layout::print_trailing_fixits): Reimplement in terms of the new
18190 (selftest::test_overlapped_fixit_printing): New function.
18191 (selftest::diagnostic_show_locus_c_tests): Call it.
18193 2017-05-03 Nathan Sidwell <nathan@acm.org>
18195 Canonicalize canonical type hashing
18196 * tree.h (type_hash_canon_hash): Declare.
18197 * tree.c (type_hash_list, attribute_hash_list): Move into
18198 type_hash_canon_hash.
18199 (build_type_attribute_qual_variant): Break out hash code calc into
18200 type_hash_canon_hash.
18201 (type_hash_canon_hash): New. Generic type hash computation.
18202 (build_range_type_1, build_array_type_1, build_function_type,
18203 build_method_type_directly, build_offset_type, build_complex_type,
18204 make_vector_type): Call it.
18206 2017-05-03 Richard Biener <rguenther@suse.de>
18208 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18209 When all DRs have unknown misaligned do not always peel
18210 when there is a store but apply the same costing model as if
18211 there were only loads.
18213 2017-05-03 Richard Biener <rguenther@suse.de>
18216 PR tree-optimization/80492
18217 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18218 compare_base_decls returning dont-know properly.
18220 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18222 * config/arm/iterators.md (CCSI): New mode iterator.
18223 (arch): New mode attribute.
18224 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18225 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18226 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18227 code iterator for success result mode.
18228 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18229 the corresponding new insn generators.
18231 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18234 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18235 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18237 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18239 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18240 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18241 (DDR_A): Wrap DDR argument in brackets.
18242 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18243 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18244 (DDR_REVERSED_P): Likewise.
18246 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18248 PR tree-optimization/79472
18249 * tree-switch-conversion.c (struct switch_conv_info): Add
18250 contiguous_range and default_case_nonstandard fields.
18251 (collect_switch_conv_info): Compute contiguous_range and
18252 default_case_nonstandard fields, don't clear final_bb if
18253 contiguous_range and only the default case doesn't have the required
18255 (check_all_empty_except_final): Set default_case_nonstandard instead
18256 of failing if contiguous_range and the default case doesn't have empty
18258 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18259 and only the default case doesn't have the required constants. Skip
18261 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18262 if default_case_nonstandard.
18263 (build_constructors): Build constant 1 just once. Assert that default
18264 values aren't inserted in between cases if contiguous_range. Skip
18266 (build_arrays): Skip virtual phis.
18267 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18268 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18269 Handle virtual phis.
18270 (gen_inbound_check): Handle default_case_nonstandard case.
18271 (process_switch): Adjust check_final_bb caller. Call
18272 gather_default_values with the first non-default case instead of
18273 default case if default_case_nonstandard.
18275 2017-05-02 Nathan Sidwell <nathan@acm.org>
18277 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18278 check. Fix formatting.
18280 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18282 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18283 errors when comparing specialized and unspecialized times.
18285 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18287 * diagnostic-show-locus.c
18288 (layout::should_print_annotation_line_p): Make private.
18289 (layout::print_annotation_line): Make private.
18290 (layout::annotation_line_showed_range_p): Make private.
18291 (layout::show_ruler): Make private.
18292 (layout::print_source_line): Make private. Pass in line and
18293 line_width, rather than calling location_get_source_line. Drop
18295 (layout::print_leading_fixits): New method.
18296 (layout::print_any_fixits): Rename to...
18297 (layout::print_trailing_fixits): ...this, and make private.
18298 Don't print newline fixits.
18299 (diagnostic_show_locus): Move logic for printing one row into...
18300 (layout::print_line): ...this new function. Move the
18301 location_get_source_line call and error-handling from
18302 print_source_line to here. Call print_leading_fixits, and rename
18303 print_any_fixits to print_trailing_fixits.
18304 (selftest::test_fixit_insert_containing_newline): Update now that
18305 newlines are partially supported.
18306 (selftest::test_fixit_insert_containing_newline_2): New test.
18307 (selftest::test_fixit_replace_containing_newline): Update comments.
18308 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18309 * edit-context.c (class added_line): New class.
18310 (class edited_line): Describe newline handling in comment.
18311 (edited_line::actually_edited_p): New method.
18312 (edited_line::print_content): Delete redundant decl.
18313 (edited_line::m_predecessors): New field.
18314 (edited_file::print_content): Call edited_line::print_content.
18315 (edited_file::print_diff): Update to support newlines.
18316 (edited_file::print_diff_hunk): Likewise.
18317 (edited_file::print_run_of_changed_lines): New function.
18318 (edited_file::print_diff_line): Convert to...
18319 (print_diff_line): ...this.
18320 (edited_file::get_effective_line_count): New function.
18321 (edited_line::edited_line): Initialize new field m_predecessors.
18322 (edited_line::~edited_line): Clean up m_predecessors.
18323 (edited_line::apply_fixit): Handle newlines.
18324 (edited_line::get_effective_line_count): New function.
18325 (edited_line::print_content): New function.
18326 (edited_line::print_diff_lines): New function.
18327 (selftest::test_applying_fixits_insert_containing_newline): New
18329 (selftest::test_applying_fixits_replace_containing_newline): New
18331 (selftest::insert_line): New function.
18332 (selftest::test_applying_fixits_multiple_lines): Add example of
18334 (selftest::edit_context_c_tests): Call the new tests.
18336 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18338 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18339 parameter cand. Update dump information.
18340 (get_computation_cost): Update uses.
18342 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18344 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18345 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18346 (get_computation_at, rewrite_use_address): Update use of
18347 get_computation_aff.
18349 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18351 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18352 (get_computation): Delete.
18353 (get_computation_cost): Implement like get_computation_cost_at.
18354 Use get_computation_at.
18355 (get_computation_cost_at): Delete.
18356 (rewrite_use_nonlinear_expr): Use get_computation_at.
18357 (rewrite_use_compare, remove_unused_ivs): Ditto.
18359 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18361 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18363 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18365 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18366 (ivopts_global_cost_for_size): Rename parameter and update uses.
18367 (iv_ca_recount_cost): Update uses.
18368 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18369 candidates seperately in n_invs and n_cands.
18370 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18372 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18374 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18375 (find_inv_vars_cb): New.
18376 (find_depends): Renamed to ...
18377 (find_inv_vars): ... this.
18378 (add_candidate_1, force_var_cost): Call find_inv_vars.
18379 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18381 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18383 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18384 inv_vars. Add inv_exprs.
18385 (struct iv_cand): Rename depends_on to inv_vars.
18386 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18387 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18388 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18389 (dump_cand): Dump inv_vars.
18390 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18391 (record_invariant, find_depends, add_candidate_1): Ditto.
18392 (set_group_iv_cost, force_var_cost): Ditto.
18393 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18394 (get_computation_cost_at, get_computation_cost): Ditto.
18395 (determine_group_iv_cost_generic): Ditto.
18396 (determine_group_iv_cost_address): Ditto.
18397 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18398 (determine_group_iv_costs): Ditto.
18399 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18400 (iv_ca_set_remove_invariants): Renamed to ...
18401 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18402 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18403 (iv_ca_set_add_invariants): Renamed to ...
18404 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18405 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18406 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18407 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18408 (create_new_ivs): Remove useless dump.
18410 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18412 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18414 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18415 (iv_ca_set_no_cp, create_new_iv): Ditto.
18417 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18419 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18421 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18423 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18424 function tree_check2.
18426 2017-05-02 Martin Liska <mliska@suse.cz>
18428 * doc/gcov.texi: Add missing preposition.
18429 * gcov.c (function_info::function_info): Properly fill up
18430 all member variables.
18432 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18434 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18436 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18438 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18440 2017-05-02 Martin Liska <mliska@suse.cz>
18443 * lto-streamer-in.c (lto_read_tree_1): Remove
18444 LTO_STREAMER_DEBUG.
18445 * lto-streamer.c (struct tree_hash_entry): Likewise.
18446 (struct tree_entry_hasher): Likewise.
18447 (tree_entry_hasher::hash): Likewise.
18448 (tree_entry_hasher::equal): Likewise.
18449 (lto_streamer_init): Likewise.
18450 (lto_orig_address_map): Likewise.
18451 (lto_orig_address_get): Likewise.
18452 (lto_orig_address_remove): Likewise.
18453 * lto-streamer.h: Likewise.
18454 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18455 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18457 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18459 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18460 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18461 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18462 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18463 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18464 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18465 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18466 (mm_maskz_sub_ss): New intrinsics.
18467 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18468 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18469 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18470 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18471 (__builtin_ia32_subss_mask_round): New builtins.
18472 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18473 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18474 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18476 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18477 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18479 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18482 2017-05-02 Martin Jambor <mjambor@suse.cz>
18484 PR tree-optimization/78687
18485 * tree-sra.c (access): New field parent.
18486 (process_subtree_disqualification): New function.
18487 (disqualify_candidate): Call it.
18488 (build_accesses_from_assign): Reset write flag if creating an
18490 (build_access_subtree): Fill in parent field and also prpagate
18491 down grp_write flag.
18492 (create_artificial_child_access): New parameter set_grp_write, set
18493 grp_write to its value.
18494 (propagate_subaccesses_across_link): Also propagate grp_write flag
18496 (propagate_all_subaccesses): Push the closest parent back to work
18497 queue if add_access_to_work_queue returned true.
18499 2017-05-02 Richard Biener <rguenther@suse.de>
18501 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18502 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18503 -fstrict-overflow documentation.
18504 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18505 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18506 flag_strict_overflow.
18507 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18508 * lto-opts.c (lto_write_options): Do not stream it.
18509 * lto-wrapper.c (merge_and_complain): Do not handle it.
18510 * opts.c (default_options_table): Do not set -fstrict-overflow.
18511 (finish_options): Likewise do not clear it when sanitizing.
18512 * simplify-rtx.c (simplify_const_relational_operation): Do not
18513 test flag_strict_overflow.
18515 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18517 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18518 using enabled attribute.
18519 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18520 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18521 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18522 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18523 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18524 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18525 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18526 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18527 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18528 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18530 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18532 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18534 2017-05-02 Richard Biener <rguenther@suse.de>
18536 PR tree-optimization/80591
18538 2017-04-10 Richard Biener <rguenther@suse.de>
18540 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18543 2017-05-02 Richard Biener <rguenther@suse.de>
18545 PR tree-optimization/80549
18546 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18547 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18548 headers if they do not have a preheader.
18550 2017-05-02 Martin Liska <mliska@suse.cz>
18553 * common.opt: Fix typo.
18554 * doc/invoke.texi: Likewise.
18556 2017-05-01 Jan Beulich <jbeulich@suse.com>
18558 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18559 swapping, add (x,x,m,x,n) alternative.
18561 2017-05-01 Nathan Sidwell <nathan@acm.org>
18563 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18564 unnecessary unadjusted_alignment check.
18566 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18569 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18571 * gimplify.c (gimplify_cilk_detach): New function.
18572 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18573 * tree-core.h: Document EXPR_CILK_SPAWN.
18574 * tree.h (EXPR_CILK_SPAWN): Define.
18576 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18578 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18579 to use new fixit_hint representation, using the "replace" logic.
18580 (get_line_span_for_fixit_hint): Likewise.
18581 (layout::print_any_fixits): Likewise.
18582 (selftest::test_one_liner_many_fixits): Rename to...
18583 (selftest::test_one_liner_many_fixits_1): ...this, and update
18584 comment and expected output to reflect that the multiple fix-it
18585 hints are now consolidated into one insertion.
18586 (selftest::test_one_liner_many_fixits_2): New test.
18587 (selftest::test_diagnostic_show_locus_one_liner): Update for
18589 (selftest::test_fixit_consolidation): Update for fix-it API
18591 * diagnostic.c (print_parseable_fixits): Likewise.
18592 * edit-context.c (edited_line::m_line_events): Convert from
18593 auto_vec <line_event *> to auto_vec <line_event>.
18594 (class line_event): Convert from abstract base class to a concrete
18595 class, taking over the role of replace_event.
18596 (class insert_event): Delete.
18597 (class replace_event): Rename to class line_event. Convert to
18599 (edit_context::add_fixits): Reimplement.
18600 (edit_context::apply_insert): Delete.
18601 (edit_context::apply_replace): Rename to...
18602 (edit_context::apply_fixit): ...this. Convert to half-open range.
18603 (edited_file::apply_insert): Delete.
18604 (edited_file::apply_replace): Rename to...
18605 (edited_file::apply_fixit): ...this.
18606 (edited_line::~edited_line): Drop deletion of events.
18607 (edited_line::apply_insert): Delete.
18608 (edited_line::apply_replace): Rename to...
18609 (edited_line::apply_fixit): ...this. Convert to half-open range.
18610 Update for change to type of m_line_events.
18611 * edit-context.h (edit_context::apply_insert): Delete.
18612 (edit_context::apply_replace): Rename to...
18613 (edit_context::apply_fixit): ...this.
18615 2017-05-01 Martin Sebor <msebor@redhat.com>
18617 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18620 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18623 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18624 __get_cpuid_max returns 0.
18625 (__get_cpuid_count): Ditto.
18627 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18629 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18630 replacement expression is another instance of one of its arguments.
18632 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18635 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18636 check for stack push/pop autoinc.
18637 * config/i386/i386.c (ix86_agi_dependent): Return false
18638 if the only reason why modified_in_p returned true is that
18639 addr is SP based and set_insn is a push or pop.
18641 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18643 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18646 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18649 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18650 (account_size_time): Use two predicates - exec_pred and
18652 (evaluate_conditions_for_known_args): Compute both clause and
18654 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18655 (inline_summary_t::duplicate): Update.
18656 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18658 (compute_inline_parameters): Likewise.
18659 (estimate_edge_size_and_time): Update caluclation of time.
18660 (estimate_node_size_and_time): Compute both time and nonspecialized
18662 (estimate_ipcp_clone_size_and_time): Update.
18663 (inline_merge_summary): Update.
18664 (do_estimate_edge_time): Update.
18665 (do_estimate_edge_size): Update.
18666 (do_estimate_edge_hints): Update.
18667 (inline_read_section, inline_write_summary): Stream both new predicates.
18668 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18670 (compute_inlined_call_time): Cleanup.
18671 (big_speedup_p): Update.
18672 (edge_badness): Update.
18673 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18674 (size_time_entry): Replace predicate by exec_predicate and
18675 nonconst_predicate.
18676 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18677 (estimate_edge_time): Return also nonspec_time.
18678 (reset_edge_growth_cache): Update.
18680 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18682 PR rtl-optimization/80491
18683 * ifcvt.c (noce_process_if_block): When looking for x setter
18684 with missing else_bb, don't check only the insn right before
18685 cond_earliest, but look for the last insn that x is modified in
18686 within the same bb.
18688 PR rtl-optimization/80491
18689 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18691 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18693 PR tree-optimization/80487
18694 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18696 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18698 PR tree-optimization/79697
18699 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18700 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18701 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18703 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18704 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18706 2017-04-28 Martin Sebor <msebor@redhat.com>
18708 PR tree-optimization/80523
18709 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18710 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18712 (maybe_warn, format_directive, parse_directive): Use new functions.
18713 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18715 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18717 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18719 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18721 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18722 target_header_dir): Set correctly.
18723 * configure: Regenerated.
18724 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18725 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18726 instead of SYSTEM_HEADER_DIR.
18728 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18730 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18731 (estimate_local_effects): Likewise.
18732 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18733 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18734 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18735 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18736 do_estimate_edge_time, estimate_edge_time): Likewise.
18737 * ipa-inline-analysis.c (estimate_node_size_and_time,
18738 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18739 (estimate_time_after_inlining): Remove.
18741 2017-04-28 Martin Liska <mliska@suse.cz>
18743 * doc/gcov.texi: Enhance documentation of gcov.
18745 2017-04-28 Martin Liska <mliska@suse.cz>
18747 * doc/gcov.texi: Sort options in alphabetic order.
18748 * doc/gcov-dump.texi: Likewise.
18749 * doc/gcov-tool.texi: Likewise.
18750 * gcov.c (print_usage): Likewise.
18751 * gcov-dump.c (print_usage): Likewise.
18752 * gcov-tool.c (print_merge_usage_message): Likewise.
18753 (print_rewrite_usage_message): Likewise.
18754 (print_overlap_usage_message): Likewise.
18756 2017-04-28 Martin Liska <mliska@suse.cz>
18758 PR gcov-profile/53915
18759 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18761 2017-04-28 Martin Liska <mliska@suse.cz>
18763 PR gcov-profile/79891
18764 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18765 is marked by compiler as living on a line.
18766 (get_cycles_count): Remove usage of the union.
18767 (output_intermediate_file): Likewise.
18768 (find_source): Fix GNU coding style.
18769 (accumulate_line_counts): Remove old non-all block mode.
18770 (output_lines): Remove usage of the union.
18771 * profile.c (output_location): Include all BBs, even if
18772 belonging to a same line (and file) as a previous BB.
18774 2017-04-28 Martin Liska <mliska@suse.cz>
18776 * gcov.c (process_args): Handle new argument 'w'.
18777 (read_graph_file): Assign ID to BBs.
18778 (output_branch_count): Display BB # if verbose flag is set.
18779 (output_lines): Likewise for arcs.
18780 (print_usage): Add '--verbose' option help.
18781 * doc/gcov.texi: Document --verbose (-w) option.
18783 2017-04-28 Martin Liska <mliska@suse.cz>
18785 * gcov.c (struct block_location_info): New struct.
18786 (process_file): Fill up the new structure.
18787 (read_graph_file): Replace usage of encoding by the newly added
18789 (add_line_counts): Likewise.
18790 (accumulate_line_counts): Remove usage of the union.
18791 (function_info::function_info): New function.
18792 (function_info::~function_info): Likewise.
18793 (process_file): Call delete instead of release_function.
18794 (release_function): Release the function.
18795 (release_structures): Call delete instead of release_function.
18796 (solve_flow_graph): Replace usage of num_blocks.
18797 (find_exception_blocks): Likewise.
18798 (output_lines): Fix GNU coding style.
18800 2017-04-28 Martin Liska <mliska@suse.cz>
18803 * coverage.c (coverage_remove_note_file): New function.
18804 * coverage.h: Declare the function.
18805 * toplev.c (finalize): Clean if an error has been seen.
18807 2017-04-28 Martin Liska <mliska@suse.cz>
18809 PR gcov-profile/80031
18810 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18811 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18812 * gcov.c (read_graph_file): Read just number of blocks.
18813 * profile.c (branch_prob): Do not stream 0 flags per a basic
18816 2017-04-28 Martin Liska <mliska@suse.cz>
18818 * gcov-dump.c (tag_*): Add new argument to declarations.
18819 (dump_gcov_file): Likewise.
18820 (tag_blocks): Add and use new argument depth.
18821 (tag_arcs): Likewise.
18822 (tag_lines): Likewise.
18823 (tag_counters): Likewise.
18824 (tag_summary): Likewise.
18825 (dump_working_sets): Use depth to do a proper indentation.
18827 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18830 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18831 * symtab.c (symtab_node::debug_symtab): Move definition here.
18833 2017-04-28 Richard Biener <rguenther@suse.de>
18835 * lto-streamer.h (LTO_major_version): Bump to 7.
18837 2017-04-28 Richard Biener <rguenther@suse.de>
18839 * tree-vrp.c (assert_info): New struct.
18840 (add_assert_info): New helper.
18841 (register_edge_assert_for_2): Refactor to add asserts to a vector
18843 (register_edge_assert_for_1): Likewise.
18844 (register_edge_assert_for): Likewise.
18845 (finish_register_edge_assert_for): New helper actually registering
18846 asserts where live on edge.
18847 (find_conditional_asserts): Adjust.
18848 (find_switch_asserts): Likewise.
18849 (evrp_dom_walker::try_find_new_range): Generalize.
18850 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18852 2017-04-27 Marek Polacek <polacek@redhat.com>
18855 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18856 arg10 and arg11 to itype.
18858 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18860 * doc/extend.texi (Object Size Checking): Improve grammar.
18862 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18865 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18866 that the logic for permitting reciprocal estimates matches that
18869 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18872 * tree.c (type_cache_hasher::equal): Only compare
18873 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18874 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18875 non-aggregate element types.
18876 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18877 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18879 2017-04-27 Richard Biener <rguenther@suse.de>
18881 PR middle-end/80533
18882 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18883 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18884 keeping a reference to a trailing array.
18886 2017-04-27 Richard Biener <rguenther@suse.de>
18888 PR middle-end/80539
18889 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18890 being in loop-closed SSA form conservatively.
18891 (chrec_fold_multiply_poly_poly): Likewise.
18893 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18895 PR middle-end/79665
18896 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18897 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18899 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18902 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18903 (aarch64_function_arg_alignment): Return unsigned int again, but still
18904 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18905 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18906 Don't emit -Wpsabi note.
18907 (aarch64_function_arg_boundary): Likewise.
18908 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18911 2017-04-26 Nathan Sidwell <nathan@acm.org>
18913 * tree.h (crc32_unsigned_n): Declare.
18914 (crc32_unsigned, crc32_unsigned): Make inline.
18915 * tree.c (crc32_unsigned_bits): Replace with ...
18916 (crc32_unsigned_n): ... this.
18917 (crc32_unsigned, crc32_byte): Remove.
18918 (crc32_string): Remove unnecessary braces.
18920 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18922 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18923 * ipa-inline-analysis.c (MAX_TIME): Remove.
18924 (account_size_time): Use sreal for time.
18925 (dump_inline_summary): Update.
18926 (estimate_function_body_sizes): Update.
18927 (estimate_edge_size_and_time): Update.
18928 (estimate_calls_size_and_time): Update.
18929 (estimate_node_size_and_time): Update.
18930 (inline_merge_summary): Update.
18931 (inline_update_overall_summary): Update.
18932 (estimate_time_after_inlining): Update.
18933 (inline_read_section): Update.
18934 (inline_write_summary): Update.
18935 * ipa-inline.c (compute_uninlined_call_time): Update.
18936 (compute_inlined_call_time): Update.
18937 (recursive_inlining): Update.
18938 (inline_small_functions): Update.
18939 (dump_overall_stats): Update.
18940 * ipa-inline.h: Include sreal.h.
18941 (size_time_entry): Turn time to sreal.
18942 (inline_summary): Turn self_time nad time to sreal.
18944 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18946 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18948 (sreal::stream_out, sreal::stream_in): New.
18949 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18951 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18953 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18956 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18959 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18961 (dimode_scalar_chain::compute_convert_gain): Ditto.
18962 (dimode_scalar_chain::make_vector_copies): Ditto.
18963 (dimode_scalar_chain::convert_reg): Ditto.
18964 (dimode_scalar_chain::convert_insn): Ditto.
18965 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18966 (VI248_AVX512BW_1): New mode iterator.
18967 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18968 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18971 2017-04-25 Martin Sebor <msebor@redhat.com>
18973 PR tree-optimization/80497
18974 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18975 constants are representable in HOST_WIDE_INT.
18976 (parse_directive): Ditto.
18978 2017-04-25 Martin Sebor <msebor@redhat.com>
18981 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18982 (new_zero_array): Adjust signature.
18983 (dom_info::dom_init): Used unsigned rather that size_t.
18984 (dom_info::dom_info): Same.
18986 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18987 Jakub Jelinek <jakub@redhat.com>
18990 * config/arm/arm.c: Include gimple.h.
18991 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18992 returns negative, increment ncrn only if it returned positive.
18993 (arm_needs_doubleword_align): Return int instead of bool,
18994 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18995 members, but if there is any such non-FIELD_DECL
18996 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18997 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18998 returns negative, increment nregs only if it returned positive.
18999 (arm_setup_incoming_varargs): Likewise.
19000 (arm_function_arg_boundary): Emit -Wpsabi note if
19001 arm_needs_doubleword_align returns negative, return
19002 DOUBLEWORD_ALIGNMENT only if it returned positive.
19004 2017-04-25 Marek Polacek <polacek@redhat.com>
19007 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19008 first argument to type.
19010 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19013 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19014 type checks to test for compatibility instead of equality.
19016 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19017 Jakub Jelinek <jakub@redhat.com>
19020 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19022 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19023 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19024 the alignment computation, but return their maximum in warn_alignment.
19025 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19026 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19028 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19029 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19032 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19034 * config/arc/simdext.md (dmpyh): Fix typo.
19036 2017-04-25 Richard Biener <rguenther@suse.de>
19038 PR tree-optimization/80492
19039 * alias.c (compare_base_decls): Handle registers with asm
19040 specification conservatively.
19041 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19042 compare_base_decls returning dont-know properly.
19044 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19046 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19047 (legitimate_offset_address_p): New function.
19048 (arc_legitimate_address_p): Use above function.
19050 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19052 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19054 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19056 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19057 ACCH registers whenever they are available.
19059 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19061 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19062 double regs fix when not used.
19064 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19066 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19068 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19069 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19071 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19073 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19074 when emitting short ADD instructions.
19076 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19078 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19080 (cmpsi_cc_c_insn): Likewise.
19081 (cbranchsi4_scratch): Compute proper instruction length using
19082 compact_hreg_operand.
19083 * config/arc/predicates.md (compact_hreg_operand): New predicate.
19085 2017-04-25 Richard Biener <rguenther@suse.de>
19087 PR middle-end/80509
19088 * passes.c (pass_manager::pass_manager): Initialize
19089 m_name_to_pass_map.
19091 2017-04-25 Richard Biener <rguenther@suse.de>
19093 PR tree-optimization/79201
19094 * tree-ssa-sink.c (statement_sink_location): Handle calls.
19096 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19099 * config/s390/vector.md: Split MEM->GPR vector moves for
19100 non-s_operand addresses.
19102 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19105 * config/s390/predicates.md (reload_const_wide_int_operand): New
19107 * config/s390/s390.md ("movti"): Remove d/P alternative.
19108 ("movti_bigconst"): New pattern definition.
19110 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19113 * s390-protos.h (s390_expand_cs_hqi): Removed.
19114 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
19115 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
19116 modes as well as CCZ1mode and CCZmode.
19117 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
19118 signature of s390_emit_compare_and_swap.
19119 (s390_expand_cs_hqi): Likewise, make static.
19120 (s390_expand_cs_tdsi): Generate an explicit compare before trying
19121 compare-and-swap, in some cases.
19122 (s390_expand_cs): Wrapper function.
19123 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
19125 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
19126 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
19127 patterns for small and large integers. Forbid symref memory operands.
19128 Move expander to s390.c. Require cc register.
19129 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19130 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19131 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19132 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19133 symref memory operands. Remove CC mode and call s390_match_ccmode
19135 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19137 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19139 * config/s390/s390.md (define_peephole2): New peephole to help
19140 combining the load-and-test pattern with volatile memory.
19142 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19144 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19145 with CCZmode for TARGET_Z196.
19147 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19149 PR rtl-optimization/80501
19150 * combine.c (make_compound_operation_int): Set subreg_code to SET
19151 even for AND with mask of the sign bit of mode.
19153 PR rtl-optimization/80500
19154 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19155 sum's initial value.
19157 2017-04-25 Julian Brown <julian@codesourcery.com>
19158 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19160 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19162 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19164 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19166 2017-04-25 Julian Brown <julian@codesourcery.com>
19167 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19169 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19170 (thunderx2t99_sha): New Reservation.
19172 2017-04-25 Julian Brown <julian@codesourcery.com>
19173 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19175 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19176 type for 1-element load.
19178 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19180 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19182 2017-04-24 Martin Jambor <mjambor@suse.cz>
19184 PR tree-optimization/80293
19185 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19186 char arrays not totally scalarizable if it is false.
19187 (analyze_all_variable_accesses): Pass correct value in the new
19188 parameter. Add a statistics counter.
19190 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19192 PR middle-end/79931
19193 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19195 2017-04-24 Richard Biener <rguenther@suse.de>
19197 PR tree-optimization/80494
19198 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19199 out for complex types.
19201 2017-04-24 Richard Biener <rguenther@suse.de>
19203 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19204 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19205 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19206 (DFS): Adjust and never fail.
19207 (sccvn_dom_walker::fail): Remove.
19208 (sccvn_dom_walker::before_dom_children): Adjust.
19209 (run_scc_vn): Likewise and never fail.
19210 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19211 (pass_fre::execute): Likewise.
19213 2017-04-24 Richard Biener <rguenther@suse.de>
19215 PR tree-optimization/79725
19216 * tree-ssa-sink.c (statement_sink_location): Return whether
19217 failure reason was zero uses. Move that check later.
19218 (sink_code_in_bb): Deal with zero uses by removing the stmt
19221 2017-04-24 Richard Biener <rguenther@suse.de>
19224 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19225 pointer-based references.
19227 2017-04-24 Richard Biener <rguenther@suse.de>
19230 * pass_manager.h (pass_manager::operator new): Remove.
19231 (pass_manager::operator delete): Likewise.
19232 * passes.c (pass_manager::operator new): Remove.
19233 (pass_manager::operator delete): Likewise.
19234 (pass_manager::pass_manager): Zero individual pass members.
19236 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19239 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19240 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19241 Check "XEXP (src, 1)" operand here.
19242 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19243 Check "XEXP (src, 1)" operand here.
19244 (dimode_scalar_chain::make_vector_copies): Detect count register
19245 of a shift instruction. Zero extend count register from QImode
19246 to DImode to satisfy vector shift pattern count operand predicate.
19247 Substitute vector shift count operand with a DImode copy.
19248 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19251 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19253 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19254 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19255 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19256 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19257 (UNSPEC_NOREX_MEM): Remove definition.
19259 2017-04-21 Richard Biener <rguenther@suse.de>
19261 PR tree-optimization/79547
19262 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19263 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19264 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19265 without any constraints.
19267 2017-04-21 Richard Biener <rguenther@suse.de>
19269 PR tree-optimization/78847
19270 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19272 2017-04-21 Richard Biener <rguenther@suse.de>
19274 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19275 (build_distinct_type_copy): Likewise.
19276 (build_variant_type_copy): Likewise.
19277 * tree.c (build_qualified_type): Pass down mem-stat info.
19278 (build_distinct_type_copy): Likewise.
19279 (build_variant_type_copy): Likewise.
19281 2017-04-21 Richard Biener <rguenther@suse.de>
19283 PR tree-optimization/80237
19284 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19286 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19287 for a simplified result.
19289 2016-04-21 Richard Biener <rguenther@suse.de>
19291 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19292 sth as strict as a simple_iv but a chrec without symbols and an
19293 operand defined in the loop we are peeling (and not some subloop).
19294 (propagate_constants_for_unrolling): Propagate all constants.
19296 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19299 * config/i386/i386.c (print_reg): Remove assert for disalowed
19300 regno values, call output_operand_lossage instead.
19302 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19305 * config/i386/constraints.md (Yc): New register constraint.
19306 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19307 Use Yc constraint for alternative 2 of operand 0. Remove
19308 preferred_for_speed attribute.
19310 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19312 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19313 lastprivate clauses in SIMT case.
19315 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19317 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19319 2017-04-20 Richard Biener <rguenther@suse.de>
19321 PR tree-optimization/57796
19322 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19323 as N scalar stores.
19324 (vect_model_load_cost): Cost gathers as N scalar loads.
19326 2017-04-20 Richard Biener <rguenther@suse.de>
19328 * ggc-page.c (ggc_allocated_p): Rename to ...
19329 (safe_lookup_page_table_entry): ... this and return the lookup
19331 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19333 2017-04-20 Richard Biener <rguenther@suse.de>
19335 PR tree-optimization/80453
19336 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19337 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19338 from the conditions.
19339 (vn_phi_eq): Pass them down.
19340 (vn_phi_lookup): Record them.
19341 (vn_phi_insert): Likewise.
19343 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19345 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19346 uninitialized variable warning to avoid buffer overrun.
19348 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19351 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19352 is suppressed for '{ 0 }' in C.
19354 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19356 * BASE-VER: Set to 8.0.0.
19358 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19360 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19361 priority .init_array and .fini_array section with SECTION_NOTYPE
19364 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19366 PR middle-end/80423
19367 * tree.h (build_array_type): Add typeless_storage default argument.
19368 * tree.c (type_cache_hasher::equal): Also compare
19369 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19370 (build_array_type): Add typeless_storage argument, set
19371 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19373 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19374 (build_array_type): Likewise. Add typeless_storage argument.
19376 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19377 Jakub Jelinek <jakub@redhat.com>
19379 PR tree-optimization/80426
19380 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19381 operation on symbolic operands, also compute the overflow for the
19382 invariant part when the operation degenerates into a negation.
19384 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19387 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19388 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19391 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19393 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19396 * config/avr/avr.c (tree.h): Include it.
19397 (cgraph.h): Include it.
19398 (avr_encode_section_info): Don't warn for uninitialized progmem
19399 variable if it's just an alias.
19401 2017-04-19 Richard Biener <rguenther@suse.de>
19404 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19405 when needed by AutoPGO.
19407 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19410 * doc/lto.texi: Remove an extra 'that'.
19412 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19414 PR rtl-optimization/80429
19415 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19416 are only used in debug insns.
19418 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19419 Vladimir Makarov <vmakarov@redhat.com>
19421 * config/sparc/predicates.md (input_operand): Add comment. Return
19422 true for any memory operand when LRA is in progress.
19423 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19425 2017-04-18 Jeff Law <law@redhat.com>
19428 * mips.md ({return,simple_return}_internal): Do not overwrite
19431 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19433 PR tree-optimization/80443
19434 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19435 instead of adding 1, subtract -1 and similarly instead of subtracting
19438 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19440 PR rtl-optimization/80357
19441 * haifa-sched.c (tmp_bitmap): New variable.
19442 (model_recompute): Handle duplicate use records.
19443 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19444 (free_global_sched_pressure_data): Free it.
19446 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19449 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19450 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19451 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19452 instead of SYSTEM_HEADER_DIR.
19454 2017-04-18 Jeff Law <law@redhat.com>
19456 PR middle-end/80422
19457 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19458 predecessors after walking up the insn chain.
19460 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19463 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19464 sizetype type into debug info.
19466 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19469 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19470 unneeded test for TARGET_UPPER_REGS_SF.
19471 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19473 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19476 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19477 instead of gsi_after_labels.
19479 2017-04-18 Jeff Law <law@redhat.com>
19481 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19485 2017-04-13 Jeff Law <law@redhat.com>
19486 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19487 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19489 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19492 * config/avr/avr.c (intl.h): Include it.
19493 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19495 2017-04-18 Martin Liska <mliska@suse.cz>
19497 PR gcov-profile/78783
19498 * gcov-tool.c (gcov_output_files): Validate that destination
19499 file is either removed by the tool or by a user.
19501 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19502 Guy Benyei <guybe@mellanox.com>
19504 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19505 block, and do not negate it, the stored id is already negative.
19507 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19509 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19511 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19514 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19515 masks of options that should be turned off if the VSX vector
19516 options are turned off.
19517 (OTHER_P8_VECTOR_MASKS): Likewise.
19518 (OTHER_VSX_VECTOR_MASKS): Likewise.
19519 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19520 rs6000_disable_incompatible_switches to validate no type switches
19522 (rs6000_incompatible_switch): New function to disallow turning on
19523 other vector options if -mno-vsx, -mno-power8-vector, or
19524 -mno-power9-vector are specified.
19526 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19528 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19530 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19532 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19533 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19534 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19535 (ARG_POINTER_CFA_OFFSET): Likewise.
19537 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19539 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19540 conditions to take advantage of various optimizations.
19542 2017-04-13 Jeff Law <law@redhat.com>
19544 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19545 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19546 (zero_extendsidi2_dext): Likewise.
19548 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19551 * fold-const.c (fold_ternary_loc): Revert
19552 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19555 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19557 PR rtl-optimization/80343
19558 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19559 new scratch pseudo.
19561 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19564 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19565 to ubsan_encode_value.
19567 2017-04-13 Jeff Law <law@redhat.com>
19569 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19570 appearing in DEBUG_INSNs.
19572 2017-04-13 Martin Liska <mliska@suse.cz>
19574 PR gcov-profile/80413
19575 * gcov-io.c (gcov_write_string): Copy to buffer just when
19576 allocated size is greater than zero.
19578 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19581 * dwarf2out.c (decls_for_scope): Ignore declarations of
19582 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19584 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19587 * ipa-visibility.c (non_local_p): Fix typos.
19588 (localize_node): When localizing symbol in same comdat group,
19589 dissolve the group only when we know external symbols are going
19591 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19593 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19595 PR tree-optimization/79390
19596 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19597 order does not result in usable sequence, retry with reversed operand
19603 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19604 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19605 op0 instead of fold_convert_loc (loc, type, arg0).
19607 2017-04-12 Jeff Law <law@redhat.com>
19609 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19610 has a delay slot in the generated code.
19612 * config/cris/cris.md (cris_preferred_reload_class): Return
19613 GENNONACR_REGS rather than GENERAL_REGS.
19615 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19618 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19619 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19620 signedness of the result type.
19622 2017-04-12 Richard Biener <rguenther@suse.de>
19623 Jeff Law <law@redhat.com>
19625 PR tree-optimization/80359
19626 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19627 trim stores to TARGET_MEM_REFs.
19629 2017-04-12 Richard Biener <rguenther@suse.de>
19631 PR tree-optimization/79390
19632 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19633 threading case even more.
19635 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19638 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19639 for quad_address_p for TImode, instead of just not indexed_address.
19641 2017-04-12 Richard Biener <rguenther@suse.de>
19642 Bernd Edlinger <bernd.edlinger@hotmail.de>
19644 PR middle-end/79671
19645 * alias.c (component_uses_parent_alias_set_from): Handle
19646 TYPE_TYPELESS_STORAGE.
19647 (get_alias_set): Likewise.
19648 * tree-core.h (tree_type_common): Add typeless_storage flag.
19649 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19650 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19651 for types containing members with TYPE_TYPELESS_STORAGE.
19652 (place_field): Likewise.
19653 (layout_type): Likewise for ARRAY_TYPE.
19654 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19655 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19656 TYPE_TYPELESS_STORAGE.
19657 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19659 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19662 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19663 first argument to type.
19665 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19669 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19670 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19671 (rs6000_expand_binop_builtin): Likewise.
19672 (rs6000_expand_ternop_builtin): Likewise; also add missing
19673 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19674 vshasigma built-ins.
19675 * doc/extend.texi: Document that vec_xxpermdi's third argument
19676 must be a constant.
19678 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19680 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19681 Use shift_const cost parameter when calculating gain of STV shifts.
19683 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19685 PR rtl-optimization/70478
19686 * lra-constraints.c (process_alt_operands): Check memory for
19687 disfavoring memory insn operand.
19689 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19691 PR middle-end/80100
19692 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19693 left shift in unsigned HOST_WIDE_INT type.
19695 PR rtl-optimization/80385
19696 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19697 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19700 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19701 if they have any depend clauses.
19703 2017-04-11 Martin Liska <mliska@suse.cz>
19706 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19707 * ipa-split.c (split_function): Create a local comdat symbol
19708 if caller is in a comdat group.
19710 2017-04-11 Martin Liska <mliska@suse.cz>
19713 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19716 2017-04-11 Martin Sebor <msebor@redhat.com>
19718 PR middle-end/80364
19719 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19720 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19722 (directive::set_width, directive::set_precision, format_character):
19724 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19727 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19730 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19731 conflict, set target->arch_name instead of target->cpu_name.
19733 2017-04-11 Richard Biener <rguenther@suse.de>
19735 PR tree-optimization/80374
19736 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19737 build_zero_cst, remove fold_convertible_p check again.
19739 2017-04-11 Martin Liska <mliska@suse.cz>
19742 * ubsan.c (instrument_object_size): Do not instrument register
19745 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19748 * config/i386/i386-builtin-types.def
19749 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19750 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19751 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19752 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19753 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19754 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19755 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19756 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19757 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19758 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19759 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19760 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19761 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19762 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19763 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19764 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19765 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19766 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19767 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19768 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19769 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19770 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19771 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19772 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19773 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19774 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19775 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19776 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19777 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19778 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19779 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19780 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19781 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19782 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19783 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19784 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19785 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19786 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19787 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19788 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19789 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19790 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19791 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19792 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19793 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19794 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19796 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19797 flag to second_arg_count, handle 4 argument function type _COUNT
19798 aliases, handle second_arg_count on second argument rather than last.
19800 2017-04-10 Jeff Law <law@redhat.com>
19802 PR tree-optimization/80374
19803 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19804 record anything if we can not convert integer_zero_node to the
19807 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19810 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19811 Enhance special handling given to the TARGET_P9_MINMAX option in
19812 relation to certain other options.
19814 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19816 PR tree-optimization/80153
19817 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19818 remove POINTER_PLUS_EXPR's base part directly, rather than through
19821 2017-04-10 Richard Biener <rguenther@suse.de>
19822 Bin Cheng <bin.cheng@arm.com>
19824 PR tree-optimization/80153
19825 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19826 the first element of pointer type aff_tree. Build result expr in
19828 (add_elt_to_tree): Convert to type unconditionally. Remove other
19829 fold_convert calls.
19830 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19831 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19833 2017-04-10 Richard Biener <rguenther@suse.de>
19835 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19838 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19840 PR rtl-optimization/70478
19841 * lra-constraints.c (curr_small_class_check): New.
19842 (update_and_check_small_class_inputs): New.
19843 (process_alt_operands): Update curr_small_class_check. Disfavor
19844 alternative insn memory operands. Check available regs for small
19847 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19850 * config/mips/mips.opt (-mvirt): Update description.
19851 * doc/invoke.texi (-mvirt): Likewise.
19853 2017-04-10 Richard Biener <rguenther@suse.de>
19855 PR middle-end/80362
19856 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19857 looking for NEGATE_EXPR in -A / -B to A / B folding.
19859 2017-04-10 Martin Liska <mliska@suse.cz>
19861 PR gcov-profile/80224
19862 * gcov.c (print_usage): Fix usage string.
19863 (get_gcov_intermediate_filename): Remove.
19864 (output_gcov_file): Use both for normal and intermediate format.
19865 (generate_results): Do not initialize special file for
19866 intermediate format.
19868 2017-04-10 Richard Biener <rguenther@suse.de>
19870 PR tree-optimization/80304
19871 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19874 2017-04-10 Nathan Sidwell <nathan@acm.org>
19877 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19878 (rs6000_init_builtins): Use it.
19880 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19882 * config/arm/arm.md (<mrc>): Add mode to SET source.
19883 (<mrrc>): Likewise.
19885 2017-04-10 Richard Biener <rguenther@suse.de>
19887 PR middle-end/80344
19888 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19890 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19893 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19894 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19895 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19896 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19897 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19898 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19899 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19900 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19901 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19902 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19903 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19904 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19905 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19906 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19907 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19908 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19909 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19910 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19911 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19912 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19913 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19914 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19915 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19917 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19919 PR rtl-optimization/70478
19920 * lra-constraints.c: Reverse the last patch.
19922 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19924 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19925 Add comment for WCHAR_T.
19927 2017-04-08 Martin Liska <mliska@suse.cz>
19930 2017-04-07 Martin Liska <mliska@suse.cz>
19933 * ipa-split.c (split_function): Add function part to a same comdat
19936 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19939 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19941 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19943 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19944 (vec_pairing): ...to this.
19945 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19946 (rs6000_sched_init): Adjust for name change.
19947 (struct rs6000_sched_context): Likewise.
19948 (rs6000_init_sched_context): Likewise.
19949 (rs6000_set_sched_context): Likewise.
19951 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19957 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19959 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19960 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19961 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19963 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19965 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19967 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19969 PR rtl-optimization/70703
19970 * ira-color.c (update_conflict_hard_regno_costs): Use
19971 int64_t instead of HOST_WIDE_INT.
19973 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19975 PR rtl-optimization/70478
19976 * lra-constraints.c (process_alt_operands): Disfavor alternative
19977 insn memory operands.
19979 2017-04-07 Jeff Law <law@redhat.com>
19981 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19982 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19984 2017-04-07 Martin Liska <mliska@suse.cz>
19987 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19988 Show error message instead of an ICE.
19990 2017-04-07 Martin Liska <mliska@suse.cz>
19993 * ipa-split.c (split_function): Add function part to a same comdat
19996 2017-04-07 Richard Biener <rguenther@suse.de>
19998 PR middle-end/80341
19999 * tree.c (get_unwidened): Also handle ! for_type case for
20001 * convert.c (do_narrow): Split out from ...
20002 (convert_to_integer_1): ... here. Do not pass final truncation
20003 type to get_unwidened for TRUNC_DIV_EXPR.
20005 2017-04-07 Richard Biener <rguenther@suse.de>
20007 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20009 (aff_combination_const): Adjust.
20010 (aff_combination_scale): Likewise.
20011 (aff_combination_add_elt): Likewise.
20012 (aff_combination_add_cst): Likewise.
20013 (aff_combination_convert): Likewise.
20014 (add_elt_to_tree): Likewise. Remove unused argument.
20015 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20017 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20019 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20021 * config/arm/arm.c (arm_default_short_enums): Use
20022 ARM_DEFAULT_SHORT_ENUMS.
20023 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20025 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20028 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20029 members with redundant out-of-class redeclaration.
20031 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20034 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20035 * config/i386/i386.md (*zero_extendsidi2):
20036 Add (?*x,*x) and (?*v,*v) alternatives.
20038 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20041 * config/i386/i386.c (ix86_expand_builtin)
20042 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20043 mode from insn data. Convert operands to insn operand mode.
20044 Copy operands that don't satisfy insn predicate to a register.
20046 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20048 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20051 2017-04-06 Richard Biener <rguenther@suse.de>
20053 PR tree-optimization/80334
20054 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20055 preserve alignment of accesses.
20057 2017-04-06 Richard Biener <rguenther@suse.de>
20059 PR tree-optimization/80262
20060 * tree-sra.c (build_ref_for_offset): Preserve address-space
20062 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20063 Drop useless address-space information on MEM_REF offsets.
20065 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20067 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20069 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20071 PR rtl-optimization/70703
20072 * ira-color.c (update_conflict_hard_regno_costs): Use
20073 HOST_WIDE_INT instead of long.
20075 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20078 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20079 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20081 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
20082 for x86_64 target. Handle -m3dnowa option.
20084 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20086 PR rtl-optimization/70703
20087 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
20088 (update_conflict_hard_regno_costs): Use long instead of unsigned
20089 arithmetic for cost calculation.
20091 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20092 Bernd Edlinger <bernd.edlinger@hotmail.de>
20095 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
20098 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
20101 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20102 ptr_mode with Pmode throughout.
20103 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20104 into probe_stack_range and use DImode.
20106 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
20109 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
20110 call_eh_return is true.
20112 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20114 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
20115 Initialize last_match_fntype_index.
20117 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20120 * tree-nvr.c: Include internal-fn.h.
20121 (pass_return_slot::execute): Ignore internal calls without
20124 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20125 Richard Biener <rguenther@suse.de>
20128 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
20129 captures used multiple times, except for the last use.
20130 * generic-match-head.c: Include gimplify.h.
20132 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20134 PR tree-optimization/79390
20135 * target.h (struct noce_if_info): Declare.
20136 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20137 * target.def (noce_conversion_profitable_p): New target hook.
20138 * ifcvt.h (struct noce_if_info): New type, moved from ...
20139 * ifcvt.c (struct noce_if_info): ... here.
20140 (noce_conversion_profitable_p): Renamed to ...
20141 (default_noce_conversion_profitable_p): ... this. No longer
20143 (noce_try_store_flag_constants, noce_try_addcc,
20144 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20145 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20146 instead of noce_conversion_profitable_p.
20147 * config/i386/i386.c: Include ifcvt.h.
20148 (ix86_option_override_internal): Don't override
20149 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20150 (ix86_noce_conversion_profitable_p): New function.
20151 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20152 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20153 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20154 * doc/tm.texi: Regenerated.
20156 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20158 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20161 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20164 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20165 instructions for small multiply cores.
20167 2017-04-04 Jeff Law <law@redhat.com>
20169 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20171 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20172 that are not set by the loop over the elements.
20174 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20177 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20178 int mode, convert_modes it to mode as unsigned, otherwise use
20179 lowpart_subreg to mode rather than SImode.
20180 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20181 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20182 Use DImode instead of SImode for the shift count operand.
20183 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20186 2017-04-04 Richard Biener <rguenther@suse.de>
20188 PR middle-end/80281
20189 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20190 arithmetic done for the negate or the plus. Simplify.
20191 (A - (-B) -> A + B): Likewise.
20192 * fold-const.c (split_tree): Make sure to not negate pointers.
20194 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20196 PR rtl-optimization/60818
20197 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20198 a compare of comparisons with the thing compared if this results
20199 in a different machine mode.
20201 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20203 * alias.c (base_alias_check): Fix typo in comment.
20204 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20205 * cgraphunit.c (symbol_table::compile): Likewise.
20206 * collect2.c (maybe_run_lto_and_relink): Likewise.
20207 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20208 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20209 * config/avr/avr.c (avr_map_op_t): Likewise.
20210 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20211 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20212 * config/epiphany/epiphany.md (movcc): Likewise.
20213 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20214 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20216 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20217 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20218 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20219 * config/sh/sh.c (sh_rtx_costs): Likewise.
20220 * fold-const.c (fold_truth_andor): Likewise.
20221 * genautomata.c (collapse_flag): Likewise.
20222 * gengtype.h (struct type::u::s): Likewise.
20223 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20224 * input.c (FORMAT_AMOUNT): Likewise.
20225 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20226 (known_aggs_to_agg_replacement_list): Likewise.
20227 * ipa-inline-analysis.c: Likewise.
20228 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20229 * ipa-polymorphic-call.c
20230 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20231 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20232 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20234 * modulo-sched.c (apply_reg_moves): Likewise.
20235 * omp-expand.c (build_omp_regions_1): Likewise.
20236 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20237 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20238 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20239 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20240 * value-prof.c: Likewise.
20241 * var-tracking.c (val_reset): Likewise.
20243 2017-04-03 Richard Biener <rguenther@suse.de>
20245 PR tree-optimization/80275
20246 * fold-const.c (split_address_to_core_and_offset): Handle
20249 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20251 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20252 descriptors is at least equal to that of functions.
20254 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20256 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20258 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20261 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20262 (mov<IMOD4:mode>): New expander.
20263 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20265 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20267 PR rtl-optimization/79405
20268 * fwprop.c (propagations_left): New variable.
20269 (forward_propagate_into): Decrement it.
20270 (fwprop_init): Initialize it.
20271 (fw_prop): If the variable has reached zero, stop propagating.
20272 (fwprop_addr): Ditto.
20274 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20277 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20278 a FUNCTION_DECL, pass it as decl instead of origin to
20281 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20283 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20286 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20289 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20290 TARGET_VSX_SMALL_INTEGER.
20292 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20294 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20295 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20297 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20299 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20300 extraction from odd-numbered MSA register.
20302 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20304 PR middle-end/80173
20305 * expmed.c (store_bit_field_1): Don't attempt to create
20306 a word subreg out of hard registers wider than word if they
20307 have HARD_REGNO_NREGS of 1 for their mode.
20309 PR middle-end/80163
20310 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20311 conversions to integer types wider than word and pointer.
20314 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20315 (rtx_equal_for_cselib_p): Pass 0 to it.
20316 * cselib.c (cselib_hasher::equal): Likewise.
20317 (rtx_equal_for_cselib_1): Add depth argument. If depth
20318 is 128, don't look up VALUE locs and punt. Increment
20319 depth in recursive calls when walking VALUE locs.
20321 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20323 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20324 (make_gcov_file_name): Use the canonical path name for generating
20326 (read_line): Fix handling of files with ascii null bytes.
20328 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20330 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20331 to initialise a vector register instead
20332 of using a const_int.
20334 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20336 PR translation/80189
20337 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20338 diagnostic messages.
20340 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20343 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20344 (dfp_diex_<mode>): Update mode of operand 1.
20345 * doc/extend.texi (dxex, dxexq): Document change to return type.
20346 (diex, diexq): Document change to argument type.
20348 2017-03-30 Martin Jambor <mjambor@suse.cz>
20351 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20352 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20353 it reflects the signature changes performed at the callee side.
20354 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20355 to cgraph_build_function_type_skip_args.
20356 (build_function_decl_skip_args): Adjust call to the above function.
20358 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20361 * config/i386/sse.md
20362 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20363 register as dest whenever it is a MEM not rtx_equal_p to the
20364 corresponding dup operand, and when forcing into reg move the
20365 reg into the memory afterwards.
20366 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20367 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20368 for the force_reg mode.
20369 (avx512vl_vextractf128<mode>): Use register as dest either
20370 always when a MEM, or when it is a MEM not rtx_equal_p to the
20371 corresponding dup operand, or even not when it is a CONST_VECTOR
20372 depending on the mode and lo vs. hi.
20373 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20375 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20376 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20377 Likewise. Require that operands[2] is even.
20378 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20379 Remove extraneous parens. Require that operands[2] is a multiple
20381 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20382 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20383 disallow memory then.
20385 2017-03-30 Richard Biener <rguenther@suse.de>
20387 PR tree-optimization/77498
20388 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20389 to non-constants over backedges.
20391 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20393 PR rtl-optimization/80233
20394 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20395 as last_combined_insn. Do not test for BARRIER_P separately.
20397 2017-03-29 Andreas Schwab <schwab@suse.de>
20400 * calls.c (prepare_call_address): Convert funexp to Pmode before
20401 copying to temp reg.
20403 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20405 PR tree-optimization/80158
20406 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20407 Handle possible future case of more than one alternate
20409 (replace_rhs_if_not_dup): Likewise.
20410 (replace_one_candidate): Likewise.
20412 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20414 PR rtl-optimization/80193
20415 * ira.c (ira): Do not check allocation for LRA.
20417 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20419 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20420 (nvptx_output_simt_exit): Declare.
20421 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20422 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20423 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20424 (nvptx_declare_function_name): ...here. Emit declaration of local
20425 memory space buffer for omp_simt_enter insn.
20426 (nvptx_output_unisimt_switch): New.
20427 (nvptx_output_softstack_switch): New.
20428 (nvptx_output_simt_enter): New.
20429 (nvptx_output_simt_exit): New.
20430 * config/nvptx/nvptx.h (struct machine_function): New fields
20431 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20432 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20433 (UNSPECV_SIMT_EXIT): Ditto.
20434 (omp_simt_enter_insn): New insn.
20435 (omp_simt_enter): New expansion.
20436 (omp_simt_exit): New insn.
20437 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20439 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20440 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20441 (expand_GOMP_SIMT_EXIT): New.
20442 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20443 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20444 (GOMP_SIMT_EXIT): Ditto.
20445 * target-insns.def (omp_simt_enter): New insn.
20446 (omp_simt_exit): Ditto.
20447 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20449 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20450 (lower_rec_input_clauses): Likewise.
20451 (lower_lastprivate_clauses): Handle SIMT privatization.
20453 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20454 (ompdevlow_adjust_simt_enter): New.
20455 (find_simtpriv_var_op): New.
20456 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20457 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20459 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20460 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20461 (copy_decl_for_dup_finish): Ditto.
20463 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20465 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20468 * config/i386/i386.c (ix86_option_override_internal): Always
20469 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20471 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20473 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20475 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20477 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20478 mark new edge's irreducible flag accordign to it.
20479 (vect_do_peeling): Check loop preheader edge's irreducible flag
20480 and pass it to function slpeel_add_loop_guard.
20482 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20484 PR tree-optimization/80218
20485 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20486 Update block frequencies and counts.
20488 2017-03-28 Richard Biener <rguenther@suse.de>
20490 PR tree-optimization/78644
20491 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20492 of a simplification result we may not use it at all.
20494 2017-03-28 Richard Biener <rguenther@suse.de>
20497 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20498 without arguments, generate default definition of a SSA name.
20500 2017-03-28 Richard Biener <rguenther@suse.de>
20502 PR middle-end/80222
20503 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20504 TYPE_REF_CAN_ALIAS_ALL references.
20505 * fold-const.c (fold_indirect_ref_1): Likewise.
20507 2017-03-28 Martin Liska <mliska@suse.cz>
20510 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20511 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20513 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20514 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20516 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20517 (EXTRA_SPECS): Define.
20518 (SUBTARGET_EXTRA_SPECS): Likewise.
20519 (SUBTARGET_CPP_SPEC): Likewise.
20520 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20521 SUBTARGET_EXTRA_SPECS.
20522 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20524 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20526 * config/arc/simdext.md (vst64_insn): Update pattern.
20527 (vld32wh_insn): Likewise.
20528 (vld32wl_insn): Likewise.
20529 (vld64_insn): Likewise.
20530 (vld32_insn): Likewise.
20532 2017-03-28 Marek Polacek <polacek@redhat.com>
20535 * fold-const.c (fold_comparison): Use protected_set_expr_location
20536 instead of SET_EXPR_LOCATION.
20538 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20540 * tree.c (add_expr): Avoid name lookup warning.
20542 2017-03-27 Jeff Law <law@redhat.com>
20544 PR tree-optimization/80216
20545 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20546 function name. Limit recursion depth.
20547 (record_temporary_equivalences): Corresponding changes.
20549 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20551 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20554 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20557 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20559 * cfgcleanup.c (reg_note_cfa_p): New array.
20560 (insns_have_identical_cfa_notes): New function.
20561 (old_insns_match_p): Don't cross-jump in between /f
20562 and non-/f instructions. If both i1 and i2 are frame related,
20563 verify all CFA notes, their order and content.
20565 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20568 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20569 HImode and SImode with zero extend to DImode to one insn.
20570 (bswap<mode>2_extenddi): Likewise.
20571 (bswapsi2_extenddi): Likewise.
20572 (bswaphi2_extendsi): Likewise.
20573 (bswaphi2): Combine bswap HImode and SImode into one insn.
20574 Separate memory insns from swapping register.
20575 (bswapsi2): Likewise.
20576 (bswap<mode>2): Likewise.
20577 (bswaphi2_internal): Delete, no longer used.
20578 (bswapsi2_internal): Likewise.
20579 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20580 store, and gpr<-gpr swap insns.
20581 (bswap<mode>2_store): Likewise.
20582 (bswaphi2_reg): Register only splitter, combine with the splitter.
20583 (bswaphi2 splitter): Likewise.
20584 (bswapsi2_reg): Likewise.
20585 (bswapsi2 splitter): Likewise.
20586 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20587 the insns into load, store, and register/register insns.
20588 (bswapdi2_ldbrx): Likewise.
20589 (bswapdi2_load): Likewise.
20590 (bswapdi2_store): Likewise.
20591 (bswapdi2_reg): Likewise.
20593 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20595 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20596 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20598 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20601 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20603 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20604 special handling for target option conflicts between dform
20605 options (-mpower9-dform, -mpower9-dform-vector,
20606 -mpower9-dform-scalar) and -mno-direct-move.
20608 2017-03-27 Richard Biener <rguenther@suse.de>
20610 PR tree-optimization/80181
20611 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20613 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20615 * config/arc/predicates.md (move_double_src_operand): Replace the
20616 call to move_double_src_operand with a call to address_operand.
20618 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20620 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20621 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20622 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20624 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20626 * config/arc/predicates.md (long_immediate_loadstore_operand):
20627 Consider scaled addresses cases.
20629 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20631 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20632 restored when in interrupt.
20633 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20634 doesn't have delay slot.
20636 2017-03-27 Richard Biener <rguenther@suse.de>
20639 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20640 inlined thunk clones.
20642 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20645 * asan.c (instrument_derefs): Copy over last operand from
20646 original COMPONENT_REF to the new COMPONENT_REF with
20647 DECL_BIT_FIELD_REPRESENTATIVE.
20648 * ubsan.c (instrument_object_size): Likewise.
20650 2017-03-27 Richard Biener <rguenther@suse.de>
20652 PR tree-optimization/80170
20653 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20654 sure DR/SCEV didnt fold in constants we do not see when looking
20655 at the reference base alignment.
20657 2017-03-27 Richard Biener <rguenther@suse.de>
20659 PR middle-end/80171
20660 * gimple-fold.c (fold_ctor_reference): Properly guard against
20661 NULL return value from canonicalize_constructor_val.
20663 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20666 * config/i386/i386.c (ix86_expand_builtin)
20667 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20668 flags reg setting and flags reg using instructions.
20669 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20670 clobbering instructions to zero extend op2.
20672 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20674 * doc/install.texi (Configuration) <--with-aix-soname>:
20675 Update link to AIX ld.
20677 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20679 PR rtl-optimization/80160
20680 PR rtl-optimization/80159
20681 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20682 reg_alternate_class into account.
20684 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20687 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20688 to consider in curr_insn_transform.
20690 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20692 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20693 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20694 and emit_mode_inner.
20696 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20698 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20699 argument to the overloaded builtin variants. Use the new flag to
20700 deprecate certain builtin variants.
20701 * config/s390/s390-builtin-types.def: Add new builtin types.
20702 * config/s390/s390-builtins.h: Support new flags field for
20703 overloaded builtins.
20704 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20705 (s390_macro_to_expand): Enable vector float data type.
20706 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20707 builtins by incrementing the __VEC__ version number.
20708 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20710 (s390_resolve_overloaded_builtin): Emit error messages depending
20711 on the builtin flags.
20712 * config/s390/s390.c (s390_expand_builtin): Support additional
20713 flags argument. Change error message to match the messages
20714 emitted in s390-c.c.
20715 * config/s390/s390.md: New UNSPEC_* constants.
20716 (op_type): Add new instruction types.
20717 * config/s390/vecintrin.h: Add new builtins and test data class
20719 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20720 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20721 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20722 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20723 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20724 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20726 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20727 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20728 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20729 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20731 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20732 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20733 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20734 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20736 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20737 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20738 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20739 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20740 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20741 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20742 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20744 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20745 ("vec_scatter_element<V_HW_4:mode>_DI")
20746 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20747 ("vec_fpint<mode>", "vflls")
20748 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20749 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20750 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20751 ("*vec_cmphe<mode>_cc"): ... these.
20753 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20754 mode constant instead of magic value.
20756 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20758 * config/s390/s390.c (s390_expand_vec_compare): Support other
20759 vector floating point modes than just V2DF.
20760 (s390_expand_vcond): Likewise.
20761 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20762 (s390_cannot_change_mode_class): Prevent mode changes between TF
20763 and V1TF in vector registers.
20764 * config/s390/s390.md (DF, SF): New mode attributes.
20765 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20766 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20767 SFmode support for VRs.
20768 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20770 (VFT, VF_HW): New mode iterators.
20771 (vw, sdx): New mode attributes.
20772 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20773 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20774 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20775 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20776 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20777 also the new vector floating point modes. Renaming to ...
20779 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20780 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20781 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20782 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20783 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20784 ("vec_unordered<mode>"): ... these.
20786 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20787 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20788 ("*vec_extendv2df"): New insn definitions.
20790 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20792 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20793 ("mulditi3_2", "*muldi3_sign"): New patterns.
20794 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20795 rename the pattern definition.
20797 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20799 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20801 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20803 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20805 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20806 instruction if possible.
20807 * config/s390/vector.md (vec_halfnumelts): New mode
20809 ("*vec_vllezlf<mode>"): New pattern.
20811 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20813 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20814 ("popcountv4si2", "popcountv2di2"): Rename to ...
20815 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20816 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20818 ("popcount<mode>2_vxe"): New pattern.
20820 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20822 * common/config/s390/s390-common.c (processor_flags_table): Add
20824 * config.gcc: Add arch12.
20825 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20826 Default to arch12 for unknown CPU model numbers.
20827 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20828 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20829 PROCESSOR_max sanity check.
20830 * config/s390/s390-opts.h (enum processor_type): Add
20832 * config/s390/s390.c (processor_table): Add arch12.
20833 (s390_expand_builtin): Add check for B_VXE flag.
20834 (s390_issue_rate): Add PROCESSOR_ARCH12.
20835 (s390_get_sched_attrmask): Likewise.
20836 (s390_get_unit_mask): Likewise.
20837 (s390_sched_score): Enable z13 scheduling for arch12.
20838 (s390_sched_reorder): Likewise.
20839 (s390_sched_variable_issue): Likewise.
20840 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20842 (s390_tune_attr): Use z13 scheduling also for arch12.
20843 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20844 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20845 (TARGET_VXE_P): New macros.
20846 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20847 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20848 * config/s390/s390.opt: Add arch12 as processor_type.
20850 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20852 * config/s390/s390.md
20853 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20854 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20855 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20857 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20858 Rename expanders to ...
20860 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20861 ("fixuns_truncdddi2_emu"): ... these.
20863 ("fixuns_trunc<mode>si2_emu"): New expander.
20865 ("*fixuns_truncdfdi2_z13"): Rename to ...
20866 ("*fixuns_truncdfdi2_vx"): ... this.
20868 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20870 * config/s390/2964.md: Remove the single element vector compare
20871 instructions which are no longer used.
20872 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20874 (s390_canonicalize_comparison): Remove handling of DFmode
20876 (s390_expand_vec_compare_scalar): Remove function.
20877 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20878 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20880 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20882 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20884 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20886 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20887 will anyway by matched by mov<mode>_64dfp.
20889 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20891 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20892 vlef/vstef. Add missing operand to vleif.
20894 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20896 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20897 pair for all vector types with 64 bit elements.
20898 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20899 * config/s390/vector.md (V_HW_64): ... here.
20900 (V_128_NOSINGLE): New mode iterator.
20901 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20902 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20903 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20904 ("*vec_load_pairv2di"): Change to ...
20905 ("*vec_load_pair<mode>"): ... this one.
20907 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20909 * config/s390/constraints.md: Add comments.
20910 (jKK): Reject element sizes > 8 bytes.
20911 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20913 * config/s390/s390.md: Add the s_operand checks formerly in
20914 s390_split_ok_p to various splitters where they are still
20916 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20917 for 128 bit vectors. Plus two splitters.
20919 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20921 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20924 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20927 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20928 error if the boundary argument is not constant.
20930 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20932 PR rtl-optimization/80112
20933 * loop-doloop.c (doloop_condition_get): Don't check condition
20934 if cmp isn't SET with IF_THEN_ELSE src.
20936 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20938 PR tree-optimization/80158
20939 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20940 replacing a candidate statement, also replace it for the
20941 candidate's alternate interpretation.
20942 (replace_rhs_if_not_dup): Likewise.
20943 (replace_one_candidate): Likewise.
20945 2017-03-24 Richard Biener <rguenther@suse.de>
20947 PR tree-optimization/80167
20948 * graphite-isl-ast-to-gimple.c
20949 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20951 (translate_isl_ast_to_gimple::get_rename): Likewise.
20953 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20955 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20956 handling of certain combinations of target options, including the
20957 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20958 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20960 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20963 * config/arm/arm.md (*load_multiple): Add reload_completed to
20964 matching condition.
20966 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20967 Richard Biener <rguenth@suse.de>
20969 PR tree-optimization/79908
20970 PR tree-optimization/80136
20971 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20972 been cast away, gimplify_and_add suffices.
20974 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20976 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20978 2017-03-23 Richard Biener <rguenther@suse.de>
20980 PR tree-optimization/80032
20981 * gimplify.c (gimple_push_cleanup): Forced unconditional
20982 cleanups still have to go to the conditional_cleanups
20985 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20987 PR tree-optimization/80072
20988 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20990 (next_operand_entry_id): Change type to unsigned int.
20991 (sort_by_operand_rank): Make sure to return the right return value
20992 even if unsigned fields are bigger than INT_MAX.
20993 (struct oecount): Change cnt and id type to unsigned int.
20994 (oecount_hasher::equal): Formatting fix.
20995 (oecount_cmp): Make sure to return the right return value
20996 even if unsigned fields are bigger than INT_MAX.
20997 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21000 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21001 TREE_READONLY on result if writing it more than once.
21004 * doc/invoke.texi (-fsanitize=thread): Document that with
21005 -fnon-call-exceptions atomics are not able to throw
21009 * tsan.c: Include tree-eh.h.
21010 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21011 maybe_clean_or_replace_eh_stmt where needed.
21012 (instrument_memory_accesses): Add cfg_changed argument.
21013 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21014 if it returned true.
21015 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21017 PR rtl-optimization/63191
21018 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21019 wrapper function, moved the whole old content into ...
21020 (ix86_delegitimize_address_1): ... this. New inline function.
21021 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21022 true as last argument instead of ix86_delegitimize_address.
21024 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21026 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21027 cortexa57_branch_cost.
21029 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21031 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21033 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21036 * doc/md.texi (Constraints): Document wA constraint.
21037 * config/rs6000/constraints.md (wA): New.
21038 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21039 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21040 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21041 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21043 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21046 * gimplify.c (is_oacc_declared): New function.
21047 (oacc_default_clause): Use it to set default flags for acc declared
21048 variables inside parallel regions.
21049 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21050 declared variables.
21051 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21052 declare attribute to any decl as necessary.
21054 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21057 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21058 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21059 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21060 (arm_arch_lpae): This.
21061 * config/arm/arm.c (arm_arch7ve): Rename into ...
21062 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21063 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21066 2017-03-22 Martin Liska <mliska@suse.cz>
21069 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21070 error message instead of an ICE.
21072 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21074 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21076 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21078 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21080 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21083 2017-03-21 Martin Sebor <msebor@redhat.com>
21085 * doc/extend.texi: Use "cannot" instead of "can't."
21086 * doc/hostconfig.texi: Same.
21087 * doc/install.texi: Same.
21088 * doc/invoke.texi: Same.
21089 * doc/loop.texi: Same.
21090 * doc/md.texi: Same.
21091 * doc/objc.texi: Same.
21092 * doc/rtl.texi: Same.
21093 * doc/tm.texi: Same.
21094 * doc/tm.texi.in: Same.
21095 * doc/trouble.texi: Same.
21097 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
21100 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
21101 (collect_checksum_attributes): Set it.
21102 (die_checksum_ordered): Use it.
21104 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21106 PR tree-optimization/79908
21107 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
21108 change: For a VA_ARG whose LHS has been cast away, use
21109 force_gimple_operand to construct the side effects.
21111 2017-03-21 David Malcolm <dmalcolm@redhat.com>
21113 PR translation/80001
21114 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
21115 more amenable to translation.
21116 (oacc_loop_auto_partitions): Likewise.
21118 2017-03-21 Marek Polacek <polacek@redhat.com>
21119 Martin Sebor <msebor@redhat.com>
21121 PR tree-optimization/80109
21122 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
21123 on INTEGRAL_TYPE_P.
21125 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21126 Segher Boessenkool <segher@kernel.crashing.org>
21129 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21130 check reg_used_between_p between insn and one of succ or succ2
21131 depending on if succ is artificial insn not inserted into insn
21134 2017-03-21 Martin Liska <mliska@suse.cz>
21136 PR gcov-profile/80081
21137 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21138 * doc/gcc.texi: Include gcov-dump stuff.
21139 * doc/gcov-dump.texi: New file.
21141 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21143 PR rtl-optimization/79150
21144 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21145 conditional jump, if the jump is the last insn of the loop.
21147 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21148 Richard Biener <rguenth@suse.de>
21150 PR tree-optimization/79908
21151 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21152 been cast away, use force_gimple_operand to construct the side
21155 2017-03-21 Martin Liska <mliska@suse.cz>
21157 PR libfortran/79956
21158 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21161 2017-03-21 Brad Spengler <spender@grsecurity.net>
21164 * plugin.c (htab_hash_plugin): New function.
21165 (add_new_plugin): Use it and adjust.
21166 (parse_plugin_arg_opt): Adjust.
21167 (init_one_plugin): Likewise.
21169 2017-03-21 Richard Biener <rguenther@suse.de>
21171 PR tree-optimization/80032
21172 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21173 if set force the cleanup to happen unconditionally.
21174 (gimplify_target_expr): Push inserted clobbers with force_uncond
21175 to avoid them being removed by control-dependent DCE.
21177 2017-03-21 Richard Biener <rguenther@suse.de>
21179 PR tree-optimization/80122
21180 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21181 va_arg_pack_len when the inlined call stmt requires pack
21183 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21185 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21188 * tsan.c (instrument_builtin_call): If the memory model argument
21189 is not a constant, assume it is valid.
21192 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21195 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21197 PR rtl-optimization/79910
21198 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21199 if its dest is used by an insn before I2 (other than the combined
21200 insns themselves, which are properly handled already).
21202 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21205 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21207 * combine.c (record_used_regs): New static function.
21208 (try_combine): Handle situations where there is an additional
21209 instruction between I2 and I3 which needs to have a LOG_LINK
21213 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21215 * combine.c (try_combine): Delete redundant i1 test. Call
21216 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21218 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21221 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21222 alternatives 13/14.
21224 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21226 PR tree-optimization/80054
21227 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21228 the optimization if a PHI or any of its arguments is not dominated
21229 by the candidate's basis. Use gphi* rather than gimple* as
21231 (replace_profitable_candidates): Clean up a gimple* variable that
21232 should be a gphi* variable.
21234 2017-03-20 Martin Sebor <msebor@redhat.com>
21237 * doc/extend.texi (attribute constructor): Document present limitation.
21239 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21242 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21243 __POWER9_VECTOR__ #ifdef control, change template definition to
21244 use Power9-specific built-in function.
21245 (vec_any_eq): Likewise.
21246 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21247 to control outcomes from this test.
21248 (vector_ae_<mode>p): For VEC_F modes, likewise.
21250 2017-03-20 Ian Lance Taylor <iant@google.com>
21252 * config/i386/i386.c (ix86_function_regparm): Save an extra
21253 register for -fsplit-stack with DECL_STATIC_CHAIN.
21255 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21258 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21259 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21261 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21263 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21265 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21268 2017-03-20 Marek Polacek <polacek@redhat.com>
21271 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21273 2017-03-20 Richard Biener <rguenther@suse.de>
21275 PR tree-optimization/80113
21276 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21277 allocate extra SSA name for PHI def.
21278 (add_close_phis_to_outer_loops): Likewise.
21279 (add_close_phis_to_merge_points): Likewise.
21280 (copy_loop_close_phi_args): Likewise.
21281 (copy_cond_phi_nodes): Likewise.
21283 2017-03-20 Martin Liska <mliska@suse.cz>
21285 PR middle-end/79753
21286 * tree-chkp.c (chkp_build_returned_bound): Do not build
21287 returned bounds for a LHS that's not a BOUNDED_P type.
21289 2017-03-20 Martin Liska <mliska@suse.cz>
21293 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21294 COMPLEX_CST and VECTOR_CST.
21296 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21299 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21300 target operand. A new splitter adds the clobber statement in case
21301 the target operand is dead anyway.
21303 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21305 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21306 to age-old versions of binutils and glibc.
21308 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21310 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21312 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21314 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21316 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21318 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21319 requirement for binutils 2.13.
21321 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21323 * combine.c (try_combine): Delete redundant i1 test. Call
21324 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21326 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21328 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21329 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21331 <riscv64-*-elf>: Re-arrange section
21332 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21333 <riscv32-*-linux>: Likewise.
21334 <riscv64-*-elf>: Likewise
21335 <riscv64-*-linux>: Likewise.
21337 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21340 * aarch64.opt(verbose-cost-dump): Fix typo.
21342 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21345 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21346 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21348 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21350 * reload.c (find_reloads): When reloading a nonoffsettable address,
21351 use RELOAD_OTHER for it and its address reloads.
21353 PR rtl-optimization/79910
21354 * combine.c (record_used_regs): New static function.
21355 (try_combine): Handle situations where there is an additional
21356 instruction between I2 and I3 which needs to have a LOG_LINK
21359 2017-03-17 Jeff Law <law@redhat.com>
21361 PR tree-optimization/71437
21362 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21363 conditional in the hash table first.
21364 (vrp_dom_walker::before_dom_children): Extract condition from
21365 ASSERT_EXPR. Record condition, its inverion and any implied
21366 conditions as well.
21368 2017-03-17 Marek Polacek <polacek@redhat.com>
21369 Markus Trippelsdorf <markus@trippelsdorf.de>
21371 PR tree-optimization/80079
21372 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21375 2017-03-17 Richard Biener <rguenther@suse.de>
21377 PR middle-end/80075
21378 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21379 Properly verify the LHS before the RHS possibly claims to be
21381 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21384 2017-03-17 Martin Jambor <mjambor@suse.cz>
21386 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21387 (List of -O2 options): Likewise.
21388 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21391 2017-03-17 Tom de Vries <tom@codesourcery.com>
21393 * gcov-dump.c (print_usage): Print bug_report_url.
21395 2017-03-17 Richard Biener <rguenther@suse.de>
21397 PR middle-end/80050
21398 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21399 (parser::peek): Likewise.
21401 2017-03-17 Richard Biener <rguenther@suse.de>
21403 PR tree-optimization/80048
21404 * sese.c (free_sese_info): Properly release rename_map and
21405 copied_bb_map elements.
21407 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21409 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21410 Add linked-list forward and backlinks. Insert on
21411 construction, remove on destruction.
21412 (class pass_store_merging): Add m_stores_head field.
21413 (pass_store_merging::terminate_and_process_all_chains):
21414 Iterate over m_stores_head list.
21415 (pass_store_merging::terminate_all_aliasing_chains):
21417 (pass_store_merging::execute): Check for debug stmts first.
21418 Push new chains onto the m_stores_head stack.
21420 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21423 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21424 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21425 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21427 2017-03-16 Jeff Law <law@redhat.com>
21429 PR tree-optimization/71437
21430 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21431 member function. Implementation moved into after_dom_children
21432 member function and into the threader's thread_outgoing_edges
21434 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21435 some code into new thread_outgoing_edges.
21436 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21437 definition. Simplify marker handling (do it here). Assume we always
21438 have the available expression and the const/copies tables.
21439 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21441 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21442 * tree-vrp.c (equiv_stack): No longer file scoped.
21443 (vrp_dom_walker): New class.
21444 (vrp_dom_walker::before_dom_children): New member function.
21445 (vrp_dom_walker::after_dom_children): Likewise.
21446 (identify_jump_threads): Setup domwalker. Use it rather than
21447 walking edges in a random order by hand. Simplify setup/finalization.
21448 (finalize_jump_threads): Remove.
21449 (vrp_finalize): Do not call identify_jump_threads here.
21450 (execute_vrp): Do it here instead and call thread_through_all_blocks
21453 PR tree-optimization/71437
21454 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21456 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21458 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21459 (dom_opt_dom_walker::thread_across_edge): Remove
21460 handle_dominating_asserts argument. All callers changed.
21461 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21462 changes. Remove calls to lhs_of_dominating_assert. Other
21463 uses of handle_dominating_asserts turn into unconditional code
21464 (simplify_control_stmt_condition_1): Likewise.
21465 (simplify_control_stmt_condition): Likewise.
21466 (thread_through_normal_block, thread_across_edge): Likewise.
21467 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21468 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21469 object if it is not an SSA_NAME.
21470 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21471 before calling into the VRP specific simplifiers.
21472 (identify_jump_threads): Remove handle_dominating_asserts
21475 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21478 * tree-diagnostic.c (default_tree_printer): No longer static.
21479 * tree-diagnostic.h (default_tree_printer): New prototype.
21481 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21483 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21484 Change ins into fmov.
21486 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21488 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21489 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21490 Use h_con constraint for operand 1.
21491 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21492 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21494 2017-03-15 Jeff Law <law@redhat.com>
21496 PR tree-optimization/71437
21497 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21498 (record_temporary_equivalences): Use it.
21500 PR tree-optimization/71437
21501 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21502 tree-ssa-scopedtables.
21503 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21504 (record_conditions, record_cond, vuse_eq): Likewise.
21505 (record_edge_info): Adjust to API tweak of record_conditions.
21506 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21507 (record_temporary_equivalences, optimize_stmt): Likewise.
21508 (eliminate_redundant_computations): Likewise.
21509 (record_equivalences_from_stmt): Likewise.
21510 * tree-ssa-scopedtables.c: Include options.h and params.h.
21511 (vuse_eq): New function, moved from tree-ssa-dom.c
21512 (build_and_record_new_cond): Likewise.
21513 (record_conditions): Likewise. Accept vector of conditions rather
21514 than edge_equivalence structure for first argument.
21515 for the first argument.
21516 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21517 from tree-ssa-dom.c.
21518 (avail_exprs_stack::record_cond): Likewise.
21519 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21520 from tree-ssa-dom.c.
21521 (avail_exprs_stack): Add new member functions lookup_avail_expr
21523 (record_conditions): Declare.
21525 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21528 * lra-constraints.c (process_alt_operands): Increase reject for
21529 reloading an input/output operand.
21531 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21534 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21535 insns to convert from signed/unsigned char/short to IEEE 128-bit
21537 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21539 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21542 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21543 subreg of inner mode for values already in registers.
21545 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21547 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21548 iteration reg is used after the loop.
21550 2017-03-14 Martin Sebor <msebor@redhat.com>
21552 PR tree-optimization/79800
21553 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21554 precision in negative-positive range.
21555 (format_floating): Call non-const overload with adjusted precision.
21557 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21560 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21563 2017-03-14 Martin Sebor <msebor@redhat.com>
21565 PR middle-end/80020
21566 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21567 * builtins.def (aligned_alloc): Use it.
21570 * Makefile.in (GTFILES): Add calls.c.
21571 * calls.c: Include "gt-calls.h".
21573 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21575 PR rtl-optimization/79728
21576 * regs.h (struct target_regs): New field
21577 x_contains_allocatable_regs_of_mode.
21578 (contains_allocatable_regs_of_mode): New macro.
21579 * reginfo.c (init_reg_sets_1): Initialize it, and change
21580 contains_reg_of_mode so it includes global regs as well.
21581 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21582 rather than contains_regs_of_mode.
21584 2017-03-14 Martin Liska <mliska@suse.cz>
21586 * doc/invoke.texi: Document options that can't be combined with
21587 -fcheck-pointer-bounds.
21589 2017-03-14 Martin Liska <mliska@suse.cz>
21591 PR middle-end/79831
21592 * doc/invoke.texi (-Wchkp): Document the option.
21594 2017-03-14 Martin Liska <mliska@suse.cz>
21596 * Makefile.in: Install gcov-dump.
21598 2017-03-14 Martin Liska <mliska@suse.cz>
21600 * multiple_target.c (expand_target_clones): Bail out for
21601 an invalid attribute.
21603 2017-03-14 Richard Biener <rguenther@suse.de>
21605 * alias.c (struct alias_set_entry): Pack properly.
21606 * cfgloop.h (struct loop): Likewise.
21607 * cse.c (struct set): Likewise.
21608 * ipa-utils.c (struct searchc_env): Likewise.
21609 * loop-invariant.c (struct invariant): Likewise.
21610 * lra-remat.c (struct cand): Likewise.
21611 * recog.c (struct change_t): Likewise.
21612 * rtl.h (struct address_info): Likewise.
21613 * symbol-summary.h (function_summary): Likewise.
21614 * tree-loop-distribution.c (struct partition): Likewise.
21615 * tree-object-size.c (struct object_size_info): Likewise.
21616 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21617 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21618 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21619 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21620 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21621 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21622 (struct _stmt_vec_info): Likewise.
21624 2017-03-14 Martin Liska <mliska@suse.cz>
21627 * multiple_target.c (create_dispatcher_calls): Check that
21628 a target can create a function dispatcher.
21630 2017-03-14 Martin Liska <mliska@suse.cz>
21633 * multiple_target.c (expand_target_clones): Drop local.local
21634 flag for default implementation.
21636 2017-03-14 Richard Biener <rguenther@suse.de>
21638 PR tree-optimization/80030
21639 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21641 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21643 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21644 gcc_fallthrough() instead of __attribute__((fallthrough));
21646 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21648 * doc/gcc.texi: Remove "up" link to (DIR).
21649 * doc/gccint.texi: Ditto.
21651 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21653 * doc/install.texi (Specific) <avr>: Remove reference to
21656 2017-03-13 Jeff Law <law@redhat.com>
21658 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21659 attribute rather than comments.
21661 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21662 match_scratch operand is highest.
21664 2017-03-13 Martin Liska <mliska@suse.cz>
21666 PR middle-end/78339
21667 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21668 is a CHKP clone, use original declaration.
21670 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21672 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21673 (arc_conditional_register_usage): Use a different allocation order
21674 when optimizing for size.
21675 * common/config/arc/arc-common.c (arc_option_optimization_table):
21676 Section anchors default on when optimizing for size.
21678 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21680 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21682 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21684 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21685 * config/arc/arc.md (cpu_facility): Add cd variant.
21686 (*movqi_insn): Add code density variant.
21687 (*movhi_insn): Likewise.
21688 (*movqi_insn): Likewise.
21689 (*addsi3_mixed): Likewise.
21690 (subsi3_insn): Likewise.
21692 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21694 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21696 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21698 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21699 expressions with MINUS and UNARY ops.
21701 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21704 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21706 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21707 between vec_select and vector argument.
21708 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21709 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21710 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21711 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21712 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21713 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21715 2017-03-13 Richard Biener <rguenther@suse.de>
21718 * params.def (vect-max-peeling-for-alignment): Fix typo.
21720 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21722 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21723 issue that only occurred with binutils below 2.18.
21725 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21727 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21728 refer to binutils 2.11/2.12 minimum.
21730 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21732 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21733 ftp.kernel.org and simplify binutils requirement.
21735 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21737 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21739 (Optimize Options): Ditto. Also remove redundancy.
21741 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21743 PR translation/79848
21744 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21746 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21747 to G_ to avoid double translation.
21749 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21751 PR translation/79923
21752 * auto-profile.c (get_combined_location): Convert leading
21753 character of diagnostics to lower case and remove trailing period.
21754 (read_profile): Likewise for various diagnostics.
21755 * config/arm/arm.c (arm_option_override): Remove trailing period
21756 from various diagnostics.
21757 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21758 (msp430_expand_delay_cycles): Likewise.
21760 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21763 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21764 full command-line argument, rather than just "str".
21765 (aarch64_validate_march): Likewise.
21766 (aarch64_validate_mtune): Likewise.
21768 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21770 PR rtl-optimization/78911
21771 * lra-assigns.c (must_not_spill_p): New function.
21772 (spill_for): Use it.
21774 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21776 PR tree-optimization/79981
21777 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21778 ATOMIC_COMPARE_EXCHANGE ifn result.
21779 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21780 IFN_ATOMIC_COMPARE_EXCHANGE.
21782 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21785 * opts.c (parse_sanitizer_options): Add missing question mark to
21786 "did you mean" message.
21788 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21790 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21792 (VMULEUH_UNS): Likewise.
21793 (VMULOUB_UNS): Likewise.
21794 (VMULOUH_UNS): Likewise.
21795 * config/rs6000/rs6000.c (builtin_function_type): Remove
21796 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21798 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21801 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21802 x with result of extra_parsing_for_operand_code_0.
21803 (function_reader::extra_parsing_for_operand_code_0): Convert
21804 return type from void to rtx, returning x. When reading
21805 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21806 larger size containing struct block_symbol.
21808 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21810 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21811 -mfloat128-hardware without -m64.
21813 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21816 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21817 entries to the case statement that marks unsigned arguments to
21818 overloaded functions.
21820 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21822 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21823 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21825 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21828 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21829 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21831 2017-03-10 Martin Liska <mliska@suse.cz>
21835 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21836 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21839 2017-03-10 Olivier Hainque <hainque@adacore.com>
21841 * tree-switch-conversion (array_value_type): Start by resetting
21842 candidate type to it's main variant.
21844 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21846 PR rtl-optimization/79909
21847 * combine.c (try_combine): Use simplify_replace_rtx on individual
21848 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21849 of the whole CALL_INSN_FUNCTION_USAGE.
21851 PR tree-optimization/79972
21852 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21853 get_range_info on SSA_NAMEs. Formatting fixes.
21855 2017-03-10 Richard Biener <rguenther@suse.de>
21856 Jakub Jelinek <jakub@redhat.com>
21858 PR tree-optimization/77975
21859 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21860 edge to be constant.
21861 (get_val_for): For constant x return it. Formatting fix.
21862 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21863 would use the same bases as the current one.
21865 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21867 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21868 instead of vec_select for V1TImode.
21869 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21871 (VSX_LE_128): Add V1TI to this mode iterator.
21872 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21873 (*vsx_le_perm_store_<mode>): Likewise.
21874 (pre-reload splitter for VSX stores): Likewise.
21875 (post-reload splitter for VSX stores): Likewise.
21876 (*vsx_xxpermdi2_le_<mode>): Likewise.
21877 (*vsx_lxvd2x2_le_<mode>): Likewise.
21878 (*vsx_stxvd2x2_le_<mode>): Likewise.
21880 2017-03-09 Michael Eager <eager@eagercon.com>
21882 Correct failures with --enable-checking=yes,rtl.
21884 * config/microblaze/microblaze.c (microblaze_expand_shift):
21885 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21886 test for const0_rtx.
21887 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21888 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21890 2017-03-09 Richard Biener <rguenther@suse.de>
21892 PR tree-optimization/79977
21893 * graphite-scop-detection.c (scop_detection::merge_sese):
21894 Handle the case of extra exits to blocks dominating the entry.
21896 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21898 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21901 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21903 PR rtl-optimization/79949
21904 * lra-constraints.c (process_alt_operands): Check memory when
21905 trying to predict a cycle. Print about the overall increase.
21907 2017-03-09 Richard Biener <rguenther@suse.de>
21909 PR middle-end/79971
21910 * gimple-expr.c (useless_type_conversion_p): Preserve
21911 TYPE_SATURATING for fixed-point types.
21913 2017-03-09 Richard Biener <rguenther@suse.de>
21916 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21917 alignment of BLKmode params.
21919 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21922 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21923 (VALL_NO_V2Q): Likewise.
21925 * config/aarch64/aarch64-simd.md
21926 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21928 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21929 VALL_NO_V2Q mode iterator.
21930 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21932 2017-03-09 Martin Liska <mliska@suse.cz>
21934 PR tree-optimization/79631
21935 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21936 tree_int_cst_sign_bit just for INTEGER constants.
21938 2017-03-09 Martin Liska <mliska@suse.cz>
21942 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21945 2017-03-09 Marek Polacek <polacek@redhat.com>
21948 * tree.c (inchash::add_expr): Handle TREE_VEC.
21950 2017-03-09 Martin Liska <mliska@suse.cz>
21953 (chkp_narrow_size_and_offset): New function.
21954 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21955 (void chkp_parse_bit_field_ref): New function.
21956 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21957 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21959 2017-03-09 Martin Liska <mliska@suse.cz>
21962 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21963 (chkp_find_bounds_1): Remove gcc_unreachable.
21965 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21968 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21969 BUILT_IN_SYNC*, determine the access type from the size suffix and
21970 always build a MEM_REF with that type. Handle forgotten
21971 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21974 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21975 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21976 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21977 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21978 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21979 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21980 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21981 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21982 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21983 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21984 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21985 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21986 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21987 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21988 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21989 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21990 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21991 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21992 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21993 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21994 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21995 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21996 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21997 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21998 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21999 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22000 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22001 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22002 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22003 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22004 definitions outside of __OPTIMIZE__ guarded section.
22007 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22008 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22009 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22010 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22013 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22015 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22016 ("vfenez<mode>"): Add missing constraints.
22018 2017-03-08 Martin Sebor <msebor@redhat.com>
22021 * config/nds32/nds32.c (nds32_option_override):
22022 Fix misspelled diagnostic.
22024 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22027 * gimplify.c (gimplify_omp_for): Replace index var in outer
22028 taskloop statement with an artificial variable and add
22029 OMP_CLAUSE_PRIVATE clause for it.
22031 2017-03-08 Richard Biener <rguenther@suse.de>
22033 PR tree-optimization/79955
22034 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22035 for accesses that are completely outside of the variable.
22037 2017-03-08 Andrew Haley <aph@redhat.com>
22039 PR tree-optimization/79943
22040 * tree-ssa-loop-split.c (compute_new_first_bound): When
22041 calculating the new upper bound, (END-BEG) should be added, not
22044 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22046 * config/avr/avr.md (setmemhi): Make sure match_dup
22047 operand number comes before match_scratch.
22049 2017-03-08 Richard Biener <rguenther@suse.de>
22051 PR tree-optimization/79920
22052 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22053 with ncopies == 1 to ...
22054 (vect_transform_slp_perm_load): ... here. Properly compute
22055 all element loads by iterating VF times over the group. Do
22056 not handle ncopies (computed in a broken way) in
22057 vect_create_mask_and_perm.
22059 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22062 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22063 is a uniform vector, use uniform_vector_p return value instead of
22064 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22066 2017-03-07 Marek Polacek <polacek@redhat.com>
22068 PR middle-end/79809
22069 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22070 (alloca_call_type): Likewise.
22072 2017-03-07 Martin Liska <mliska@suse.cz>
22074 * gcov.c (process_args): Put comment to correct location.
22076 2017-03-07 Martin Liska <mliska@suse.cz>
22078 PR middle-end/68270
22079 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22080 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22081 (chkp_narrow_bounds_for_field): Likewise.
22082 (chkp_parse_array_and_component_ref): Pass one more argument to
22085 2017-03-07 Richard Biener <rguenther@suse.de>
22087 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
22090 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22092 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
22093 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
22095 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22098 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
22099 to end of description.
22100 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22102 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22104 PR rtl-optimization/79901
22105 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22107 (*avx512f_<code><mode>3<mask_name>): ... this.
22108 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
22109 iterator instead of VI8_AVX2_AVX512BW.
22111 PR rtl-optimization/79901
22112 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
22113 min/max expander, expand it using expand_vec_cond_expr.
22116 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
22119 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22122 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
22123 to void * for PCH reasons.
22124 * dwarf2out.c (output_loc_operands, output_die): Cast
22125 v.val_vec.array to unsigned char *.
22127 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22130 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22133 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22135 PR rtl-optimization/79571
22136 * lra-constraints.c (process_alt_operands): Calculate static
22137 reject and subtract it from overall when only addresses will be
22140 2017-03-06 Julia Koval <julia.koval@intel.com>
22143 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22144 incoming stack boundary to 128 for 64-bit targets.
22146 2017-03-06 Richard Biener <rguenther@suse.de>
22148 PR tree-optimization/79894
22149 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22150 to NULL after folding it.
22152 2017-03-06 Richard Biener <rguenther@suse.de>
22154 PR tree-optimization/79824
22155 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22156 check disabling peeling for gaps.
22158 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22160 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22161 attributes): Document gettimeofday.
22163 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22165 * config/s390/s390.c (s390_option_override_internal): Set
22166 PARAM_MIN_VECT_LOOP_BOUND
22168 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22170 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22171 * config/s390/s390.md: Likewise.
22173 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22176 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22177 (<avx2_avx512>_perm<mode>): Rename to ...
22178 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22180 (<avx512>_perm<mode>_mask): Rename to ...
22181 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22183 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22184 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22185 instead of VI8F_256_512.
22186 (avx512f_perm<mode>): New define_expand.
22187 (avx512f_perm<mode>_mask): Likewise.
22188 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22189 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22191 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22193 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22194 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22196 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22198 2017-03-06 Martin Liska <mliska@suse.cz>
22201 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22202 when having a SSA NAME w/o VAR_DECL assigned to it.
22204 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22206 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22207 msa_dpsub_<su>_d): Fix MODE for vec_select.
22209 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22211 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22213 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22215 2017-03-06 Richard Biener <rguenther@suse.de>
22217 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22218 * plugin.c (register_plugin_info): Likewise.
22219 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22221 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22223 * config/i386/sse.md (sse_storehps, sse_storelps,
22224 avx_<castmode><avxsizesuffix>_<castmode>,
22225 avx512f_<castmode><avxsizesuffix>_<castmode>,
22226 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22227 in condition that at least one operand is not a MEM.
22229 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22231 PR middle-end/79805
22232 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22233 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22235 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22236 gimple_call_nothrow_p flag based on whether original builtin can throw.
22237 If it can, emit following stmts on the fallthrough edge.
22238 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22239 don't create new bb if inserting just debug stmts on the edge, try to
22240 insert them on the fallthru bb or just reset debug stmts.
22242 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22245 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22246 restore recog_data (including the operand rtxes inside it) around
22247 the call to get_insn_template.
22249 2017-03-03 Martin Sebor <msebor@redhat.com>
22251 PR tree-optimization/79699
22252 * context.c (context::~context): Free MPFR caches to avoid
22253 a memory leak on program exit.
22255 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22257 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22258 Use wide_int::ulow () instead of .elt (0).
22260 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22262 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22263 (*pushxf): Limit oF constraint to 32bit targets and add oC
22264 constraint for 64bit targets.
22265 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22266 (*pushdf): Change rmF constraint to rmC.
22268 2017-03-03 Martin Liska <mliska@suse.cz>
22270 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22271 Remove unused variable.
22273 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22276 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22277 is a memory operand, increase num_memory.
22278 (ix86_expand_args_builtin): Likewise.
22280 2017-03-03 Jan Hubicka <jh@suse.cz>
22283 * ipa-devirt.c (maybe_record_node): Properly handle
22284 __cxa_pure_virtual visibility.
22286 2017-03-03 Martin Liska <mliska@suse.cz>
22288 PR tree-optimization/79803
22289 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22291 (pass_loop_prefetch::execute): Disabled optimization if an
22292 assumption about L1 cache size is not met.
22294 2017-03-03 Martin Liska <mliska@suse.cz>
22296 PR rtl-optimization/79574
22297 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22298 (hash_scan_set): Likewise.
22299 (dump_hash_table): Likewise.
22300 (hoist_code): Likewise.
22302 2017-03-03 Richard Biener <rguenther@suse.de>
22304 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22305 in place of uhigh ().
22306 (fixed_convert_from_real): Likewise.
22308 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22311 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22313 2017-03-03 Richard Biener <rguenther@suse.de>
22315 PR middle-end/79818
22316 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22317 TYPE_OVERFLOW_UNDEFINED check.
22319 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22321 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22323 (vector_ae_<mode>_p): Likewise.
22324 (vector_nez_<mode>_p): Likewise.
22325 (vector_ne_v2di_p): Likewise.
22326 (vector_ae_v2di_p): Likewise.
22327 (vector_ne_<mode>_p): Likewise.
22328 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22330 (vsx_tsqrt<mode>2_fe): Likewise.
22332 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22335 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22337 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22339 PR rtl-optimization/79780
22340 * cprop.c (one_cprop_pass): When second and further conditional trap
22341 in a single basic block is turned into an unconditional trap, turn it
22342 into a deleted note to avoid RTL verification failures.
22344 2017-03-02 Richard Biener <rguenther@suse.de>
22346 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22348 2017-03-02 Richard Biener <rguenther@suse.de>
22350 PR tree-optimization/79345
22352 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22353 param and abort the walk, returning -1 if it is hit.
22354 (walk_aliased_vdefs): Take a limit param and pass it on.
22355 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22356 defaulting to 0 and return a signed int.
22357 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22358 (check_defs): New helper.
22359 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22360 about uninitialized memory.
22361 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22362 bogus uninitialized warning.
22363 (fixed_convert_from_real): Likewise.
22365 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22367 PR tree-optimization/66768
22368 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22369 iv_use if base object can't be determined.
22371 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22373 PR tree-optimization/79345
22374 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22375 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22376 (get_pattern_stats): Initialize it.
22377 * genemit.c (gen_expand): Verify match_scratch numbers come after
22378 match_operand/match_dup numbers.
22379 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22380 match_scratch numbers.
22381 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22383 * config/s390/s390.md (trunctdsd2): Likewise.
22385 2017-03-02 Richard Biener <rguenther@suse.de>
22387 * wide-int.h (wide_int_storage::operator=): Implement in terms
22390 2017-03-02 Richard Biener <rguenther@suse.de>
22392 PR tree-optimization/79777
22393 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22394 the to insert expression to sth existing.
22396 2017-03-01 Martin Sebor <msebor@redhat.com>
22398 PR middle-end/79692
22399 * gimple-ssa-sprintf.c
22400 (directive::known_width_and_precision): New function.
22401 (format_integer): Use it.
22402 (get_mpfr_format_length): Consider the full range of precision
22403 when computing %g output with the # flag. Set the likely byte
22404 count to 3 rather than 1 when precision is indeterminate.
22405 (format_floating): Correct the lower bound of precision.
22407 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22409 * doc/invoke.texi: Document default code model for 64-bit Linux.
22411 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22414 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22415 udiv rather than div since input pattern is unsigned.
22417 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22419 * config/i386/i386.c (print_reg): Warn for values of
22420 unsupported size in integer register.
22422 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22425 * config/rs6000/predicates.md (current_file_function_operand): Do
22426 not allow self calls to be local if the function is replaceable.
22428 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22431 * config/rs6000/altivec.h (vec_ctz and others): Change the
22432 preprocessor macro that controls conditional compilation from
22433 _ARCH_PWR9 to __POWER9_VECTOR__.
22434 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22435 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22436 control (instead of _ARCH_PWR9 control) so that template
22437 definition uses power9-specific function.
22438 (vec_any_eq): Likewise.
22439 (vec_all_ne): Change macro definition to use a power9-specific
22440 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22441 _ARCH_PWR9 control).
22442 (vec_any_eq) Likewise.
22443 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22444 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22445 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22446 support for xvcmpnedp instruction.
22447 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22448 macro expansion so that Power9 implementation of vec_all_ne does
22449 not use the AltiVec predicate framework.
22450 (VCMPNEH_P): Likewise.
22451 (VCMPNEW_P): Likewise.
22452 (VCMPNED_P): Likewise.
22453 (VCMPNEFP_P): Likewise.
22454 (VCMPNEDP_P): Likewise.
22455 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22456 implementation of vec_any_eq to not use AltiVec predicate
22458 (VCMPAEH_P): Likewise.
22459 (VCMPAEW_P): Likewise.
22460 (VCMPAED_P): Likewise.
22461 (VCMPAEFP_P): Likewise.
22462 (VCMPAEDP_P): Likewise.
22463 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22464 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22465 not use the AltiVec predicate framework.
22466 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22467 of vec_any_eq to not use AltiVec predicate framework.
22468 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22469 support for predefined __POWER9_VECTOR__ macro to indicate that
22470 Power9 instruction selection is enabled.
22471 (altivec_overloaded_builtins): Remove extraneous
22472 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22473 function argument types RS6000_BTI_bool_V16QI and
22474 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22475 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22476 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22477 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22478 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22479 Power9 for implementations of vec_cmpne. Change the signature for
22480 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22481 (representing vec_all_ne) to remove the previously described first
22482 argument of type RS6000_BTI_INTSI, as this was an artifact of
22483 reliance on the AltiVec predicate framework, which is no longer
22484 used in the implementation of these functions. Add
22485 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22486 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22487 since, unlike the AltiVec predicate framework implementation, we
22488 do not share function descriptors between vec_alle and vec_anyeq.
22489 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22490 set of modes that receive special treatment even when
22491 TARGET_P9_VECTOR is true. The special treatment emits code that
22492 does not depend on Power9 instructions.
22493 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22494 define_expand to not rely on AltiVec predicate framework.
22495 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22497 (vector_ne_v2di_p): Change this define_expand to not rely on
22498 AltiVec predicate framework.
22499 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22501 (vector_ne_<mode>_p): Change this define_expand to not rely on
22502 AltiVec predicate framework.
22503 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22505 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22506 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22507 define_insn pattern.
22508 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22509 define_insn pattern because the xvcmpne<VSs>. instruction is not
22511 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22512 instruction is not supported.
22514 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22516 * config/nvptx/nvptx.c: Include intl.h.
22518 2017-03-01 Martin Jambor <mjambor@suse.cz>
22521 * ipa-prop.h (ipa_bits): Removed field known.
22522 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22523 to pointers. Adjusted their comments to warn about their sharing.
22524 (ipcp_transformation_summary): Change bits to a vector of pointers.
22525 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22526 (ipa_get_ipa_bits_for_value): Declare.
22527 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22528 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22529 (ipa_bits_hash_table): Likewise.
22530 (ipa_vr_ggc_hash_traits): Likewise.
22531 (ipa_vr_hash_table): Likewise.
22532 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22533 being pointers and vr_known being removed.
22534 (ipa_set_jf_unknown): Likewise.
22535 (ipa_get_ipa_bits_for_value): New function.
22536 (ipa_set_jfunc_bits): Likewise.
22537 (ipa_get_value_range): New overloaded functions.
22538 (ipa_set_jfunc_vr): Likewise.
22539 (ipa_compute_jump_functions_for_edge): Use the above functions to
22540 construct bits and vr parts of jump functions.
22541 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22542 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22544 (ipcp_grow_transformations_if_necessary): Also allocate
22545 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22547 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22548 them. Fix too long lines.
22549 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22550 vr_known being removed.
22551 (ipa_read_jump_function): Use new setter functions to construct bits
22552 and vr parts of jump functions or set them to NULL.
22553 (write_ipcp_transformation_info): Adjust for bits being pointers.
22554 (read_ipcp_transformation_info): Likewise.
22555 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22557 Include gt-ipa-prop.h.
22558 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22560 (ipcp_store_bits_results): Likewise.
22561 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22562 Do not write to existing jump functions but use a temporary instead.
22564 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22567 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22568 attempt to use its first operand as BIT_FIELD_REF base.
22570 2017-03-01 Richard Biener <rguenther@suse.de>
22572 PR middle-end/79721
22573 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22574 interpolating formula in wrapping arithmetic.
22575 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22577 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22579 PR tree-optimization/79734
22580 * tree-vect-generic.c (expand_vector_condition): Optimize
22581 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22582 Handle VEC_COND_EXPR where comparison has different inner width from
22583 type's inner width.
22585 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22587 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22588 markup, and similar issues. Remove @opindex entries for things
22589 that aren't options. Add missing -mmpy-option entries.
22591 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22593 PR tree-optimization/79737
22594 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22595 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22596 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22597 instead of byte_size. Formatting fix.
22598 (shift_bytes_in_array_right): Formatting fix.
22600 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22603 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22604 condition on optimize for the leaf function test.
22606 2017-02-28 Martin Liska <mliska@suse.cz>
22609 * read-rtl-function.c (function_reader::handle_unknown_directive):
22610 Bail out when one uses -flto.
22612 2017-02-28 Martin Liska <mliska@suse.cz>
22614 * common.opt: Replace space with tabular for options of <number>
22616 * config/i386/i386.opt: Show <number> value for
22617 -mlarge-data-threshold.
22618 * opts.c (print_filtered_help): Do not display number in hexadecimal
22621 2017-02-28 Martin Liska <mliska@suse.cz>
22623 * common.opt: Fix --help=option -Q for options which are of
22626 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22628 * config/i386/i386.c (print_reg): Error out for values
22629 of 8-bit size in invalid integer register.
22631 2017-02-28 Martin Sebor <msebor@redhat.com>
22633 PR tree-optimization/79691
22634 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22636 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22639 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22640 gcc_unreachable with output_operand_lossage.
22642 2017-02-28 Richard Biener <rguenther@suse.de>
22644 PR tree-optimization/79740
22645 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22647 (visit_nary_op): Insert the nary into the hashtable if we
22648 pattern-matched sth.
22649 * tree-ssa-pre.c (eliminate_insert): Robustify.
22651 2017-02-28 Richard Biener <rguenther@suse.de>
22653 PR middle-end/79731
22654 * fold-const.c (decode_field_reference): Reject out-of-bound
22657 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22659 * config/i386/i386.c: Include intl.h.
22660 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22661 instead of just cond ? "..." : "...".
22662 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22663 * coverage.c (read_counts_file): Likewise.
22664 * omp-offload.c: Include intl.h.
22665 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22666 of just cond ? "..." : "...".
22667 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22668 of just cond ? "..." : "...".
22670 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22673 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22675 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22676 'tune for' CPU name.
22677 * config/arm/arm-cpu-data.h: Regenerated.
22679 2017-02-28 Richard Biener <rguenther@suse.de>
22681 PR tree-optimization/79732
22682 * tree-inline.c (expand_call_inline): Do not shadow var.
22684 2017-02-28 Richard Biener <rguenther@suse.de>
22686 PR tree-optimization/79723
22687 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22688 address-space properly.
22690 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22692 * doc/optinfo.texi (Optimization groups): Fix option used for
22694 * doc/invoke.texi (-fopt-info): Document "omp".
22695 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22696 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22697 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22698 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22699 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22701 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22703 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22704 OPTGROUP_OMP with "omp".
22706 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22709 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22710 for arithmetic shift of unsigned V2DI.
22712 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22714 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22715 arc/linux.h headers.
22716 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22717 (LINK_SPEC): Likewise.
22718 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22719 (EXTRA_SPECS): Likewise.
22720 (STARTFILE_SPEC): Likewise.
22721 (ENDFILE_SPEC): Likewise.
22722 (LIB_SPEC): Likewise.
22723 (TARGET_SDATA_DEFAULT): Likewise.
22724 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22725 (MULTILIB_DEFAULTS): Likewise.
22726 (DWARF2_UNWIND_INFO): Likewise.
22727 * config/arc/big.h: New file.
22728 * config/arc/elf.h: Likewise.
22729 * config/arc/linux.h: Likewise.
22730 * config/arc/t-uClibc: Remove.
22732 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22734 PR tree-optimization/77536
22735 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22736 (tree_transform_and_unroll_loop): Use above function to compute the
22737 estimated niter of unrolled loop and use it when scaling profile.
22738 Also use count info rather than frequency if it's non-zero.
22739 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22740 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22741 (vect_transform_loop): Call above function.
22743 2017-02-27 Richard Biener <rguenther@suse.de>
22745 PR tree-optimization/45397
22746 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22747 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22748 (visit_nary_op): Add pattern matching for CSEing sign-changed
22749 or truncated operations with wider ones.
22751 2017-02-27 Richard Biener <rguenther@suse.de>
22753 PR tree-optimization/79690
22754 * tree-vect-stmts.c (vectorizable_store): Use vector type
22755 built from the DR with address-space.
22757 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22759 * doc/invoke.texi (Optimize Options): Refine the description
22760 of asan-use-after-return.
22762 2017-02-25 Alan Modra <amodra@gmail.com>
22764 PR rtl-optimization/79584
22765 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22766 base, not ad->base_term, the reg within base. Remove assertion
22767 that ad->base == ad->base_term. Replace gen_int_mode using
22768 bogus mode with const0_rtx.
22770 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22772 PR middle-end/79396
22773 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22774 FMA_EXPR like tcc_binary or tcc_unary.
22776 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22779 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22781 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22782 (output_loc_operands): Handle DW_OP_call_ref and
22783 DW_OP_GNU_variable_value.
22784 (struct variable_value_struct): New type.
22785 (struct variable_value_hasher): Likewise.
22786 (variable_value_hash): New variable.
22787 (string_types): Remove.
22788 (copy_loc_descr): New function.
22789 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22790 (prepend_loc_descr_to_each): New function.
22791 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22792 instead of add_loc_descr_to_each if the first argument is single
22793 location list and the second has multiple.
22794 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22795 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22796 when looking for variable value which doesn't have other location info.
22797 (loc_list_from_tree): Formatting fix.
22798 (gen_array_type_die): Simplify DW_AT_string_length handling.
22799 (adjust_string_types): Remove.
22800 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22801 string_types. Call resolve_variable_values.
22802 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22803 (resolve_addr_in_expr): Likewise. Add A argument.
22804 (copy_deref_exprloc): Remove deref argument. Adjust for the
22805 original expression being DW_OP_GNU_variable_value with optionally
22806 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22807 optionally after it.
22808 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22809 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22810 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22811 (variable_value_hasher::hash, variable_value_hasher::equal): New
22813 (resolve_variable_value_in_expr, resolve_variable_value,
22814 resolve_variable_values, note_variable_value_in_expr,
22815 note_variable_value): New functions.
22816 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22819 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22822 * opts.h (handle_generated_option): Add GENERATED_P argument.
22823 * opts-common.c (handle_option): Adjust function comment.
22824 (handle_generated_option): Add GENERATED_P argument, pass it to
22826 (control_warning_option): Pass false to handle_generated_option
22828 * opts.c (maybe_default_option): Pass true to handle_generated_option
22830 * optc-gen.awk: Likewise.
22832 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22834 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22835 a REG, look at the REG it is a SUBREG of.
22836 (splitter for cmpeqsi_t): Ditto.
22838 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22840 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22841 the special USEs with the pattern of the insn, not the insn itself.
22843 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22846 * doc/invoke.texi: Document -mload-store-pairs.
22848 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22849 Sandra Loosemore <sandra@codesourcery.com>
22851 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22852 argument isn't a CONST_INT.
22853 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22854 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22855 (nios2_validate_compare): Bypass alternate compare logic if *op2
22856 is not a CONST_INT.
22857 (ldstwm_operation_p): Return false if first_base is not a REG or
22858 if first_offset is not a CONST_INT.
22860 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22862 * config/cris/cris.md: Use correct operand in a define_peephole2.
22864 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22866 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22868 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22870 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22871 this_insn if it is an INSN or JUMP_INSN.
22872 (force_offsettable): Look at base, not at addr.
22873 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22874 on things that aren't necessarily CONST_INTs.
22876 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22878 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22879 -mfpmath=sse is the default also for x86-32 targets with SSE2
22880 instruction set when @option{-ffast-math} is enabled
22882 2017-02-24 Jeff Law <law@redhat.com>
22884 PR rtl-optimizatoin/79286
22885 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22888 2017-02-24 Richard Biener <rguenther@suse.de>
22890 PR tree-optimization/79389
22891 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22894 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22896 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22897 function comment to reflect reality.
22898 (loop_exits_before_overflow): Fix typo in function description.
22900 2017-02-24 Richard Biener <rguenther@suse.de>
22902 PR tree-optimization/79389
22903 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22904 properly that a threading opportunity exists. Detect conditional
22905 copy/constant propagation opportunities.
22907 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22909 * config/visium/visium.md (type): Add trap.
22910 (b): New mode attribute.
22911 (*btst): Rename into...
22912 (*btst<mode>): ...this and adjust.
22913 (*cbranchsi4_btst_insn): Rename into...
22914 (*cbranch<mode>4_btst_insn): ...this and adjust.
22915 (trap): New define_insn.
22917 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22919 PR tree-optimization/79389
22920 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22921 (noce_reversed_cond_code): New function.
22922 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22923 reversed_comparison_code. Formatting fix.
22924 (noce_try_store_flag): Test rev_cond != NULL in addition to
22925 reversed_comparison_code.
22926 (noce_try_store_flag_constants): Likewise.
22927 (noce_try_store_flag_mask): Likewise.
22928 (noce_try_addcc): Use rev_cond if non-NULL instead of
22929 reversed_comparison_code.
22930 (noce_try_cmove_arith): Likewise. Formatting fixes.
22931 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22932 (noce_find_if_block): Initialize rev_cond.
22933 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22934 instead of false as last argument never attempt to reverse it
22937 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22939 PR tree-optimization/79663
22940 * tree-predcom.c (combine_chains): Process refs in reverse order
22941 only for ZERO length chains, and add explaining comment.
22943 2017-02-23 Jeff Law <law@redhat.com>
22945 PR tree-optimization/79578
22946 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22947 in call to operand_equal_p.
22949 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22952 * config/i386/cpuid.h: Fix another undefined behavior.
22954 2017-02-23 Richard Biener <rguenther@suse.de>
22956 PR tree-optimization/79683
22957 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22958 vector types for data-refs.
22960 2017-02-23 Martin Liska <mliska@suse.cz>
22962 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22964 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22966 PR middle-end/79665
22967 * internal-fn.c (get_range_pos_neg): Moved to ...
22968 * tree.c (get_range_pos_neg): ... here. No longer static.
22969 * tree.h (get_range_pos_neg): New prototype.
22970 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22971 are known to be in between 0 and signed maximum inclusive, try to
22972 expand both unsigned and signed divmod and use the cheaper one from
22975 2017-02-22 Jeff Law <law@redhat.com>
22977 PR tree-optimization/79578
22978 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22979 to compare base operands.
22981 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22984 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22985 gpc_reg_operand instead of fpr_reg_operand.
22987 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22989 * config/mips/mips.c (mips_return_in_memory): Force FP
22990 vector types to be returned in memory for o32 ABI.
22992 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22994 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22995 instead of DW_TAG_member for static data member declarations and don't
22996 set no_linkage_name for static inline data members.
22997 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23000 2017-02-22 Martin Liska <mliska@suse.cz>
23002 * doc/invoke.texi: Replace inequality signs with square brackets
23005 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23007 PR tree-optimization/68644
23008 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23010 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23013 * lra-constraints.c (simplify_operand_subreg): Handle
23014 WORD_REGISTER_OPERATIONS targets.
23016 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23019 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23020 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23021 elimination by swapping fld*.
23023 2017-02-22 Richard Biener <rguenther@suse.de>
23025 PR tree-optimization/79673
23026 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23027 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23028 irrelevant address-space qualifiers and avoiding a
23029 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23031 2017-02-22 Richard Biener <rguenther@suse.de>
23033 PR tree-optimization/79666
23034 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23035 to not symbolically negate if that may introduce undefined
23038 2017-02-22 Martin Liska <mliska@suse.cz>
23041 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23042 * data-streamer-out.c (streamer_write_gcov_count_stream):
23044 * value-prof.c (stream_out_histogram_value): Make assert more
23045 precise based on type of counter.
23047 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23050 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23051 Use nonimmediate_operand instead of memory_operand for operand 1.
23052 (float-extend standard_x87sse_constant_load splitter): Ditto.
23054 2017-02-21 Jeff Law <law@redhat.com>
23056 PR tree-optimization/79621
23057 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23058 blocks with edges to themselves.
23060 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23063 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23064 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23065 Use gimple_call_builtin_p.
23068 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23069 on temporarily removed DEBUG_INSNs.
23071 PR tree-optimization/79649
23072 * tree-loop-distribution.c (classify_partition): Give up on
23073 non-generic address space loads/stores.
23075 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23077 * doc/loop.texi (Loop manipulation): Remove nonexistent
23078 tree_ssa_loop_version from the documentation.
23079 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23081 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23084 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
23085 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
23086 * config/rs6000/rs6000.c: Include except.h.
23087 (rs6000_expand_split_stack_prologue): Call
23088 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
23090 2017-02-21 Martin Jambor <mjambor@suse.cz>
23093 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
23094 have been analyzed.
23096 2017-02-21 Martin Jambor <mjambor@suse.cz>
23098 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
23099 for backward compatibility only.
23100 * doc/invoke.texi (Option Summary): Remove all references to
23101 -fipa-cp-alignment.
23103 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23107 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23109 * lra-constraints.c (curr_insn_transform): Handle
23110 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23112 2017-02-21 Martin Liska <mliska@suse.cz>
23114 * config/i386/i386.opt: Replace -masm-dialect with -masm.
23116 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
23118 PR translation/79638
23119 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
23121 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
23124 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
23125 (arm_function_ok_for_sibcall): Return false for an indirect call by
23126 descriptor if all the argument registers are used.
23127 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
23128 alignment of the function.
23130 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23132 PR tree-optimization/61441
23133 * simplify-rtx.c (simplify_const_unary_operation): For
23134 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23135 the sNaN unmodified.
23137 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23139 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23140 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23141 instead of SYSTEM_HEADER_DIR.
23143 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23144 Martin Liška <mliska@suse.cz>
23146 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23147 Fix typos and grammar, use active voice, and clarify.
23149 2017-02-20 Marek Polacek <polacek@redhat.com>
23151 PR middle-end/79537
23152 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23155 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23157 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23160 * config/i386/i386.c (ix86_expand_builtin): Handle
23161 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23162 ix86_builtins_isa[fcode].isa as a requirement of those
23163 flags and any other flag in the bitmask.
23164 (ix86_init_mmx_sse_builtins): Use 0 instead of
23165 ~OPTION_MASK_ISA_64BIT as mask.
23166 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23167 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23168 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23169 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23171 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23174 * lra-constraints.c (split_reg): Check requested split mode
23175 is supported by the register.
23177 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23179 * lra-constraints.c (simplify_operand_subreg): Remove early
23182 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23185 * lra-constraints.c (curr_insn_transform): Tighten condition
23186 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23188 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23191 * lra-constraints.c (curr_insn_transform): Handle
23192 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23194 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23197 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23199 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23201 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23204 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23207 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23209 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23212 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23214 * final.c (last_columnnum, override_columnnum): New variables.
23215 (final_start_function): Set last_columnnum, pass it to begin_prologue
23216 hook and pass 0 to dwarf2out_begin_prologue.
23217 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23218 to source_line debug hook.
23219 (notice_source_line): Compute last_columnnum and for debug_column_info
23220 return true on column changes.
23221 * debug.h (struct gcc_debug_hooks): Add column argument to
23222 source_line and begin_prologue hooks.
23223 (debug_nothing_int_charstar_int_bool): Remove prototype.
23224 (debug_nothing_int_int_charstar,
23225 debug_nothing_int_int_charstar_int_bool): New prototypes.
23226 (dwarf2out_begin_prologue): Add column argument.
23227 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23228 begin_prologue hooks.
23229 (debug_nothing_int_charstar_int_bool): Remove.
23230 (debug_nothing_int_int_charstar,
23231 debug_nothing_int_int_charstar_int_bool): New functions.
23232 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23233 through to dwarf2out_source_line.
23234 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23235 (dwarf2out_source_line): Add column argument, emit it if requested.
23236 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23238 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23239 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23240 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23241 through to dwarf2out_begin_prologue.
23242 (vmsdbgout_source_line): Add column argument, pass it through to
23243 dwarf2out_source_line.
23244 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23245 dbxout_source_line caller.
23246 (dbxout_source_line): Add column argument.
23248 * common.opt (gno-column-info, gcolumn-info): New options.
23249 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23250 (check_die): Also test for multiple DW_AT_decl_column attributes.
23251 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23252 DW_AT_decl_column if requested.
23253 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23255 (gen_variable_die): Likewise.
23256 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23257 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23260 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23261 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23262 (ix86_handle_option): Handle OPT_m3dnowa.
23263 * doc/invoke.texi (-m3dnowa): Document.
23264 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23265 -m3dnowa instead of -m3dnow -march=athlon.
23268 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23269 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23271 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23274 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23275 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23276 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23277 generator for vsx_xxpermdi_<mode>_be.
23278 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23279 force big-endian semantics.
23280 (vsx_xxpermdi_<mode>_be): New define_expand with same
23281 implementation as previous version of vsx_xxpermdi_<mode>.
23283 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23285 PR tree-optimization/79327
23286 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23287 variable, its initialization and use.
23289 2017-02-17 Julia Koval <julia.koval@intel.com>
23291 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23292 (OPTION_MASK_ISA_PKU_UNSET): New.
23293 (ix86_handle_option): Handle -mrdpid.
23294 * config/i386/cpuid.h (bit_RDPID): New.
23295 * config/i386/driver-i386.c (host_detect_local_cpu):
23296 Detect RDPID feature.
23297 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23298 * config/i386/i386-c.c (ix86_target_macros_internal):
23300 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23301 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23302 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23303 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23304 * config/i386/i386.md (define_insn "rdpid"): New.
23305 * config/i386/i386.opt Add -mrdpid.
23306 * config/i386/immintrin.h (_rdpid_u32): New.
23308 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23310 PR rtl-optimization/79541
23311 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23312 instead of transforming it into USE.
23314 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23316 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23317 If HONOR_SNANS (SFmode) force the input to a register.
23318 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23319 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23320 an frsp or similar insn.
23322 2017-02-17 Martin Liska <mliska@suse.cz>
23324 PR rtl-optimization/79577
23325 * params.def (selsched-max-sched-times): Increase minimum to 1.
23327 2017-02-17 Martin Liska <mliska@suse.cz>
23329 PR rtl-optimization/79574
23330 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23332 2017-02-17 Martin Liska <mliska@suse.cz>
23334 PR tree-optimization/79529
23335 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23336 ssa_defined_default_def_p to handle cases which are implicitly
23338 * tree-ssa.c (ssa_defined_default_def_p): New function.
23339 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23340 which are implicitly defined.
23341 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23343 2017-02-17 Richard Biener <rguenther@suse.de>
23345 PR middle-end/79576
23346 * params.def (max-ssa-name-query-depth): Limit to 10.
23348 2017-02-17 Richard Biener <rguenther@suse.de>
23350 PR tree-optimization/79552
23351 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23354 2017-02-17 Richard Biener <rguenther@suse.de>
23357 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23359 2017-02-17 Marek Polacek <polacek@redhat.com>
23361 PR middle-end/79536
23362 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23363 (fold_negate_expr): New wrapper.
23365 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23367 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23368 Correct terminology and de-emphasize pre-standard behavior.
23370 2017-02-16 Alan Modra <amodra@gmail.com>
23372 PR rtl-optimization/79286
23373 * ira.c (def_dominates_uses): New function.
23374 (update_equiv_regs): Don't create an equivalence for insns that
23375 may trap where the register def does not dominate the use.
23377 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23379 PR rtl-optimization/78127
23380 * lra.c (lra): Call lra_eliminate before finish the loop after
23383 2017-02-16 Richard Biener <rguenther@suse.de>
23385 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23387 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23388 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23389 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23390 (isl_val_int_from_wi): New function.
23391 (extract_affine_gmp): Rename to ...
23392 (extract_affine_wi): ... this, take a widest_int.
23393 (extract_affine_int): Just wrap extract_affine_wi.
23394 (add_param_constraints): Use isl_val_int_from_wi.
23395 (add_loop_constraints): Likewise, and extract_affine_wi.
23397 2017-02-15 Jeff Law <law@redhat.com>
23399 PR middle-end/79521
23400 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23401 ira_init_register_move_cost_if_necessary.
23403 2017-02-15 Martin Sebor <msebor@redhat.com>
23405 PR middle-end/32003
23406 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23407 removed in a prior commit.
23409 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23411 PR tree-optimization/79347
23412 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23413 counters during peeling.
23415 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23417 * Makefile.in (site.exp): Remove "set ISLVER".
23419 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23422 * real.c (real_from_integer): Call real_convert even for decimal.
23424 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23427 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23429 2017-02-14 Andrew Pinski <apinski@cavium.com>
23431 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23432 cores and change the partno/implementer to be correct.
23433 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23434 the 'B" as the implementer.
23435 * config/aarch64/aarch64-tune.md: Regenerate.
23437 2017-02-14 Carl Love <cel@us.ibm.com>
23439 * config/rs6000/rs6000.c: Add case statement entry to make the
23440 xvcvuxdsp built-in argument unsigned.
23441 * config/rs6000/vsx.md: Fix the source and return operand types so they
23442 match the instruction definitions from the ISA document. Fix typo
23443 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23446 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23449 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23450 member early_clobber_alts.
23451 * lra-lives.c (reg_early_clobber_p): New.
23452 (process_bb_lives): Use it.
23453 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23454 (debug_operand_data): Initialize early_clobber_alts.
23455 (setup_operand_alternative): Set up early_clobber_alts.
23456 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23457 alternatives to new_insn_reg.
23458 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23460 (lra_update_insn_regno_info): Pass the new arg.
23462 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23464 PR middle-end/79505
23465 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23466 (new_oacc_loop_raw): Don't clear already cleared fields.
23469 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23470 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23471 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23473 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23476 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23478 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23481 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23482 the extra instruction to the right place to store 128-bit constant
23485 2017-02-14 Martin Sebor <msebor@redhat.com>
23487 PR middle-end/79448
23488 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23489 warning for strings of unknown length.
23491 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23493 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23495 2017-02-14 Jeff Law <law@redhat.com>
23498 * ira-costs.c (scan_one_insn): Initialize register move costs
23499 for pseudos seen in USE/CLOBBER insns.
23501 PR tree-optimization/79095
23502 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23503 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23504 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23505 if the operands are known to be not equal, then the resulting range
23507 (intersect_ranges): If the new range is ~[0,0] and the old range is
23508 wide, then prefer ~[0,0].
23509 * tree-vrp.c (overflow_comparison_p_1): New function.
23510 (overflow_comparison_p): New function.
23511 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23512 if NAME is used in an overflow test.
23513 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23514 overflow check that can be expressed as an equality test, then adjust
23515 ops to be that equality test.
23517 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23519 * config/s390/s390-builtin-types.def: Remove flags argument.
23520 * config/s390/s390.c (s390_init_builtins): Likewise.
23522 2017-02-14 Martin Liska <mliska@suse.cz>
23524 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23525 vector. Fix trailing white spaces.
23527 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23529 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23532 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23534 PR rtl-optimization/68664
23535 * config/arm/arm.c (arm_sched_can_speculate_insn):
23536 New function. Declare prototype.
23537 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23539 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23541 PR rtl-optimization/68664
23542 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23544 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23546 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23548 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23549 max skip bytes for function, loop and jump.
23551 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23553 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23554 ABS_EXPR for gimple dump.
23556 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23559 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23561 PR tree-optimization/79408
23562 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23563 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23564 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23565 also if rhs1 is INTEGER_CST.
23567 2017-02-14 Richard Biener <rguenther@suse.de>
23569 PR middle-end/79432
23570 * tree-into-ssa.c (insert_phi_nodes): When the function can
23571 have abnormal edges rewrite SSA names with broken use-def
23572 dominance out of SSA and register them for PHI insertion.
23574 2017-02-13 Martin Sebor <msebor@redhat.com>
23576 PR middle-end/79496
23577 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23578 clearing info.nowrite flag when snprintf size argument is a range.
23580 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23582 * cprop.c (cprop_jump): Add missing space in string literal.
23583 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23584 (get_constraint_for_component_ref): Likewise.
23585 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23586 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23587 * lra-constraints.c (process_alt_operands): Likewise.
23588 * ipa-inline.c (inline_small_functions): Likewise.
23589 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23590 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23591 * trans-mem.c (diagnose_tm_1_op): Likewise.
23592 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23593 (grid_parallel_clauses_gridifiable): Likewise.
23595 * config/nvptx/mkoffload.c (process): Add space in between
23598 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23599 "MOD4_SSE_REGS" and "ALL_REGS".
23601 * spellcheck.c (test_data): Add , in between "foo" and "food".
23603 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23606 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23607 boundary crossing check and subsequent code generation agree.
23609 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23611 * config/aarch64/aarch64.c (has_memory_op): Delete.
23612 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23615 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23617 PR rtl-optimization/79388
23618 PR rtl-optimization/79450
23619 * combine.c (distribute_notes): When removing TEM_INSN for which
23620 corresponding dest has last value recorded, invalidate that last
23623 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23625 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23626 of explicit '@'. Add missing assembly comment marker on branch costs
23629 2017-02-13 Nathan Sidwell <nathan@acm.org>
23631 * gengtype-lex.l (<in_struct>): Add '/'.
23633 2017-02-13 Martin Liska <mliska@suse.cz>
23636 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23638 2017-02-13 Richard Biener <rguenther@suse.de>
23640 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23642 * configure: Re-generate.
23643 * config.in: Likewise.
23644 * graphite-dependences.c: Simplify as if
23645 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23646 * graphite-isl-ast-to-gimple.c: Likewise.
23647 * graphite-optimize-isl.c: Likewise.
23648 * graphite-poly.c: Likewise.
23649 * graphite-sese-to-poly.c: Likewise.
23650 * graphite.h: Likewise.
23651 * toplev.c: Include isl/version.h and use isl_version () for
23652 printing the ISL version.
23653 * doc/install.texi: Update ISL requirement.
23655 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23657 * doc/standards.texi (Standards): Update reference to
23660 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23662 * doc/extend.texi (Named Address Spaces): sourceware.org now
23664 * doc/install.texi (Binaries): Ditto.
23667 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23669 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23670 standard terminology "stringize"/"stringizing" throughout.
23671 * doc/cppinternals.texi: Likewise.
23673 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23675 * doc/extend.texi: Fix some spelling mistakes and typos.
23676 * doc/invoke.texi: Likewise.
23678 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23681 * params.def (inline-min-speedup) Change from 10 to 8.
23683 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23685 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23688 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23691 * ipa-inline-analysis.c (get_minimal_bb): New function.
23692 (record_modified): Use it.
23693 (remap_edge_change_prob): Handle also ancestor functions.
23695 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23697 * doc/contrib.texi (Contributors): Remove broken link into
23698 the Mauve CVS repository.
23700 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23702 PR middle-end/79454
23703 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23704 result computation whenever lhs doesn't have vector mode, not
23705 just when it has BLKmode.
23707 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23709 * doc/makefile.texi (profiledbootstrap): Refer to the
23710 installation instructions only in textual form.
23712 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23715 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23717 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23719 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23720 (Specific): Update mingw-w64 reference.
23722 (Specific): Remove broken link to Renesas RX processor.
23724 2017-02-10 Richard Biener <rguenther@suse.de>
23726 * toplev.c (process_options): Do not mention obsolete graphite
23727 options when printing sorry message about missing graphite support.
23728 Mention -floop-nest-optimize.
23730 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23732 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23733 (vtst_p16): Likewise.
23734 (vtstq_p8): Likewise.
23735 (vtstq_p16): Likewise.
23737 (vtstq_p64): Likewise.
23738 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23739 (vset_lane_p64): New.
23740 (vsetq_lane_p64): New.
23742 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23744 PR tree-optimization/79411
23745 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23746 stmt operands are SSA_NAMEs used in abnormal phis.
23747 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23750 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23753 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23756 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23758 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23760 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23762 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23766 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23767 not arbitrary TREE_CONSTANT.
23770 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23771 "omp declare target link" attribute unless is_global_var.
23772 * omp-offload.c (find_link_var_op): Likewise.
23774 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23775 Chung-Lin Tang <cltang@codesourcery.com>
23777 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23779 (gimplify_adjust_omp_clauses): Don't delete TILE.
23780 (gimplify_omp_for): Deal with TILE.
23781 * internal-fn.c (expand_GOACC_TILE): New function.
23782 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23784 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23785 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23787 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23788 avoid DIV for outermost collapse var.
23789 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23790 Remove out of date comments, fix whitespace.
23791 * omp-general.c (omp_extract_for_data): Deal with tiling.
23792 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23793 adjust OLF_DIM_BASE value.
23794 (struct omp_for_data): Add tiling field.
23795 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23796 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23797 for auto loops. Remove default auto determining, moved to
23798 oacc_loop_fixed_partitions.
23799 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23800 stmts, add e_mask field.
23801 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23802 (oacc_thread_numbers): Use oacc_dim_call.
23803 (oacc_xform_tile): New.
23804 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23805 (finish_oacc_loop): Adjust for ifns vector.
23806 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23807 add case for GOACC_TILE fns.
23808 (oacc_loop_xform_loop): Delete.
23809 (oacc_loop_process): Iterate over call list directly, and add
23810 handling for GOACC_TILE fns.
23811 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23813 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23814 vector partitioning to outer loops. Assign 2 partitions to loops
23815 when available. Add TILE handling.
23816 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23817 (execite_oacc_device_lower): Process GOACC_TILE fns,
23818 ignore unknown specs.
23819 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23820 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23821 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23823 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23825 * configure.ac (ACX_BUGURL): Update.
23826 * configure: Regenerate.
23828 2017-02-09 Richard Biener <rguenther@suse.de>
23830 PR tree-optimization/69823
23831 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23832 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23834 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23836 * config/arc/arc-c.def: Add __NPS400__ definition.
23837 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23838 (TARGET_NPS400): Define.
23840 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23842 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23844 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23845 pointer, arch_info.
23846 (arc_cpu_types): Fill the arch_info field with a pointer into the
23847 arc_arch_types table.
23848 (arc_selected_cpu): Declare.
23849 * config/arc/arc.c (arc_selected_cpu): Make global.
23850 (arc_selected_arch): Delete.
23851 (arc_base_cpu): Delete.
23852 (arc_override_options): Remove references to deleted variables,
23853 update access to arch information.
23854 (ARC_OPT): Update access to arch information.
23855 (ARC_OPTX): Likewise.
23856 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23857 (TARGET_ARC600): Update access to arch information.
23858 (TARGET_ARC601): Likewise.
23859 (TARGET_ARC700): Likewise.
23860 (TARGET_EM): Likewise.
23861 (TARGET_HS): Likewise.
23862 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23865 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23868 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23869 condition/operands for integer GE/LE/GEU/LEU operations.
23871 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23873 PR translation/79397
23874 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23877 2017-02-08 Martin Jambor <mjambor@suse.cz>
23880 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23881 whether allocation happened.
23882 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23883 nothing was allocated.
23885 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23887 PR tree-optimization/79408
23888 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23889 constant, but SSA_NAME with a known integer range, use the minimum
23890 of that range instead of op1 to determine if modulo can be replaced
23891 with its first operand.
23893 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23895 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23897 2017-02-08 Richard Biener <rguenther@suse.de>
23899 PR tree-optimization/71824
23900 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23901 Check all loops contained in the merged region.
23903 2017-02-07 Andrew Pinski <apinski@cavium.com>
23905 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23907 2017-02-07 Andrew Pinski <apinski@cavium.com>
23909 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23910 (thunderxt88): Likewise.
23911 (thunderxt81): Disable LSE and change v8.1 to v8.
23912 (thunderxt83): Likewise.
23914 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23915 Richard Biener <rguenther@suse.de>
23917 PR middle-end/79399
23918 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23919 type from int to size_t.
23920 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23922 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23924 PR rtl-optimization/79386
23925 * cprop.c (bypass_conditional_jumps): Initialize
23926 bypass_last_basic_block already before splitting bbs after
23927 unconditional traps...
23928 (bypass_conditional_jumps): ... rather than here.
23931 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23932 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23933 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23934 fix -masm=intel patterns.
23936 2017-02-07 Richard Biener <rguenther@suse.de>
23938 PR tree-optimization/79256
23939 PR middle-end/79278
23940 * builtins.c (get_object_alignment_2): Use min_align_of_type
23941 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23942 and ADJUST_FIELD_ALIGN.
23944 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23946 * doc/tm.texi: Regenerate.
23947 * stor-layout.c (layout_decl): Adjust.
23948 (update_alignment_for_field): Likewise.
23949 (place_field): Likewise.
23950 (min_align_of_type): Likewise.
23951 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23952 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23953 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23954 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23955 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23956 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23957 * config/i386/i386.c (x86_field_alignment): Likewise.
23958 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23959 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23960 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23961 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23962 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23963 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23967 2017-01-30 Richard Biener <rguenther@suse.de>
23969 PR tree-optimization/79256
23970 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23971 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23974 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23976 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23977 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23978 builtins to SImode and emit a zero-extend, if necessary.
23980 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23982 * docs/invoke.texi (RISC-V Options): Alphabetize.
23984 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23986 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23989 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23991 * config/riscv/riscv.c: New file.
23992 * common/config/riscv/riscv-common.c: Likewise.
23993 * config.gcc: Likewise.
23994 * config/riscv/constraints.md: Likewise.
23995 * config/riscv/elf.h: Likewise.
23996 * config/riscv/generic.md: Likewise.
23997 * config/riscv/linux.h: Likewise.
23998 * config/riscv/multilib-generator: Likewise.
23999 * config/riscv/peephole.md: Likewise.
24000 * config/riscv/pic.md: Likewise.
24001 * config/riscv/predicates.md: Likewise.
24002 * config/riscv/riscv-builtins.c: Likewise.
24003 * config/riscv/riscv-c.c: Likewise.
24004 * config/riscv/riscv-ftypes.def: Likewise.
24005 * config/riscv/riscv-modes.def: Likewise.
24006 * config/riscv/riscv-opts.h: Likewise.
24007 * config/riscv/riscv-protos.h: Likewise.
24008 * config/riscv/riscv.h: Likewise.
24009 * config/riscv/riscv.md: Likewise.
24010 * config/riscv/riscv.opt: Likewise.
24011 * config/riscv/sync.md: Likewise.
24012 * config/riscv/t-elf-multilib: Likewise.
24013 * config/riscv/t-linux: Likewise.
24014 * config/riscv/t-linux-multilib: Likewise.
24015 * config/riscv/t-riscv: Likewise.
24016 * configure.ac: Likewise.
24017 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24018 Waterman as RISC-V maintainers.
24019 * doc/install.texi: Add RISC-V entries.
24020 * doc/invoke.texi: Add RISC-V options section.
24021 * doc/md.texi: Add RISC-V constraints section.
24022 * configure: Regenerated.
24024 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24027 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24028 false values to be constant vectors with all 0 or all 1 bits set.
24029 (vcondu<mode><mode>): Likewise.
24030 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24032 (fpmask_comparison_operator): Update comment.
24033 (vecint_comparison_operator): New predicate.
24034 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24035 vector conditionals when the true and false values are constant
24036 vectors with all 0 bits or all 1 bits set.
24038 2017-02-06 Martin Sebor <msebor@redhat.com>
24040 PR tree-optimization/79376
24041 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24043 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24045 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24046 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24047 to simplify split condition.
24049 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24051 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24052 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24055 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24057 PR rtl-optimization/68664
24058 * target.def (can_speculate_insn): New hook.
24059 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24060 * doc/tm.texi: Regenerate.
24061 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24062 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24063 (rs6000_sched_can_speculate_insn): New function.
24065 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24067 PR tree-optimization/79284
24068 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24069 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24070 vectorizable_mask_load_store, vectorizable_operation,
24071 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24072 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24073 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24074 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24075 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24076 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24077 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24078 is_gimple_assign (stmt). Replace another such test with
24079 is_gimple_assign (stmt).
24081 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24084 * config/avr/avr.c (rtl-iter.h): Include it.
24085 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
24086 (avr_legitimate_combined_insn): ...and implementation.
24088 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
24090 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
24091 * config/s390/s390.c (s390_const_operand_ok)
24092 (s390_canonicalize_comparison, s390_extract_part)
24093 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
24094 (s390_contiguous_bitmask_p, s390_rtx_costs)
24095 (legitimize_pic_address): Likewise.
24096 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
24097 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
24098 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
24099 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
24100 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24102 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24104 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24105 REGNO($0) == REGNO($1).
24107 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24109 * config/s390/linux.h(SIZE_TYPE): Add comment.
24111 2017-02-06 Julian Brown <julian@codesourcery.com>
24112 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24113 Virendra Pathak <virendra.pathak@broadcom.com>
24115 * config/aarch64/aarch64-cores.def: Change the scheduler
24117 * config/aarch64/aarch64.md: Include thunderx2t99.md.
24118 * config/aarch64/thunderx2t99.md: New file.
24120 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24122 * doc/standards.texi (Go Language): Update link to language
24125 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24127 * tree-eh.c (lower_resx): Sanitize profile.
24128 (cleanup_empty_eh_move_lp): Likewise.
24130 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24133 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24135 * cfgloopmanip.h (loop_version): Update prototype.
24136 * modulo-sched.c (sms_schedule): Update call of loop_version.
24137 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24138 * tree-parloops.c (gen_parallel_loop): Likewise.
24139 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24140 * tree-ssa-loop-split.c (split_loop): Likewise.
24141 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24142 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24144 2017-02-05 Martin Liska <mliska@suse.cz>
24147 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24149 (print_operand_address): Initialize a struct to zero.
24151 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24153 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24154 garbage collector only in textual form.
24156 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24158 * doc/extend.texi (x86 specific memory model extensions for
24159 transactional memory): Simplify a phrase.
24161 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24164 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24165 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24166 (atomic_storedi_1): Likewise.
24168 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24170 PR tree-optimization/79338
24171 * tree-parloops.c (gather_scalar_reductions): Don't call
24172 vect_analyze_loop_form for loop->inner before destroying loop's
24175 2017-02-03 Martin Sebor <msebor@redhat.com>
24177 PR tree-optimization/79327
24178 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24179 when precision has resulted in leading zeros.
24180 (format_integer): Adjust the likely counter to assume an unknown
24181 argument that may be zero is non-zero.
24183 2017-02-03 Jason Merrill <jason@redhat.com>
24186 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24187 avoid copying non-taken branch.
24189 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24191 PR tree-optimization/79340
24192 * tree-vect-loop.c (vectorizable_reduction): Release
24193 vec_defs elements after safe_splicing them into other vectors.
24196 PR tree-optimization/79327
24197 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24198 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24200 (format_integer): Use wide_int_to_tree instead of build_int_cst
24201 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24202 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24203 of shortest and longest sequence.
24205 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24207 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24208 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24210 2017-02-03 Walter Lee <walt@tilera.com>
24213 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24214 after initial stackframe link reg save.
24215 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24217 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24220 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24221 wu for stxssp alternative.
24223 2017-02-03 Martin Sebor <msebor@redhat.com>
24225 PR tree-optimization/79352
24226 * gimple-fold.c (get_range_strlen): Add argument.
24227 (get_range_strlen): Change return type to bool.
24228 (get_maxval_strlen): Pass in a dummy argument.
24229 * gimple-fold.h (get_range_strlen): Change return type to bool.
24230 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24231 * tree.h (array_at_struct_end_p): Add argument.
24232 * tree.c (array_at_struct_end_p): Handle it.
24234 2017-02-03 Martin Liska <mliska@suse.cz>
24237 * multiple_target.c (create_dispatcher_calls): Redirect edge
24238 from a caller of a dispatcher.
24239 (expand_target_clones): Make the clones local.
24240 (ipa_target_clone): Do both target clones and resolvers.
24241 (ipa_dispatcher_calls): Remove the pass.
24242 (pass_dispatcher_calls::gate): Likewise.
24243 (make_pass_dispatcher_calls): Likewise.
24244 * passes.def (pass_target_clone): Put as very first IPA early
24247 2017-02-03 Martin Liska <mliska@suse.cz>
24249 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24250 in case of a function with ifunc attribute.
24252 2017-02-03 Martin Liska <mliska@suse.cz>
24254 * cgraph.c (cgraph_node::dump): Dump function version info.
24255 * symtab.c (symtab_node::dump_base): Add missing new line.
24257 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24259 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24260 (ifcombine_ifandif): Use it.
24262 2017-02-03 Martin Liska <mliska@suse.cz>
24264 * doc/invoke.texi: Document default value for
24265 use-after-scope-direct-emission-threshold.
24267 2017-02-03 Martin Liska <mliska@suse.cz>
24269 PR tree-optimization/79339
24270 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24271 (format_floating): Likewise.
24273 2017-02-03 Martin Liska <mliska@suse.cz>
24276 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24278 (ipa_node_params_t::remove): Likewise.
24279 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24280 initialization from removed ipa_node_params_t::insert.
24281 (ipa_node_params::~ipa_node_params): Move from removed
24282 ipa_node_params_t::release.
24283 * symbol-summary.h (symbol_summary::m_released): New member.
24284 Do not release a summary twice. Do not allow to call finalizer
24285 for types of a summary that live in GGC memory.
24287 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24289 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24292 2017-02-02 Martin Sebor <msebor@redhat.com>
24294 PR middle-end/79275
24295 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24296 (format_string): Tighten up the range of output for non-constant
24297 strings and correct the expected range for wide non-constant strings.
24299 2017-02-02 Martin Sebor <msebor@redhat.com>
24301 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24303 PR middle-end/32003
24304 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24306 (-fdump-tree-@var): Add to index and document how to come up
24307 with pass-specific option and dump file names.
24308 (-fdump-passes): Clarify where to look for output.
24310 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24312 PR middle-end/77445
24313 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24314 statistics of the analyzed path; allow threading for speed when
24315 any of BBs along the path are optimized for speed.
24317 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24319 PR middle-end/78468
24320 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24321 settings of the virtual registers.
24324 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24326 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24327 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24330 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24332 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24333 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24335 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24337 * config/s390/s390.md: Add missing comments with the expanded
24339 * config/s390/vector.md: Likewise.
24340 * config/s390/vx-builtins.md: Likewise.
24342 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24345 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24346 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24347 conditions on a single line.
24349 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24351 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24352 __S390_VX__ to __VX__.
24354 2017-02-01 Andrew Pinski <apinski@cavium.com>
24356 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24357 stmt_info to record_stmt_cost.
24358 (vect_get_known_peeling_cost): Pass stmt_info if known to
24360 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24361 cpu_vector_cost field into
24362 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24363 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24364 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24365 splitting of scalar_stmt_cost and vec_stmt_cost.
24366 (thunderx_vector_cost): Likewise.
24367 (cortexa57_vector_cost): LIkewise.
24368 (exynosm1_vector_cost): Likewise.
24369 (xgene1_vector_cost): Likewise.
24370 (thunderx2t99_vector_cost): Improve after the splitting of the two
24372 (aarch64_builtin_vectorization_cost): Update for the splitting of
24373 scalar_stmt_cost and vec_stmt_cost.
24375 2017-02-01 Torvald Riegel <triegel@redhat.com>
24376 Richard Henderson <rth@redhat.com>
24378 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24379 conditional on existance of a fast atomic load.
24380 * optabs-query.c (can_atomic_load_p): New function.
24381 * optabs-query.h (can_atomic_load_p): Declare it.
24382 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24383 no fast atomic load is available for the particular size of access.
24384 (expand_atomic_compare_and_swap): Likewise.
24385 (expand_atomic_load): Likewise.
24386 (expand_atomic_store): Likewise.
24387 (expand_atomic_fetch_op): Likewise.
24388 * testsuite/lib/target-supports.exp
24389 (check_effective_target_sync_int_128): Remove x86 because it provides
24390 no fast atomic load.
24391 (check_effective_target_sync_int_128_runtime): Likewise.
24393 2017-02-01 Richard Biener <rguenther@suse.de>
24395 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24396 (graphite_transform_loops): Provide opt-info for optimized nests.
24397 * tree-parloop.c (parallelize_loops): Provide opt-info for
24398 parallelized loops.
24400 2017-02-01 Richard Biener <rguenther@suse.de>
24402 PR middle-end/79315
24403 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24404 was not set before.
24406 2017-02-01 Richard Biener <rguenther@suse.de>
24408 PR tree-optimization/71824
24409 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24410 Verify the loops are valid in the merged SESE region.
24411 (scop_detection::can_represent_loop_1): Check analyzing the
24412 evolution of the number of iterations in the region succeeds.
24414 2017-01-31 Ian Lance Taylor <iant@golang.org>
24416 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24417 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24419 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24421 PR preprocessor/79210
24422 * input.c (get_substring_ranges_for_loc): Replace line_width
24423 assertion with error-handling.
24425 2017-01-31 Richard Biener <rguenther@suse.de>
24427 PR tree-optimization/77318
24428 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24429 (create_pw_aff_from_tree): Take loop parameter.
24430 (add_condition_to_pbb): Pass loop of the condition to
24431 create_pw_aff_from_tree.
24433 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24435 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24436 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24438 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24442 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24444 (convert_int_to_float128): Likewise.
24445 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24446 (convert_int_to_float128): Likewise.
24447 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24448 (UNSPEC_IEEE128_CONVERT): Likewise.
24449 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24450 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24451 Use local variables for IBM extended format.
24452 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24453 (fix_trunc<mode>si2_fprs): Likewise.
24454 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24455 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24456 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24457 to know that we can now have integers of all sizes in vector
24459 (fix<uns>_<mode>di2_hw): Likewise.
24460 (float<uns>_<mode>si2_hw): Likewise.
24461 (fix_<mode>si2_hw): Likewise.
24462 (fixuns_<mode>si2_hw): Likewise.
24463 (float<uns>_<mode>di2_hw): Likewise.
24464 (float_<mode>di2_hw): Likewise.
24465 (float_<mode>si2_hw): Likewise.
24466 (floatuns_<mode>di2_hw): Likewise.
24467 (floatuns_<mode>si2_hw): Likewise.
24468 (xscvqp<su>wz_<mode>): Delete, no longer used.
24469 (xscvqp<su>dz_<mode>): Likewise.
24470 (xscv<su>dqp_<mode>): Likewise.
24471 (ieee128_mfvsrd_64bit): Likewise.
24472 (ieee128_mfvsrd_32bit): Likewise.
24473 (ieee128_mfvsrwz): Likewise.
24474 (ieee128_mtvsrw): Likewise.
24475 (ieee128_mtvsrd_64bit): Likewise.
24476 (ieee128_mtvsrd_32bit): Likewise.
24478 2017-01-31 Martin Liska <mliska@suse.cz>
24481 * ipa-prop.c (ipa_free_all_node_params): Call release method
24482 instead of ~sumbol_summary to not to trigger double times
24485 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24487 PR tree-optimization/71691
24488 * bitmap.h (class auto_bitmap): New.
24489 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24490 is_maybe_undefined instead of ssa_undefined_value_p.
24492 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24494 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24495 __S390_ARCH_LEVEL__ to __ARCH__.
24497 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24499 PR tree-optimization/79267
24500 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24501 if should_remove_lhs_p is true.
24503 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24506 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24507 (add_alignment_attribute): New.
24508 (base_type_die): Add alignment attribute.
24509 (subrange_type_die): Likewise.
24510 (modified_type_die): Likewise.
24511 (gen_array_type_die): Likewise.
24512 (gen_descr_array_type_die: Likewise.
24513 (gen_enumeration_type_die): Likewise.
24514 (gen_subprogram_die): Likewise.
24515 (gen_variable_die): Likewise.
24516 (gen_field_die): Likewise.
24517 (gen_ptr_to_mbr_type_die): Likewise.
24518 (gen_struct_or_union_type_die): Likewise.
24519 (gen_subroutine_type_die): Likewise.
24520 (gen_typedef_die): Likewise.
24521 (base_type_cmp): Compare alignment attribute.
24523 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24526 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24527 (setb_unsigned) New pattern for setb with CCUNS.
24528 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24529 subfc./subfe sequence to avoid overflow problems. Generate a
24530 shorter sequence with cmpld/setb for power9.
24531 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24532 for generating subfc. instruction.
24533 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24534 now uses this instruction.
24536 2017-01-30 Ian Lance Taylor <iant@google.com>
24539 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24540 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24542 2017-01-30 Martin Sebor <msebor@redhat.com>
24544 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24545 Move constant to the right of a relational operator.
24546 (get_mpfr_format_length, format_character, format_string): Ditto.
24547 (should_warn_p, maybe_warn): Same.
24549 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24551 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24554 * asan.c (get_translation_unit_decl): Remove function.
24555 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24557 2017-01-30 Martin Liska <mliska@suse.cz>
24559 PR gcov-profile/79259
24560 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24561 -fprofile-generate.
24563 2017-01-30 Martin Liska <mliska@suse.cz>
24566 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24567 Initialize variables with NULL value.
24569 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24572 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24574 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24576 2017-01-30 Richard Biener <rguenther@suse.de>
24578 PR tree-optimization/79276
24579 * tree-vrp.c (process_assert_insertions): Properly adjust common
24580 when removing a duplicate.
24582 * gcc.dg/torture/pr79276.c: New testcase.
24584 2017-01-30 Richard Biener <rguenther@suse.de>
24586 PR tree-optimization/79256
24587 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24588 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24590 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24592 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24595 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24596 ("*r<noxa>sbg_<mode>_sll_bitmask")
24597 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24598 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24599 Use contiguous_bitmask_nowrap_operand.
24601 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24604 * config/rs6000/altivec.h (vec_xl): Revise #define.
24605 (vec_xst): Likewise.
24607 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24609 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24611 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24613 PR rtl-optimization/79194
24614 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24615 traps before call to bypass_conditional_jumps.
24617 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24619 PR tree-optimization/71374
24620 * lra-constraints.c (check_conflict_input_operands): New.
24621 (match_reload): Use it.
24623 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24626 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24627 account to calculate conflict_set.
24629 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24631 PR rtl-optimization/78559
24632 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24633 other_insn in combine.
24635 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24637 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24638 uint16_type_node for BT_UINT16.
24640 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24642 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24643 "RTL Tests" to menu.
24644 (GIMPLE Tests): New node.
24645 (RTL Tests): New node.
24647 2017-01-27 Richard Biener <rguenther@suse.de>
24649 PR tree-optimization/79245
24650 * tree-loop-distribution.c (distribute_loop): Apply cost
24651 modeling also to detected patterns.
24653 2017-01-27 Richard Biener <rguenther@suse.de>
24655 PR tree-optimization/71433
24656 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24657 (compare_assert_loc): New function.
24658 (process_assert_insertions): Sort and optimize assert locations
24659 to remove duplicates and push down identical assertions on
24660 edges to their destination block.
24662 2017-01-27 Richard Biener <rguenther@suse.de>
24664 PR tree-optimization/79244
24665 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24666 out SSA names even if abnormal.
24668 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24670 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24671 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24672 instead of MPFR_RNDN.
24674 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24677 * arm.c (arm_option_override): Don't call build_target_option_node
24678 until after doing all option overrides.
24679 (arm_valid_target_attribute_tree): Likewise.
24681 2017-01-27 Martin Liska <mliska@suse.cz>
24683 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24684 for {cd}tors and C++ {cd}tors.
24686 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24688 * config/s390/s390.md ("*setmem_long_and")
24689 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24691 2017-01-26 Martin Sebor <msebor@redhat.com>
24693 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24696 2017-01-26 Martin Sebor <msebor@redhat.com>
24698 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24699 HAVE_DFmode before using XFmode or DFmode.
24700 (parse_directive): Avoid using the z length modifier to avoid
24701 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24703 PR middle-end/78703
24704 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24705 to accept adjustment as an array.
24706 (get_int_range): New function.
24707 (struct directive): Make width and prec arrays.
24708 (directive::set_width, directive::set_precision): Call get_int_range.
24709 (format_integer, format_floating): Handle width and precision ranges.
24710 (format_string, parse_directive): Same.
24712 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24715 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24716 comdat_type_p set, just clone them, but keep the children in the
24720 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24721 which have direct callers with -fvar-tracking-assignments enabled
24723 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24724 inside of type units.
24726 2017-01-26 Martin Sebor <msebor@redhat.com>
24728 PR middle-end/78703
24729 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24731 (struct format_result): Replace number_chars, number_chars_min,
24732 and number_chars_max with a single member of struct result_range.
24734 (format_result::operator+=): Adjust.
24735 (struct fmtresult): Remove bounded. Handle likely and unlikely
24737 (fmtresult::adjust_for_width_or_precision): New function.
24738 (fmtresult:type_max_digits): New function.
24739 (bytes_remaining): Handle likely and unlikely counters.
24740 (min_bytes_remaining): Remove.
24741 (format_percent): Simplify.
24742 (format_integer, format_floating): Set likely and unlikely counters.
24743 (get_string_length, format_character, format_string): Same.
24744 (format_plain, should_warn_p): New function.
24745 (maybe_warn): Call should_warn_p. Update diagnostic messages
24746 and handle those for all directives, including plain strings.
24747 (format_directive): Handle likely and unlikely counters.
24748 Remove unnecessary quoting from diagnostics. Add an informational
24750 (add_bytes): Remove.
24751 (pass_sprintf_length::compute_format_length): Simplify.
24752 (try_substitute_return_value): Handle likely and unlikely counters.
24754 2017-01-26 Carl Love <cel@us.ibm.com>
24756 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24757 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24759 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24762 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24763 endianess for subregs into account.
24764 * lra-constraints.c (lra_constraints): Do risky transformations
24765 always on the first iteration.
24766 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24767 last_call_used_reg_set.
24768 (process_bb_lives): Define and use last_call_used_reg_set.
24769 * lra.c (lra): Always continue after lra_constraints on the first
24772 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24774 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24776 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24778 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24780 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24781 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24782 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24783 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24784 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24785 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24786 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24787 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24788 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24790 2017-01-26 Marek Polacek <polacek@redhat.com>
24793 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24794 for the third operand.
24796 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24798 PR middle-end/79236
24799 * omp-low.c (struct omp_context): Add simt_stmt field.
24800 (scan_omp_for): Return omp_context *.
24801 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24802 context to the _simt_ SIMD stmt.
24803 (lower_omp_for): For combined SIMD with sibling _simt_
24804 SIMD, make sure to use the same decls in _looptemp_
24805 clauses as in the sibling.
24807 2017-01-26 David Sherwood <david.sherwood@arm.com>
24809 PR middle-end/79212
24810 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24813 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24816 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24817 emit fld b; fld a; if possible.
24819 * brig-builtins.def: Update copyright years.
24820 * config/arm/arm_acle_builtins.def: Update copyright years.
24822 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24825 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24826 constraint instead of o for the stxsd instruction.
24828 2017-01-25 Carl Love <cel@us.ibm.com>
24830 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24831 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24833 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24835 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24837 2017-01-25 Richard Biener <rguenther@suse.de>
24839 PR tree-optimization/69264
24840 * target.def (vector_alignment_reachable): Improve documentation.
24841 * doc/tm.texi: Regenerate.
24842 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24844 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24845 earlier changes with respect to TYPE_USER_ALIGN.
24846 (vector_alignment_reachable_p): Likewise. Improve dumping.
24848 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24851 * config/arm/arm.md (xordi3): Force constant operand into a register
24854 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24856 * doc/invoke.texi (-fstore-merging): Correct default optimization
24857 levels at which it is enabled.
24858 (-O): Move -fstore-merging from list to...
24861 2017-01-25 Richard Biener <rguenther@suse.de>
24864 * omp-expand.c: Include debug.h.
24865 (expand_omp_taskreg): Make sure to generate early debug before
24866 outlining anything from a function.
24867 (expand_omp_target): Likewise.
24868 (grid_expand_target_grid_body): Likewise.
24870 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24873 * asan.c (get_translation_unit_decl): New function.
24874 (asan_add_global): Extract modules file name from globals
24875 TRANSLATION_UNIT_DECL name.
24877 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24880 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24881 for long calls with APCS frame and VFP.
24883 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24885 * cfg.c (original_copy_tables_initialized_p): New function.
24886 * cfg.h (original_copy_tables_initialized_p): New decl.
24887 * cfgrtl.c (relink_block_chain): Guard the call to
24888 free_original_copy_tables with a call to
24889 original_copy_tables_initialized_p.
24890 * cgraph.h (symtab_node::native_rtl_p): New decl.
24891 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24892 (symtab_node::needed_p): Don't assert for early assembly output
24893 for __RTL functions.
24894 (cgraph_node::finalize_function): Set "force_output" for __RTL
24896 (cgraph_node::analyze): Bail out early for __RTL functions.
24897 (analyze_functions): Update assertion to support __RTL functions.
24898 (cgraph_node::expand): Bail out early for __RTL functions.
24899 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24901 * function.h (struct function): Update comment for field
24903 * gimple-expr.c: Include "tree-pass.h".
24904 (gimple_has_body_p): Return false for __RTL functions.
24905 * Makefile.in (OBJS): Add run-rtl-passes.o.
24906 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24908 (gcc::pass_manager::get_clean_slate): New accessor.
24909 * passes.c: Include "insn-addr.h".
24910 (should_skip_pass_p): Add logging. Update logic for running
24911 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24912 property-provider override so it is only done for gimple passes.
24914 (skip_pass): New function.
24915 (execute_one_pass): Call skip_pass when skipping passes.
24916 * read-md.c (md_reader::read_char): Support filtering
24917 the input to a subset of line numbers.
24918 (md_reader::md_reader): Initialize fields
24919 m_first_line and m_last_line.
24920 (md_reader::read_file_fragment): New function.
24921 * read-md.h (md_reader::read_file_fragment): New decl.
24922 (md_reader::m_first_line): New field.
24923 (md_reader::m_last_line): New field.
24924 * read-rtl-function.c (function_reader::create_function): Only
24925 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24926 curr_properties. Set DECL_INITIAL to a dummy block.
24927 (read_rtl_function_body_from_file_range): New function.
24928 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24930 * run-rtl-passes.c: New file.
24931 * run-rtl-passes.h: New file.
24933 2017-01-24 Jeff Law <law@redhat.com>
24935 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24938 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24940 PR tree-optimization/79159
24941 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24942 (record_nonwrapping_iv): Improve boundary using above function if no
24943 value range information.
24945 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24946 Martin Jambor <mjambor@suse.cz>
24948 * brig-builtins.def: New file.
24949 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24950 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24951 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24952 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24953 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24954 * builtin-types.def (BT_INT8): New.
24955 (BT_INT16): Likewise.
24956 (BT_UINT8): Likewise.
24957 (BT_UINT16): Likewise.
24958 (BT_FN_ULONG): Likewise.
24959 (BT_FN_UINT_INT): Likewise.
24960 (BT_FN_UINT_ULONG): Likewise.
24961 (BT_FN_UINT_LONG): Likewise.
24962 (BT_FN_UINT_PTR): Likewise.
24963 (BT_FN_ULONG_PTR): Likewise.
24964 (BT_FN_INT8_FLOAT): Likewise.
24965 (BT_FN_INT16_FLOAT): Likewise.
24966 (BT_FN_UINT32_FLOAT): Likewise.
24967 (BT_FN_UINT16_FLOAT): Likewise.
24968 (BT_FN_UINT8_FLOAT): Likewise.
24969 (BT_FN_UINT64_FLOAT): Likewise.
24970 (BT_FN_UINT16_UINT32): Likewise.
24971 (BT_FN_UINT32_UINT16): Likewise.
24972 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24973 (BT_FN_INT_PTR_INT): Likewise.
24974 (BT_FN_UINT_PTR_UINT): Likewise.
24975 (BT_FN_LONG_PTR_LONG): Likewise.
24976 (BT_FN_ULONG_PTR_ULONG): Likewise.
24977 (BT_FN_VOID_UINT64_UINT64): Likewise.
24978 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24979 (BT_FN_INT8_INT8_INT8): Likewise.
24980 (BT_FN_INT16_INT16_INT16): Likewise.
24981 (BT_FN_INT_INT_INT): Likewise.
24982 (BT_FN_UINT_FLOAT_UINT): Likewise.
24983 (BT_FN_FLOAT_UINT_UINT): Likewise.
24984 (BT_FN_ULONG_UINT_UINT): Likewise.
24985 (BT_FN_ULONG_UINT_PTR): Likewise.
24986 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24987 (BT_FN_UINT_UINT_UINT): Likewise.
24988 (BT_FN_VOID_UINT_PTR): Likewise.
24989 (BT_FN_UINT_UINT_PTR: Likewise.
24990 (BT_FN_UINT32_UINT64_PTR): Likewise.
24991 (BT_FN_INT_INT_UINT_UINT): Likewise.
24992 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24993 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24994 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24995 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24996 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24997 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24998 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24999 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25000 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25001 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25002 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25003 * doc/frontends.texi: List BRIG FE.
25004 * doc/install.texi (Testing): Add BRIG tesring requirements.
25005 * doc/invoke.texi (Overall Options): Mention BRIG.
25006 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25008 2017-01-24 Richard Biener <rguenther@suse.de>
25010 PR translation/79208
25011 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25013 2017-01-24 Martin Jambor <mjambor@suse.cz>
25016 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25017 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25018 and known_contexts.
25020 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25022 PR middle-end/79123
25023 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25024 casts from signed to unsigned really don't have a range.
25026 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25028 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25029 GMP_RNDx for compatiblity.
25031 2017-01-24 Martin Liska <mliska@suse.cz>
25034 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25035 that would prevent us to call alloca with -1 as argument.
25037 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25039 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25040 Avoid -Wformat-security warning.
25042 2017-01-23 Andrew Pinski <apinski@cavium.com>
25044 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25047 2017-01-23 Martin Sebor <msebor@redhat.com>
25049 PR middle-end/78703
25050 * gimple-ssa-sprintf.c (warn_level): New global.
25051 (format_integer): Use it here and throughout the rest of the file.
25052 Use the same switch to compute sign as base.
25053 (maybe_warn): New function.
25054 (format_directive): Factor out warnings into maybe_warn.
25055 Add debugging output. Use warn_level.
25056 (add_bytes): Use warn_level.
25057 (pass_sprintf_length::compute_format_length): Add debugging output.
25058 (try_substitute_return_value): Same.
25059 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25061 PR middle-end/78703
25062 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25063 (struct fmtresult, format_integer, format_floating): Adjust.
25064 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25065 (get_string_length, format_string,format_directive): Same.
25066 (pass_sprintf_length::compute_format_length): Same.
25067 (try_substitute_return_value): Simplify slightly.
25069 PR middle-end/78703
25070 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25071 (fmtresult::operator+=): Outlined.
25072 (struct fmtresult): Add ctors.
25073 (struct conversion_spec): Rename...
25074 (struct directive): ...to this. Add and remove data members.
25075 (directive::set_width, directive::set_precision): New functions.
25076 (format_percent): Use fmtresult ctor.
25077 (get_width_and_precision): Remove.
25078 (format_integer): Make naming changes. Avoid computing width and
25080 (format_floating): Same. Adjust indentation.
25081 (format_character, format_none): New functions.
25082 (format_string): Moved character handling to format_character.
25083 (format_directive): Remove arguments, change return type.
25084 (parse_directive): New function.
25085 (pass_sprintf_length::compute_format_length): Move directive
25086 parsing to parse_directive.
25088 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25090 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
25091 (assign_assembler_name_if_needed): ... this.
25092 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
25093 (assign_assembler_name_if_needed): ... this.
25094 (free_lang_data_in_cgraph): Adjust callers.
25095 * cgraphunit.c (cgraph_node::analyze): Likewise.
25096 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
25099 2017-01-23 Richard Biener <rguenther@suse.de>
25101 PR tree-optimization/79088
25102 PR tree-optimization/79188
25103 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25104 resetting loop bounds after last path deletion. Reset loop
25105 bounds of the target loop, make code match the comments.
25106 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25107 Make sure loops need no fixups.
25109 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25111 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
25112 exponent support with double type for first argument.
25113 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
25114 type returned by __builtin_vec_extract_sig,
25115 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
25116 functions from "vector int" to "vector unsigned int" or from
25117 "vector long long int" to "vector unsigned long long int".
25118 Changed type returned by __builtin_vec_extract_exp,
25119 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
25120 functions from "vector int" to "vector unsigned int" or from
25121 "vector long long int" to "vector unsigned long long int".
25122 Changed return type of __builtin_vec_test_data_class,
25123 __builtin_vec_test_data_class_sp, and
25124 __builtin_vec_test_data_class_dp from "vector int" to
25125 "vector bool int" or from "vector long long int" to "vector bool
25126 long long int" and changed second argument type from "unsigned
25127 int" to "int". Added new overloaded function forms "vector float
25128 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
25129 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25130 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25131 double, vector unsigned long long int)" and "vector double
25132 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25133 long int)". Changed return type of
25134 __builtin_scalar_test_data_class and
25135 __builtin_scalar_test_data_class_sp and
25136 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25137 int" and changed second argument from "unsigned int" to "int".
25138 Changed type returned by __builtin_scalar_test_neg,
25139 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25140 from "int" to "bool int". Added new overloaded function form
25141 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25142 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25143 exponent double-precision with floating point first argument.
25144 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25145 documentation of scalar_test_data_class, scalar_test_neg,
25146 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25147 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25148 vec_test_data_class built-in functions to reflect refinements in
25149 their type signatures.
25151 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25153 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25155 (aarch64_elf_asm_destructor): Likewise.
25157 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25159 PR rtl-optimization/78634
25160 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25161 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25162 * ifcvt.c (noce_try_cmove): Add missing cost check.
25164 PR rtl-optimization/71724
25165 * combine.c (if_then_else_cond): Look for situations where it is
25166 beneficial to undo the work of one of the recursive calls.
25168 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25170 PR tree-optimization/70754
25171 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25172 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25173 combined stmt before it if not NULL.
25174 (combine_chains): Process refs reversely and compute dominance point
25177 2017-01-23 Martin Liska <mliska@suse.cz>
25179 PR tree-optimization/79196
25180 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25181 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25183 (strlen_optimize_stmt): Call the renamed function.
25185 2017-01-23 Michael Matz <matz@suse.de>
25187 PR tree-optimization/78384
25188 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25190 2017-01-23 Richard Biener <rguenther@suse.de>
25192 PR tree-optimization/79186
25193 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25194 both incoming edges before moving an assert.
25196 2017-01-23 Martin Jambor <mjambor@suse.cz>
25198 * ipa-prop.c (load_from_param_1): Removed.
25199 (load_from_unmodified_param): Bits from load_from_param_1 put back
25201 (load_from_param): Removed.
25202 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25203 with stmt. Reverted back to use of load_from_unmodified_param.
25205 2017-01-23 Martin Jambor <mjambor@suse.cz>
25208 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25209 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25210 field a pointer to garbage collected vector, mark lattices and
25211 ipcp_orig_node with GTY((skip)).
25212 (ipa_get_param_count): Adjust to descriptors being a pointer.
25213 (ipa_get_param): Likewise.
25214 (ipa_get_type): Likewise.
25215 (ipa_get_param_move_cost): Likewise.
25216 (ipa_set_param_used): Likewise.
25217 (ipa_get_controlled_uses): Likewise.
25218 (ipa_set_controlled_uses): Likewise.
25219 (ipa_is_param_used): Likewise.
25220 (ipa_node_params_t): Move into garbage collector. New methods insert
25222 (ipa_node_params_sum): Annotate wth GTY(()).
25223 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25225 (ipa_load_from_parm_agg): Adjust declaration.
25226 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25227 * ipa-profile.c (ipa_profile): Likewise.
25228 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25229 (ipa_populate_param_decls): Make descriptors parameter garbage
25231 (ipa_dump_param): Adjust to descriptors being a pointer.
25232 (ipa_alloc_node_params): Likewise.
25233 (ipa_initialize_node_params): Likewise.
25234 (load_from_param_1): Make descriptors parameter garbage collected.
25235 (load_from_unmodified_param): Likewise.
25236 (load_from_param): Likewise.
25237 (ipa_load_from_parm_agg): Likewise.
25238 (ipa_node_params::~ipa_node_params): Removed.
25239 (ipa_free_all_node_params): Remove call to delete operator.
25240 (ipa_node_params_t::insert): New.
25241 (ipa_node_params_t::remove): Likewise.
25242 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25243 copy known_csts and known_contexts vectors.
25244 (ipa_read_node_info): Adjust to descriptors being a pointer.
25245 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25247 (ipcp_transform_function): Make descriptors variable garbage
25250 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25252 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25253 * config/i386/avx512dqintrin.h: Ditto.
25254 * config/i386/avx512fintrin.h: Ditto.
25255 * config/i386/i386.c: Handle new builtins.
25256 * config/i386/i386-builtin.def: Add new builtins.
25257 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25258 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25260 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25261 Martin Liska <mliska@suse.cz>
25263 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25264 * asan.c (asan_expand_poison_ifn): Support stores and use
25265 appropriate ASAN report function.
25266 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25267 * internal-fn.def (ASAN_POISON_USE): Declare.
25268 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25269 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25270 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25271 ASAN_POISON calls w/o LHS.
25272 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25273 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25274 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25275 * gimplify.c (asan_poison_variables): Add attribute
25276 use_after_scope_memory to variables that really needs to live
25278 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25279 having the attribute.
25281 2017-01-23 Martin Liska <mliska@suse.cz>
25283 * asan.c (create_asan_shadow_var): New function.
25284 (asan_expand_poison_ifn): Likewise.
25285 * asan.h (asan_expand_poison_ifn): New declaration.
25286 * internal-fn.c (expand_ASAN_POISON): Likewise.
25287 * internal-fn.def (ASAN_POISON): New builtin.
25288 * sanopt.c (pass_sanopt::execute): Expand
25289 asan_expand_poison_ifn.
25290 * tree-inline.c (copy_decl_for_dup_finish): Make function
25292 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25293 * tree-ssa.c (is_asan_mark_p): New function.
25294 (execute_update_addresses_taken): Rewrite local variables
25295 (identified just by use-after-scope as addressable) into SSA.
25297 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25299 * doc/install.texi (Specific): opensource.apple.com uses https
25300 now. Remove trailing slash.
25302 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25304 * README.Portability: Remove note on an Irix compatibility issue.
25306 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25308 * gcov.c (INCLUDE_ALGORITHM): Define.
25309 (INCLUDE_VECTOR): Define.
25310 No longer include <vector> and <algorithm> directly.
25312 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25314 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25316 * doc/invoke.texi (Code Gen Options): Ditto.
25318 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25321 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25323 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25325 rtl-optimization/79125
25326 * cprop.c (local_cprop_pass): Handle cases where we make an
25327 unconditional trap.
25329 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25333 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25334 read from, for big endian.
25336 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25338 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25339 register pauth builtins for LP64 only.
25341 2017-01-20 Marek Polacek <polacek@redhat.com>
25344 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25347 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25349 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25351 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25352 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25353 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25355 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25358 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25359 in big-endian mode when they are not a single duplicated value.
25361 2017-01-20 Richard Biener <rguenther@suse.de>
25363 * BASE-VER: Bump to 7.0.1.
25365 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25367 * omp-low.c (omplow_simd_context): New struct. Use it...
25368 (lower_rec_simd_input_clauses): ...here and...
25369 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25370 references to idx, lane, max_vf, is_simt.
25372 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25374 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25377 2017-01-20 Martin Jambor <mjambor@suse.cz>
25379 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25380 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25382 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25383 (GTFILES): Rename hsa.c to hsa-common.c.
25384 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25385 * hsa-dump.c: Likewise.
25386 * hsa-gen.c: Likewise.
25387 * hsa-regalloc.c: Likewise.
25388 * ipa-hsa.c: Likewise.
25389 * omp-expand.c: Likewise.
25390 * omp-low.c: Likewise.
25391 * toplev.c: Likewise.
25393 2017-01-20 Marek Polacek <polacek@redhat.com>
25396 * doc/invoke.texi: Document -Wduplicated-branches.
25397 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25398 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25399 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25400 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25401 return 0 only when not OEP_LEXICOGRAPHIC.
25402 (fold_build_cleanup_point_expr): Use the expression
25403 location when building CLEANUP_POINT_EXPR.
25404 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25405 * tree.c (add_expr): Handle error_mark_node.
25407 2017-01-20 Martin Liska <mliska@suse.cz>
25410 * tree-profile.c (init_ic_make_global_vars): Do not call
25412 (gimple_init_gcov_profiler): Likewise.
25414 2017-01-20 Martin Liska <mliska@suse.cz>
25417 * cgraph.h (maybe_create_reference): Remove argument and
25419 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25421 * ipa-cp.c (create_specialized_node): Likewise.
25422 * symtab.c (symtab_node::maybe_create_reference): Handle
25423 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25425 2017-01-20 Martin Liska <mliska@suse.cz>
25427 * read-rtl-function.c (function_reader::create_function): Use
25428 build_decl instread of build_decl_stat.
25430 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25432 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25433 * config/i386/avx512dqintrin.h: Ditto.
25434 * config/i386/avx512fintrin.h: Ditto.
25435 * config/i386/i386-builtin-types.def: Add new types.
25436 * config/i386/i386.c: Handle new types.
25437 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25438 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25439 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25440 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25441 (__builtin_ia32_kshiftridi): New.
25442 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25444 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25448 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25449 define to rs6000_init_stack_protect_guard.
25450 (rs6000_init_stack_protect_guard): New function.
25452 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25453 Yunqiang Su <yunqiang.su@imgtec.com>
25455 * config.gcc (supported_defaults): Add madd4.
25456 (with_madd4): Add validation.
25457 (all_defaults): Add madd4.
25458 * config/mips/mips.opt (mmadd4): New option.
25459 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25461 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25463 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25464 (ISA_HAS_FUSED_MADD4): Likewise.
25465 * doc/invoke.texi (-mmadd4): Document the new option.
25466 * doc/install.texi (--with-madd4): Document the new option.
25468 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25470 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25471 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25472 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25473 (aarch64_init_pauth_hint_builtins): New.
25474 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25475 (aarch64_expand_builtin): Expand new builtins.
25477 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25479 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25480 * combine-stack-adj.c (no_unhandled_cfa): Handle
25481 REG_CFA_TOGGLE_RA_MANGLE.
25482 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25483 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25484 info for return address signing.
25485 (aarch64_expand_epilogue): Likewise.
25487 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25489 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25490 * config/aarch64/aarch64-protos.h
25491 (aarch64_return_address_signing_enabled): New declaration.
25492 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25494 (aarch64_expand_prologue): Sign return address before it's pushed onto
25496 (aarch64_expand_epilogue): Authenticate return address fetched from
25498 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25499 (aarch64_attributes): New function attributes for "sign-return-address".
25500 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25501 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25502 ("*do_return"): Generate combined instructions according to key index.
25503 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25504 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25506 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25507 * config/aarch64/aarch64.opt (msign-return-address=): New.
25508 * doc/extend.texi (AArch64 Function Attributes): Documents
25509 "sign-return-address=".
25510 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25512 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25514 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25515 overall option summary.
25517 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25519 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25520 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25521 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25522 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25524 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25526 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25527 -mpower9-minmax by default for -mcpu=power9.
25528 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25529 128-bit floating point.
25531 2017-01-20 Alan Modra <amodra@gmail.com>
25533 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25534 optimizing for size.
25536 2017-01-20 Alan Modra <amodra@gmail.com>
25539 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25540 for strcmp and strncmp from corresponding builtin decl.
25542 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25544 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25545 instead of i386/rtems-64.h.
25546 * config/i386/rtems-64.h: Remove.
25548 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25552 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25554 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25556 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25558 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25559 Change int to HOST_WIDE_INT.
25560 * config/aarch64/aarch64-protos.h
25561 (aarch64_simd_gen_const_vector_dup): Likewise.
25562 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25564 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25566 * langhooks-def.h (lhd_type_for_size): New decl.
25567 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25568 * langhooks.c (lhd_type_for_size): New function, taken from
25571 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25573 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25574 define_bypass for CR latency.
25575 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25576 (power9-alu2): Add define_bypass for CR latency.
25578 (power9-mul): Update insn latency.
25579 (power9-mul-compare): Update insn latency, bypass latency and remove
25582 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25584 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25586 * config/aarch64/aarch64.md
25587 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25588 aarch64_nopcrelative_literal_loads.
25589 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25591 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25593 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25594 TARGET_LOONGSON_3A.
25595 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25597 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25600 * config.gcc (supported_defaults): Add lxc1-sxc1.
25601 (with_lxc1_sxc1): Add validation.
25602 (all_defaults): Add lxc1-sxc1.
25603 * config/mips/mips.opt (mlxc1-sxc1): New option.
25604 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25606 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25607 __mips_no_lxc1_sxc1.
25608 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25609 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25610 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25612 2017-01-19 Richard Biener <rguenther@suse.de>
25614 PR tree-optimization/72488
25615 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25616 sure to restore SSA info.
25617 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25619 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25621 PR rtl-optimization/79121
25622 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25623 of the inner type when shifting an extended value.
25625 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25628 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25629 interposable aliases.
25631 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25634 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25635 Use the evmergelohi instruction.
25636 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25637 (mov_sitf_e500_subreg8_2_be): Likewise.
25638 (mov_sitf_e500_subreg12_2_le): Likewise.
25639 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25640 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25641 (mov_sitf_e500_subreg8_2_le): Likewise.
25642 (mov_sitf_e500_subreg12_2_be): Likewise.
25644 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25646 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25647 attribute from vecsimple to vecperm.
25648 (altivec_vbpermq2): Likewise.
25650 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25653 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25655 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25656 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25657 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25658 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25659 case where N arg is SIZE_MAX.
25660 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25661 (cmpstrsi): Add pattern.
25663 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25665 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25666 __builtin_vec_revb builtins.
25667 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25668 built-in functions to support generation of the ISA 3.0 XXBR<x>
25669 vector byte reverse instructions.
25670 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25671 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25672 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25673 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25674 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25675 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25676 (P9V_BUILTIN_VEC_REVB): Likewise.
25677 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25678 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25679 (p9_xxbrq_v16qi): Likewise.
25680 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25681 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25682 (p9_xxbrh_v8hi): Likewise.
25683 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25684 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25685 vec_revb built-in functions.
25687 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25689 PR rtl-optimization/78952
25690 * config/i386/i386.md (any_extract): New code iterator.
25691 (*insvqi_2): Use any_extract for source operand.
25692 (*insvqi_3): Use any_shiftrt for source operand.
25694 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25696 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25698 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25700 2017-01-18 Matthias Klose <doko@ubuntu.com>
25702 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25704 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25706 * config/rs6000/altivec.h (vec_bperm): Change #define.
25707 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25708 (altivec_vbpermq2): New define_insn.
25709 (altivec_vbpermd): Likewise.
25710 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25711 function interface.
25712 (VBPERMD): Likewise.
25713 (VBPERM): New polymorphic function interface.
25714 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25715 Add entries for P9V_BUILTIN_VEC_VBPERM.
25716 * doc/extend.texi: Add interfaces for vec_bperm.
25718 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25720 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25721 first letter of error messages.
25722 (s390_resolve_overloaded_builtin): Likewise.
25723 * config/s390/s390.c (s390_expand_builtin): Likewise.
25724 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25725 (s390_valid_target_attribute_inner_p): Likewise.
25726 * config/s390/s390.md ("tabort"): Likewise.
25728 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25730 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25731 (ISA_AVOID_DIV_HILO): New macro.
25732 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25733 (ISA_HAS_DDIV): Likewise.
25735 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25737 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25739 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25741 * doc/invoke.texi (fabi-version): Spelling fix.
25743 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25746 * doc/invoke.texi (fabi-version): Mention mangling fix for
25749 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25752 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25754 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25757 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25758 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25760 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25762 (stack_protect_test): Ditto.
25763 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25764 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25766 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25767 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25768 -mstack-protector-guard-offset=.
25769 (RS/6000 and PowerPC Options): Ditto.
25771 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25773 * config/i386/i386.h (MASK_CLASS_P): New define.
25774 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25775 there are no registers from different register sets also when
25776 mask registers are used. Update function comment.
25777 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25778 to (*k/*r) and (*k/*km) alternatives.
25780 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25782 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25783 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25784 (EH_RETURN_HANDLER_RTX): New define.
25785 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25786 Force frame pointer in EH return functions.
25787 (aarch64_expand_epilogue): Add barrier for eh_return.
25788 (aarch64_final_eh_return_addr): Remove.
25789 (aarch64_eh_return_handler_rtx): New function.
25790 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25792 (aarch64_eh_return_handler_rtx): New prototype.
25794 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25796 * config/rs6000/altivec.h (vec_rlmi): New #define.
25797 (vec_vrlnm): Likewise.
25798 (vec_rlnm): Likewise.
25799 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25800 (UNSPEC_VRLNM): Likewise.
25801 (VIlong): New mode iterator.
25802 (altivec_vrl<VI_char>mi): New define_insn.
25803 (altivec_vrl<VI_char>nm): Likewise.
25804 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25806 (VRLDNM): Likewise.
25807 (RLNM): New polymorphic function entry.
25808 (VRLWMI): New monomorphic function entry.
25809 (VRLDMI): Likewise.
25810 (RLMI): New polymorphic function entry.
25811 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25812 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25813 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25816 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25819 * dwarf2out.c (field_byte_offset): Restore the
25820 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25821 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25822 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25825 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25828 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25830 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25833 * dwarf2out.c (add_data_member_location_attribute): For constant
25834 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25835 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25836 DW_AT_byte_size attributes.
25838 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25840 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25841 after forcing to constant memory when the code model is medium.
25843 2017-01-17 Julia Koval <julia.koval@intel.com>
25846 * config/i386/avx512fintrin.h
25847 (_mm512_i32gather_ps): Change __addr type to void const*.
25848 (_mm512_mask_i32gather_ps): Ditto.
25849 (_mm512_i32gather_pd): Ditto.
25850 (_mm512_mask_i32gather_pd): Ditto.
25851 (_mm512_i64gather_ps): Ditto.
25852 (_mm512_mask_i64gather_ps): Ditto.
25853 (_mm512_i64gather_pd): Ditto.
25854 (_mm512_mask_i64gather_pd): Ditto.
25855 (_mm512_i32gather_epi32): Ditto.
25856 (_mm512_mask_i32gather_epi32): Ditto.
25857 (_mm512_i32gather_epi64): Ditto.
25858 (_mm512_mask_i32gather_epi64): Ditto.
25859 (_mm512_i64gather_epi32): Ditto.
25860 (_mm512_mask_i64gather_epi32): Ditto.
25861 (_mm512_i64gather_epi64): Ditto.
25862 (_mm512_mask_i64gather_epi64): Ditto.
25863 (_mm512_i32scatter_ps): Change __addr type to void*.
25864 (_mm512_mask_i32scatter_ps): Ditto.
25865 (_mm512_i32scatter_pd): Ditto.
25866 (_mm512_mask_i32scatter_pd): Ditto.
25867 (_mm512_i64scatter_ps): Ditto.
25868 (_mm512_mask_i64scatter_ps): Ditto.
25869 (_mm512_i64scatter_pd): Ditto.
25870 (_mm512_mask_i64scatter_pd): Ditto.
25871 (_mm512_i32scatter_epi32): Ditto.
25872 (_mm512_mask_i32scatter_epi32): Ditto.
25873 (_mm512_i32scatter_epi64): Ditto.
25874 (_mm512_mask_i32scatter_epi64): Ditto.
25875 (_mm512_i64scatter_epi32): Ditto.
25876 (_mm512_mask_i64scatter_epi32): Ditto.
25877 (_mm512_i64scatter_epi64): Ditto.
25878 (_mm512_mask_i64scatter_epi64): Ditto.
25879 * config/i386/avx512pfintrin.h
25880 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25881 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25882 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25883 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25884 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25885 (_mm512_prefetch_i32scatter_ps): Ditto.
25886 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25887 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25888 (_mm512_prefetch_i64scatter_pd): Ditto.
25889 (_mm512_prefetch_i64scatter_ps): Ditto.
25890 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25891 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25892 * config/i386/avx512vlintrin.h
25893 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25894 (_mm_mmask_i32gather_ps): Ditto.
25895 (_mm256_mmask_i32gather_pd): Ditto.
25896 (_mm_mmask_i32gather_pd): Ditto.
25897 (_mm256_mmask_i64gather_ps): Ditto.
25898 (_mm_mmask_i64gather_ps): Ditto.
25899 (_mm256_mmask_i64gather_pd): Ditto.
25900 (_mm_mmask_i64gather_pd): Ditto.
25901 (_mm256_mmask_i32gather_epi32): Ditto.
25902 (_mm_mmask_i32gather_epi32): Ditto.
25903 (_mm256_mmask_i32gather_epi64): Ditto.
25904 (_mm_mmask_i32gather_epi64): Ditto.
25905 (_mm256_mmask_i64gather_epi32): Ditto.
25906 (_mm_mmask_i64gather_epi32): Ditto.
25907 (_mm256_mmask_i64gather_epi64): Ditto.
25908 (_mm_mmask_i64gather_epi64): Ditto.
25909 (_mm256_i32scatter_ps): Change __addr type to void*.
25910 (_mm256_mask_i32scatter_ps): Ditto.
25911 (_mm_i32scatter_ps): Ditto.
25912 (_mm_mask_i32scatter_ps): Ditto.
25913 (_mm256_i32scatter_pd): Ditto.
25914 (_mm256_mask_i32scatter_pd): Ditto.
25915 (_mm_i32scatter_pd): Ditto.
25916 (_mm_mask_i32scatter_pd): Ditto.
25917 (_mm256_i64scatter_ps): Ditto.
25918 (_mm256_mask_i64scatter_ps): Ditto.
25919 (_mm_i64scatter_ps): Ditto.
25920 (_mm_mask_i64scatter_ps): Ditto.
25921 (_mm256_i64scatter_pd): Ditto.
25922 (_mm256_mask_i64scatter_pd): Ditto.
25923 (_mm_i64scatter_pd): Ditto.
25924 (_mm_mask_i64scatter_pd): Ditto.
25925 (_mm256_i32scatter_epi32): Ditto.
25926 (_mm256_mask_i32scatter_epi32): Ditto.
25927 (_mm_i32scatter_epi32): Ditto.
25928 (_mm_mask_i32scatter_epi32): Ditto.
25929 (_mm256_i32scatter_epi64): Ditto.
25930 (_mm256_mask_i32scatter_epi64): Ditto.
25931 (_mm_i32scatter_epi64): Ditto.
25932 (_mm_mask_i32scatter_epi64): Ditto.
25933 (_mm256_i64scatter_epi32): Ditto.
25934 (_mm256_mask_i64scatter_epi32): Ditto.
25935 (_mm_i64scatter_epi32): Ditto.
25936 (_mm_mask_i64scatter_epi32): Ditto.
25937 (_mm256_i64scatter_epi64): Ditto.
25938 (_mm256_mask_i64scatter_epi64): Ditto.
25939 (_mm_i64scatter_epi64): Ditto.
25940 (_mm_mask_i64scatter_epi64): Ditto.
25941 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25942 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25943 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25944 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25945 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25946 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25947 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25948 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25949 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25950 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25951 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25952 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25953 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25954 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25955 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25956 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25957 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25958 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25959 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25960 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25961 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25962 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25963 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25964 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25965 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25966 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25967 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25968 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25969 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25970 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25971 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25972 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25973 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25974 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25975 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25976 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25977 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25978 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25979 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25980 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25981 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25982 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25983 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25984 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25985 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25986 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25987 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25988 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25989 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25990 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25991 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25992 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25993 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25994 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25995 definitions accordingly.
25997 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25998 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26001 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26004 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26007 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26008 hard regs for inner regno.
26010 2017-01-17 Martin Liska <mliska@suse.cz>
26013 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26014 assumption and add comment.
26016 2017-01-17 Nathan Sidwell <nathan@acm.org>
26018 * ipa-visibility.c (localize_node): New function, broken out of ...
26019 (function_and_variable_visibility): ... here. Call it.
26021 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26023 PR middle-end/77445
26024 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26025 correctly set frequency of oudgoing edge.
26026 (duplicate_thread_path): Fix profile updating.
26028 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26031 * configure.ac: Add GCC_BASE_VER.
26032 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26033 version from BASE-VER file.
26034 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26035 (gcc.o): Depend on $(BASEVER).
26036 * common.opt (dumpfullversion): New option.
26037 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26038 * doc/invoke.texi: Document -dumpfullversion.
26039 * doc/install.texi: Document --with-gcc-major-version-only.
26040 * configure: Regenerated.
26042 2017-01-17 Richard Biener <rguenther@suse.de>
26044 PR tree-optimization/71433
26045 * tree-vrp.c (register_new_assert_for): Merge same asserts
26046 on all incoming edges.
26047 (process_assert_insertions_for): Handle insertions at the
26050 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26052 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26053 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26055 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26058 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26061 2017-01-17 Alan Modra <amodra@gmail.com>
26064 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26065 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26066 symbolic stack limit when pic.
26068 2017-01-16 Martin Sebor <msebor@redhat.com>
26070 PR tree-optimization/78608
26071 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26073 2017-01-16 Jeff Law <law@redhat.com>
26076 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26077 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26078 for several include directories that may be relative to sysroot.
26079 * config/i386/x-mingw32 (gplus_includedir): Define.
26080 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26081 (native_system_includedir): Likewise.
26082 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
26083 override if TARGET_SYSTEM_ROOT is defined.
26084 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
26086 PR tree-optimization/79090
26087 PR tree-optimization/33562
26088 PR tree-optimization/61912
26089 PR tree-optimization/77485
26090 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
26091 and computed trims into the dump file.
26093 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26095 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
26097 2017-01-16 Jakub Jelinek <jakub@redhat.com>
26100 * gimplify.c (gimplify_init_constructor): If want_value and
26101 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26105 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26106 sequence. Formatting fixes.
26107 (doloop_optimize): Formatting fixes.
26110 * gcc.c (debug_level_greater_than_spec_func): New function.
26111 (static_spec_functions): Add debug-level-gt spec function.
26112 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
26114 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26115 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26116 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
26117 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
26118 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
26119 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
26121 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
26123 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
26124 QImode fixups to general and mask registers only.
26126 2017-01-16 Carl Love <cel@us.ibm.com>
26128 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
26129 for built-in functions
26130 vector signed char vec_nabs (vector signed char)
26131 vector signed short vec_nabs (vector signed short)
26132 vector signed int vec_nabs (vector signed int)
26133 vector signed long long vec_nabs (vector signed long long)
26134 vector float vec_nabs (vector float)
26135 vector double vec_nabs (vector double)
26136 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26138 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26139 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26140 * doc/extend.texi: Update the documentation file for the new built-in
26143 2017-01-16 Martin Sebor <msebor@redhat.com>
26145 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26148 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26150 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26151 UNSPEC_VSX__XXSPLTD to require special splat handling.
26153 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26156 * system.h: Poison strndup.
26158 2017-01-16 Alan Modra <amodra@gmail.com>
26161 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26164 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26166 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26168 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26170 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26171 call recog here. Assert that INSN_CODE (insn) is non-negative.
26173 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26176 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26178 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26179 in the currently scheduled RTL fragment.
26181 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26183 PR rtl-optimization/78751
26184 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26187 2017-01-14 Jeff Law <law@redhat.com>
26189 PR tree-optimization/79090
26190 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26191 variable length stores.
26192 (compute_trims): Delete dead assignment to *trim_tail.
26193 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26196 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26198 PR rtl-optimization/78626
26199 PR rtl-optimization/78727
26200 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26201 of a block, and split such blocks after everything else is finished.
26203 2017-01-14 Alan Modra <amodra@gmail.com>
26206 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26207 target legitimate_combined_insn.
26208 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26209 (rs6000_legitimate_combined_insn): New function.
26210 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26212 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26213 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26214 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26216 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26218 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26220 2017-01-13 Jeff Law <law@redhat.com>
26222 PR tree-optimization/33562
26223 PR tree-optimization/61912
26224 PR tree-optimization/77485
26225 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26227 (delete_dead_assignment): Likewise.
26228 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26229 statement to delete_dead_call and delete_dead_assignment.
26231 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26234 * substring-locations.c (format_warning_va): Strengthen case 1 so
26235 that both endpoints of the substring must be within the format
26236 range for just the substring to be printed.
26238 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26240 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26241 * config/i386/i386.c (ix86_target_string): Add missing options
26242 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26243 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26244 flags_other and ix86_target_other to flags2_other. Display unknown
26246 (ix86_valid_target_attribute_inner_p): Add missing options and
26247 reorder options by implied ISAs, as in ix86_target_string.
26249 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26251 * hash-table.h (hash_table::too_empty_p): New function.
26252 (hash_table::expand): Use it.
26253 (hash_table::traverse): Likewise.
26254 (hash_table::empty_slot): Use sizeof (value_type) instead of
26255 sizeof (PTR) to convert bytes to elements. Shrink the table
26256 if the current size is excessive for the current number of
26259 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26261 * ira-costs.c (record_reg_classes): Break from the inner loop
26262 early once alt_fail is known to be true. Update outer loop
26263 handling accordingly.
26265 2017-01-13 Jeff Law <law@redhat.com>
26267 * tree-ssa-dse.c (decrement_count): New function.
26268 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26269 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26270 when we know the partially dead statement is a mem* function.
26272 PR tree-optimization/61912
26273 PR tree-optimization/77485
26274 * tree-ssa-dse.c: Include expr.h.
26275 (maybe_trim_constructor_store): New function.
26276 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26278 PR tree-optimization/33562
26279 PR tree-optimization/61912
26280 PR tree-optimization/77485
26281 * doc/invoke.texi: Document new dse-max-object-size param.
26282 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26283 * tree-ssa-dse.c: Include params.h.
26284 (dse_store_status): New enum.
26285 (initialize_ao_ref_for_dse): New, partially extracted from
26287 (valid_ao_ref_for_dse, normalize_ref): New.
26288 (setup_live_bytes_from_ref, compute_trims): Likewise.
26289 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26290 (maybe_trim_partially_dead_store): Likewise.
26291 (maybe_trim_complex_store): Likewise.
26292 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26293 Track what bytes live from the original store. Return tri-state
26294 for dead, partially dead or live.
26295 (dse_dom_walker): Add constructor, destructor and new private members.
26296 (delete_dead_call, delete_dead_assignment): New extracted from
26298 (dse_optimize_stmt): Make a member of dse_dom_walker.
26299 Use initialize_ao_ref_for_dse.
26301 PR tree-optimization/33562
26302 PR tree-optimization/61912
26303 PR tree-optimization/77485
26304 * sbitmap.h (bitmap_count_bits): Prototype.
26305 (bitmap_clear_range, bitmap_set_range): Likewise.
26306 * sbitmap.c (bitmap_clear_range): New function.
26307 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26309 2017-01-13 Martin Liska <mliska@suse.cz>
26312 * function.c (set_cfun): Add new argument force.
26313 * function.h (set_cfun): Likewise.
26314 * ipa-inline-transform.c (inline_call): Use the function when
26315 strict alising from is dropped for function we inline to.
26317 2017-01-13 Richard Biener <rguenther@suse.de>
26319 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26320 for dumping GIMPLE INTEGER_CSTs.
26322 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26324 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26325 to 201112L since C++17.
26327 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26330 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26331 if -fsanitize=kernel-address is present.
26333 2017-01-13 Richard Biener <rguenther@suse.de>
26335 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26336 as _Literal ( type ) number in case usual suffixes do not
26337 preserve all information.
26339 2017-01-13 Richard Biener <rguenther@suse.de>
26341 PR tree-optimization/77283
26342 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26343 and ssa-iterators.h.
26344 (is_feasible_trace): Implement a cost model based on joiner
26347 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26350 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26351 char or short to __float128/_Float128 directly.
26353 2017-01-12 Martin Sebor <msebor@redhat.com>
26355 to -Wformat-overflow.
26356 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26357 (min_bytes_remaining): Same.
26358 (get_string_length): Same.
26359 (format_string): Same.
26360 (format_directive): Same.
26362 (pass_sprintf_length::handle_gimple_call): Same.
26364 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26366 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26367 info.nowrite calls with no lhs that can't throw. Return bool
26368 whether gsi_remove has been called or not.
26369 (pass_sprintf_length::handle_gimple_call): Return bool whether
26370 try_substitute_return_value called gsi_remove. Formatting fix.
26371 (pass_sprintf_length::execute): Don't use gsi_remove if
26372 handle_gimple_call returned true.
26375 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26376 be removed due to side-effects, don't remove following barrier nor
26377 turn the successor edge into fallthru edge.
26379 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26382 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26383 element-reversing loads and stores as not swappable.
26385 2017-01-12 Nathan Sidwell <nathan@acm.org>
26386 Nicolai Stange <nicstange@gmail.com>
26388 * combine.c (try_combine): Don't ignore result of overlap checking
26389 loop. Combine overlap & asm check into single loop.
26391 2017-01-12 Richard Biener <rguenther@suse.de>
26393 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26394 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26396 2017-01-12 Richard Biener <rguenther@suse.de>
26398 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26399 and TS_TARGET_OPTION directly derive from TS_BASE.
26400 * tree-core.h (tree_optimization_option): Derive from tree_base.
26401 (tree_target_option): Likewise.
26403 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26405 * config/i386/i386.c (memory_address_length): Increase len
26406 only when rip_relative_addr_p returns false.
26408 2017-01-11 Julia Koval <julia.koval@intel.com>
26410 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26411 (OPTION_MASK_ISA_SGX_SET): New.
26412 (ix86_handle_option): Handle OPT_msgx.
26413 * config.gcc: Added sgxintrin.h.
26414 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26415 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26416 * config/i386/i386.c (ix86_target_string): Add -msgx.
26418 (ix86_option_override_internal): Handle new options.
26419 (ix86_valid_target_attribute_inner_p): Add sgx.
26420 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26421 * config/i386/i386.opt: Add msgx.
26422 * config/i386/sgxintrin.h: New file.
26423 * config/i386/x86intrin.h: Add sgxintrin.h.
26425 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26428 * fold-const.c (maybe_nonzero_address): Return 1 for function
26430 (tree_single_nonzero_warnv_p): Don't handle function local objects
26434 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26437 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26440 * opts.c: Include "spellcheck.h"
26441 (struct string_fragment): New struct.
26442 (struct edit_distance_traits<const string_fragment &>): New
26444 (get_closest_sanitizer_option): New function.
26445 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26447 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26449 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26451 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26452 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26453 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26454 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26455 for initial die_offset if dwarf_split_debug_info.
26456 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26457 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26459 (output_skeleton_debug_sections): Formatting fix. Use
26460 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26461 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26463 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26465 * config/arm/cortex-a53.md: Add bypasses for
26466 cortex_a53_r2f_cvt.
26467 (cortex_a53_r2f): Only use for transfers.
26468 (cortex_a53_f2r): Likewise.
26469 (cortex_a53_r2f_cvt): Add reservation for conversions.
26470 (cortex_a53_f2r_cvt): Likewise.
26472 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26474 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26475 to all inlined functions, change static to extern.
26477 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26480 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26482 (arm_assemble_integer): Likewise.
26484 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26486 * config.gcc: Use new awk script to check CPU, FPU and architecture
26487 parameters for --with-... options.
26488 * config/arm/parsecpu.awk: New file
26489 * config/arm/arm-cpus.in: New file.
26490 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26492 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26494 * config/arm/t-arm: Update dependency rules.
26495 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26496 of processing .def files.
26497 * config/arm/genopt.sh: Deleted.
26498 * config/arm/gentune.sh: Deleted.
26499 * config/arm/arm-cores.def: Deleted.
26500 * config/arm/arm-arches.def: Deleted.
26501 * config/arm/arm-fpus.def: Deleted.
26502 * config/arm/arm-tune.md: Regenerated.
26503 * config/arm/arm-tables.opt: Regenerated.
26504 * config/arm/arm-cpu.h: New generated file.
26505 * config/arm/arm-cpu-data.h: New generated file.
26506 * config/arm/arm-cpu-cdata.h: New generated file.
26508 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26511 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26513 (input_varpool_node): Unpack dynamically_initialized bit.
26515 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26517 PR rtl-optimization/79032
26518 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26519 the alignment of the adjusted memory reference against that of MODE,
26520 instead of the alignment of the original memory reference.
26522 2017-01-11 Martin Jambor <mjambor@suse.cz>
26524 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26526 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26527 decorated functions.
26529 2017-01-11 Richard Biener <rguenther@suse.de>
26531 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26532 set range/nonnull info for PHI results. Do not set it on
26533 stmts marked for removal.
26535 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26537 * expr.c (store_field): In the bitfield case, fetch the return value
26538 from the registers before applying a single big-endian adjustment.
26539 Always do a final load for a BLKmode value not larger than a word.
26541 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26544 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26545 that we correctly handle column numbers greater than
26546 LINE_MAP_MAX_COLUMN_NUMBER.
26548 2017-01-10 Martin Sebor <msebor@redhat.com>
26550 PR middle-end/78245
26551 * gimple-ssa-sprintf.c (get_destination_size): Call
26552 {init,fini}object_sizes.
26553 * tree-object-size.c (addr_object_size): Adjust.
26554 (pass_through_call): Adjust.
26555 (pass_object_sizes::execute): Adjust.
26556 * tree-object-size.h (fini_object_sizes): Declare.
26558 2017-01-10 Martin Sebor <msebor@redhat.com>
26560 PR tree-optimization/78775
26561 * builtins.c (get_size_range): Move...
26562 * calls.c: ...to here.
26563 (alloc_max_size): Accept zero argument.
26564 (operand_signed_p): Remove.
26565 (maybe_warn_alloc_args_overflow): Call get_size_range.
26566 * calls.h (get_size_range): Declare.
26568 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26570 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26571 from TI's devices.csv file as of September 2016.
26572 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26574 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26576 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26577 * doc/invoke.texi: Likewise.
26578 * doc/md.texi: Likewise.
26579 * doc/objc.texi: Likewise.
26581 2017-01-10 Joshua Conner <joshconner@google.com>
26583 * config/arm/fuchsia-elf.h: New file.
26584 * config/fuchsia.h: New file.
26585 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26586 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26588 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26590 2016-01-10 Richard Biener <rguenther@suse.de>
26592 PR tree-optimization/79034
26593 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26594 Propagate out degenerate PHIs in the joiner.
26596 2017-01-10 Martin Liska <mliska@suse.cz>
26598 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26599 (sort_congruence_classes_by_decl_uid): Likewise.
26600 (sort_congruence_class_groups_by_decl_uid): Likewise.
26601 (sem_item_optimizer::merge_classes): Sort class, groups in these
26602 classes and members in the groups by DECL_UID of declarations.
26603 This would make merge operations stable.
26605 2017-01-10 Martin Liska <mliska@suse.cz>
26607 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26608 usage of m_classes_vec.
26609 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26610 (sem_item_optimizer::get_group_by_hash): Likewise.
26611 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26612 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26613 (sem_item_optimizer::verify_classes): Likewise.
26614 (sem_item_optimizer::process_cong_reduction): Likewise.
26615 (sem_item_optimizer::dump_cong_classes): Likewise.
26616 (sem_item_optimizer::merge_classes): Likewise.
26617 * ipa-icf.h (congruence_class_hash): Rename from
26618 congruence_class_group_hash. Remove declaration of m_classes_vec.
26620 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26622 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26623 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26624 * config.gcc: Add avx512vpopcntdqintrin.h.
26625 * config/i386/avx512vpopcntdqintrin.h: New.
26626 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26627 * config/i386/i386-builtin-types.def: Add new types.
26628 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26629 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26630 __builtin_ia32_vpopcountq_v8di_mask): New.
26631 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26632 __AVX512VPOPCNTDQ__.
26633 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26634 (PTA_AVX512VPOPCNTDQ): Define.
26635 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26636 TARGET_AVX512VPOPCNTDQ_P): Define.
26637 * config/i386/i386.opt: Add mavx512vpopcntdq.
26638 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26639 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26641 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26643 PR middle-end/77484
26644 * predict.def (PRED_CALL): Set to 67.
26646 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26648 * expr.c (store_field): In the bitfield case, if the value comes from
26649 a function call and is of an aggregate type returned in registers, do
26650 not modify the field mode; extract the value in all cases if the mode
26651 is BLKmode and the size is not larger than a word.
26653 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26656 * config/i386/cpuid.h: Fix undefined behavior.
26658 2017-01-04 Jeff Law <law@redhat.com>
26660 PR tree-optimization/79007
26661 PR tree-optimization/67955
26662 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26663 conservative for pt.null when flag_non_call_exceptions is on.
26665 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26667 PR translation/79019
26668 PR translation/79020
26669 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26670 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26671 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26673 * config/avr/avr.opt (maccumulate-args): Likewise.
26674 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26675 * common.opt (freport-bug): Likewise.
26676 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26677 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26678 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26679 translatable string.
26680 * config/i386/i386.c (function_value_32): Likewise.
26681 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26682 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26684 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26685 * common/config/msp430/msp430-common.c (msp430_handle_option):
26687 * symtab.c (symtab_node::verify_base): Likewise.
26688 * opts.c (set_debug_level): Likewise.
26689 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26690 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26691 missing whitespace to translatable strings.
26692 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26693 * config/sh/superh.h: Likewise.
26694 * config/i386/xopintrin.h: Likewise.
26695 * config/i386/znver1.md: Likewise.
26696 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26697 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26698 * double-int.h (struct double_int): Likewise.
26699 * double-int.c (div_and_round_double): Likewise.
26700 * wide-int.cc: Likewise.
26701 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26702 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26703 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26704 (crossjumps_occurred): ... this.
26705 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26708 PR tree-optimization/78899
26709 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26710 returning bool return struct loop *, NULL for failure and the new
26712 (versionable_outer_loop_p): Don't version outer loop if it has
26713 dont_vectorized bit set.
26714 (tree_if_conversion): When versioning outer loop, ensure
26715 tree_if_conversion is performed also on the inner loop of the
26716 non-vectorizable outer loop copy.
26717 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26718 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26719 prevent vectorization of it.
26720 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26721 the outer loop vectorization of the non-scalar version is attempted
26722 before vectorization of the inner loop in scalar version. If
26723 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26724 vectorization of its inner loop.
26725 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26726 has 2 inner loops, rename also on edges from bb whose single pred
26727 is outer_loop->header. Fix typo in function comment.
26729 2017-01-09 Martin Sebor <msebor@redhat.com>
26732 * asan.c (asan_emit_stack_protection): Increase local buffer size
26733 to avoid snprintf truncation warning.
26735 2017-01-09 Andrew Pinski <apinski@cavium.com>
26737 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26738 to reference thunderx2t99 for the tuning structure
26739 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26741 (thunderx2t99_extra_costs): This.
26742 * config/aarch64/aarch64-tune.md: Regenerate.
26743 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26744 (vulcan_addrcost_table): This.
26745 (vulcan_regmove_cost): Rename to ...
26746 (thunderx2t99_regmove_cost): This.
26747 (vulcan_vector_cost): Rename to ...
26748 (thunderx2t99_vector_cost): this.
26749 (vulcan_branch_cost): Rename to ...
26750 (thunderx2t99_branch_cost): This.
26751 (vulcan_tunings): Rename to ...
26752 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26753 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26755 2017-01-09 Martin Jambor <mjambor@suse.cz>
26759 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26760 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26761 (propagate_vr_accross_jump_function): Use the above function for all
26762 value range computations for pass-through jump functions and type
26763 converasion from explicit value range values.
26764 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26765 parameters from TYPE_ARG_TYPES.
26766 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26767 (ipa_write_node_info): Stream type of the actual argument.
26768 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26770 2017-01-09 Martin Liska <mliska@suse.cz>
26773 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26774 (lookup_compiler): Do not show error message with have_E.
26776 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26778 PR tree-optimization/78938
26779 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26780 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26781 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26782 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26785 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26787 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26790 2017-01-09 Richard Biener <rguenther@suse.de>
26792 PR tree-optimization/78997
26793 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26794 name condition properly.
26796 2017-01-09 Richard Biener <rguenther@suse.de>
26799 * dwarf2out.c (is_cxx): New overload with context.
26800 (is_naming_typedef_decl): Use it.
26802 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26804 * invoke.texi (Option Summary): Correct spacing in option lists
26805 and add line breaks to fix over-long lines.
26807 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26809 PR middle-end/17660
26811 * extend.texi (Common Variable Attributes): Add xref to GCC
26812 Internals manual to explain mode attribute keywords.
26814 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26817 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26818 and Preprocessor Options.
26819 (Options for Linking): Document -pthread here....
26820 (RS/6000 and PowerPC Options): ...not here.
26821 (Solaris 2 Options): ...or here.
26822 * doc/cppopts.texi: Document -pthread.
26824 2017-01-08 Martin Sebor <msebor@redhat.com>
26826 PR middle-end/77708
26827 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26828 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26829 New member functions.
26830 (format_directive): Used them.
26832 (pass_sprintf_length::handle_gimple_call): Same.
26833 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26834 to avoid truncation for any argument.
26835 (extract_affine_mul): Same.
26836 * tree.c (get_file_function_name): Same.
26838 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26840 PR middle-end/77484
26841 * predict.def (PRED_INDIR_CALL): Set to 86.
26843 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26845 PR preprocessor/54124
26846 * doc/cppopts.texi: Reformat -d subtable to list the full name
26847 of the options. Add cross-reference to the docs for the general
26848 compiler -d options.
26849 * doc/invoke.texi (Developer Options): Add cross-reference to the
26850 preprocessor-specific -d option documentation.
26852 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26854 PR preprocessor/13498
26855 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26856 redudant material, and reflect new command-line options.
26857 (System Headers): Likewise.
26859 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26861 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26862 -isystem, and -idirafter. Copy-edit.
26863 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26864 default for -ftrack-macro-expansion. Delete obsolete and
26865 badly-formatted implementation details about -fdebug-cpp output.
26866 * doc/cppwarnopts.texi: Copy-edit.
26868 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26871 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26872 that the transition from a max line width >= 1<<10 to narrower
26873 lines works correctly.
26875 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26877 * doc/options.texi (PerFunction): New.
26878 * opt-functions.awk (switch_flags): Map both Optimization and
26879 PerFunction to CL_OPTIMIZATION.
26880 * opth-gen.awk: Test for PerFunction flag along with
26882 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26883 it only when the latter is present. Skip those that don't in
26884 the hash function generator.
26885 * common.opt (fvar-tracking): Mark as PerFunction instead of
26887 (fvar-tracking-assignments): Likewise.
26888 (fvar-tracking-assignments-toggle): Likewise.
26889 (fvar-tracking-uninit): Likewise.
26891 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26893 PR translation/79018
26894 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26897 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26900 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26901 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26902 TARGET_LONG_JUMP_TABLE_OFFSETS.
26903 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26904 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26905 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26906 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26907 * config/m68k/m68k.md (tablejump expander): Likewise.
26908 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26909 TARGET_LONG_JUMP_TABLE_OFFSETS.
26910 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26911 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26913 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26914 David Holsgrove <david.holsgrove@xilinx.com>
26916 * common/config/microblaze/microblaze-common.c
26917 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26918 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26920 * config/microblaze/microblaze.c (microblaze_must_save_register)
26921 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26923 (microblaze_eh_return): New function.
26924 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26925 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26926 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26927 * config/microblaze/microblaze.md (eh_return): New pattern.
26929 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26931 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26932 GCC_DIAGNOSTIC_STRINGIFY): Define.
26934 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26936 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26938 * config/arm/arm.md (<mcrr>): New.
26940 * config/arm/arm.c (arm_arch5te): New.
26941 (arm_option_override): Set arm_arch5te.
26942 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26944 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26945 (arm_mcrr_qualifiers): ... this. New.
26946 (MRRC_QUALIFIERS): Define to...
26947 (arm_mrrc_qualifiers): ... this. New.
26948 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26950 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26951 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26952 (MRRCI, mrrc, MRRC): New.
26953 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26954 VUNSPEC_MRRC2): New.
26956 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26958 * config/arm/arm.md (<mcr>): New.
26960 * config/arm/arm.c (arm_coproc_builtin_available): Add
26961 support for mcr, mrc, mcr2 and mrc2.
26962 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26963 (arm_mcr_qualifiers): ... this. New.
26964 (MRC_QUALIFIERS): Define to ...
26965 (arm_mrc_qualifiers): ... this. New.
26966 (MCR_QUALIFIERS): Define to ...
26967 (arm_mcr_qualifiers): ... this. New.
26968 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26970 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26971 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26972 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26973 VUNSPEC_MRC2): New.
26975 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26977 * config/arm/arm.md (*ldc): New.
26981 * config/arm/arm.c (arm_coproc_builtin_available): Add
26982 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26983 (arm_coproc_ldc_stc_legitimate_address): New.
26984 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26985 'qualifier_const_pointer'.
26986 (LDC_QUALIFIERS): Define to...
26987 (arm_ldc_qualifiers): ... this. New.
26988 (STC_QUALIFIERS): Define to...
26989 (arm_stc_qualifiers): ... this. New.
26990 * config/arm/arm-protos.h
26991 (arm_coproc_ldc_stc_legitimate_address): New.
26992 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26993 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26994 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26995 stc2, stcl, stc2l): New.
26996 * config/arm/constraints.md (Uz): New.
26997 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26998 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26999 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27000 VUNSPEC_STC2L): New.
27002 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27004 * config/arm/arm.md (<cdp>): New.
27005 * config/arm/arm.c (neon_const_bounds): Rename this ...
27006 (arm_const_bounds): ... this.
27007 (arm_coproc_builtin_available): New.
27008 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27009 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27010 (CDP_QUALIFIERS): Define to...
27011 (arm_cdp_qualifiers): ... this. New.
27013 (arm_expand_builtin_args): Add case for 6 arguments.
27014 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27015 (arm_const_bounds): ... this.
27016 (arm_coproc_builtin_available): New.
27017 * config/arm/arm_acle.h (__arm_cdp): New.
27019 * config/arm/arm_acle_builtins.def (cdp): New.
27021 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27022 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27023 'arm_const_bounds'.
27024 * config/arm/types.md (coproc): New.
27025 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27026 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27027 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27028 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27030 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27032 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27033 (UBINOP_QUALIFIERS): New.
27035 (acle_builtin_data): New. Change comment.
27036 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27037 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27038 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27039 arm_acle_builtins.def.
27040 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27041 (arm_init_acle_builtins): New.
27042 (CRC32_BUILTIN): Remove.
27043 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27044 crc32cb, crc32ch and crc32cw.
27045 (arm_init_crc32_builtins): Remove.
27046 (arm_init_builtins): Use arm_init_acle_builtins rather
27047 than arm_init_crc32_builtins.
27048 (arm_expand_acle_builtin): New.
27049 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27050 * config/arm/arm_acle_builtins.def: New.
27052 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27054 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27055 (arm_builtin_datum): ... this.
27056 (arm_init_neon_builtin): Rename to ...
27057 (arm_init_builtin): ... this. Add a new parameters PREFIX
27058 and USE_SIG_IN_NAME.
27059 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27060 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27061 'arm_builtin_datum'.
27062 (arm_init_vfp_builtins): Likewise.
27063 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27064 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27065 (arm_expand_neon_args): Rename to ...
27066 (arm_expand_builtin_args): ... this. Rename builtin_arg
27067 enum values and differentiate between ARG_BUILTIN_MEMORY
27068 and ARG_BUILTIN_NEON_MEMORY.
27069 (arm_expand_neon_builtin_1): Rename to ...
27070 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27071 values, arm_expand_builtin_args and add bool parameter NEON.
27072 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27073 (arm_expand_vfp_builtin): Likewise.
27074 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27076 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27078 PR middle-end/77484
27079 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27080 * predict.c (tree_estimate_probability_bb): Reverse direction of
27081 polymorphic call predictor.
27083 2017-01-06 David Malcolm <dmalcolm@redhat.com>
27085 * passes.c (execute_one_pass): Split out pass-skipping logic into...
27086 (determine_pass_name_match): ...this new function and...
27087 (should_skip_pass_p): ...this new function.
27089 2017-01-06 Nathan Sidwell <nathan@acm.org>
27091 * ipa-visibility.c (function_and_variable_visibility): Reformat
27092 comments and long lines. Remove extrneous if.
27093 * symtab.c (symtab_node::make_decl_local): Fix code format.
27094 (symtab_node::set_section_for_node): Fix comment typo.
27096 2017-01-06 Martin Liska <mliska@suse.cz>
27099 * lra-constraints.c: Rename invariant to lra_invariant.
27100 * predict.c (set_even_probabilities): Initialize e to NULL.
27102 2017-01-05 Martin Sebor <msebor@redhat.com>
27104 PR tree-optimization/78910
27105 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27106 (format_integer): Correct off-by-one error in the handling
27107 of precision with negative numbers in signed conversions..
27109 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
27111 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
27113 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27115 PR tree-optimization/71016
27116 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
27117 factor_out_conditional_conversion. Formatting fix.
27118 (factor_out_conditional_conversion): Add cond_stmt argument.
27119 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
27120 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
27123 2017-01-05 David Malcolm <dmalcolm@redhat.com>
27125 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
27126 read-rtl-function.o, and selftest-rtl.o.
27127 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
27128 (selftest::aarch64_test_loading_full_dump): New function.
27129 (selftest::aarch64_run_selftests): New function.
27130 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27131 selftest::aarch64_run_selftests.
27132 * config/i386/i386.c
27133 (selftest::ix86_test_loading_dump_fragment_1): New function.
27134 (selftest::ix86_test_loading_call_insn): New function.
27135 (selftest::ix86_test_loading_full_dump): New function.
27136 (selftest::ix86_test_loading_unspec): New function.
27137 (selftest::ix86_run_selftests): Call the new functions.
27138 * emit-rtl.c (maybe_set_max_label_num): New function.
27139 * emit-rtl.h (maybe_set_max_label_num): New decl.
27140 * function.c (instantiate_decls): Guard call to
27141 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27142 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27144 * gensupport.c (gen_reader::gen_reader): Pass "false"
27145 for new "compact" param of rtx_reader.
27146 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27147 rather than an empty string for NULL strings.
27148 * read-md.c: Potentially include config.h rather than bconfig.h.
27149 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27150 (have_error): New global, copied from errors.c.
27151 (md_reader::read_name): Rename to...
27152 (md_reader::read_name_1): ...this, adding "out_loc" param,
27153 and converting "missing name or number" to returning false, rather
27155 (md_reader::read_name): Reimplement in terms of read_name_1.
27156 (md_reader::read_name_or_nil): New function.
27157 (md_reader::read_string): Handle "(nil)" by returning NULL.
27158 (md_reader::md_reader): Add new param "compact".
27159 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27160 (md_reader::read_file): New method.
27161 * read-md.h (md_reader::md_reader): Add new param "compact".
27162 (md_reader::read_file): New method.
27163 (md_reader::is_compact): New accessor.
27164 (md_reader::read_name): Convert return type from void to file_location.
27165 (md_reader::read_name_or_nil): New decl.
27166 (md_reader::read_name_1): New decl.
27167 (md_reader::m_compact): New field.
27168 (noop_reader::noop_reader): Pass "false" for new "compact" param
27170 (rtx_reader::rtx_reader): Add new "compact" param.
27171 (rtx_reader::read_rtx_operand): Make virtual and convert return
27172 type from void to rtx.
27173 (rtx_reader::read_until): New decl.
27174 (rtx_reader::handle_any_trailing_information): New virtual function.
27175 (rtx_reader::postprocess): New virtual function.
27176 (rtx_reader::finalize_string): New virtual function.
27177 (rtx_reader::m_in_call_function_usage): New field.
27178 (rtx_reader::m_reuse_rtx_by_id): New field.
27179 * read-rtl-function.c: New file.
27180 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27181 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27182 (selftest::verify_three_block_rtl_cfg): New decl.
27183 * read-rtl-function.h: New file.
27184 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27185 For host, include function.h, memmodel.h, and emit-rtl.h.
27186 (one_time_initialization): New function.
27187 (struct compact_insn_name): New struct.
27188 (compact_insn_names): New array.
27189 (find_code): Handle insn codes in compact dumps.
27190 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27191 (bind_subst_iter_and_attr): Likewise.
27192 (add_condition_to_string): Likewise.
27193 (add_condition_to_rtx): Likewise.
27194 (apply_attribute_uses): Likewise.
27195 (add_current_iterators): Likewise.
27196 (apply_iterators): Likewise.
27197 (initialize_iterators): Guard usage of apply_subst_iterator with
27198 #ifdef GENERATOR_FILE.
27199 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27200 (md_reader::read_mapping): Likewise.
27201 (add_define_attr_for_define_subst): Likewise.
27202 (add_define_subst_attr): Likewise.
27203 (read_subst_mapping): Likewise.
27204 (check_code_iterator): Likewise.
27205 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27207 (one_time_initialization): New function.
27208 (rtx_reader::read_until): New method.
27209 (read_flags): New function.
27210 (parse_reg_note_name): New function.
27211 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27212 Handle reuse_rtx ids.
27213 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27214 Add parsing support for RTL dumps, mirroring the special-cases in
27215 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27216 values, and calling handle_any_trailing_information.
27217 (rtx_reader::read_rtx_operand): Convert return type from void
27218 to rtx, returning return_rtx. Handle case 'e'. Call
27219 finalize_string on XSTR and XTMPL fields.
27220 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27221 "(nil)" values were omitted. Call the postprocess vfunc on the
27223 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27224 class ctor. Initialize m_in_call_function_usage. Call
27225 one_time_initialization.
27226 * rtl-tests.c (selftest::test_uncond_jump): Call
27227 set_new_first_and_last_insn.
27228 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27229 * selftest-rtl.c: New file.
27230 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27231 (selftest::get_insn_by_uid): New decl.
27232 * selftest-run-tests.c (selftest::run_tests): Call
27233 read_rtl_function_c_tests.
27234 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27235 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27238 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27240 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27241 operands in a special way. Assert that pos+len <= mode precision.
27243 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27245 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27246 3 argument Alias with unlimited for the negative form.
27247 (fno-vect-cost-model): Removed.
27249 2017-01-05 Martin Liska <mliska@suse.cz>
27251 * hsa-gen.c (gen_hsa_divmod): New function.
27252 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27254 2017-01-05 Martin Liska <mliska@suse.cz>
27257 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27260 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27262 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27263 small constant length operands.
27265 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27267 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27268 between loop iterations.
27270 2017-01-05 Martin Liska <mliska@suse.cz>
27273 * gimplify.c (gimplify_decl_expr): Compare to
27274 asan_poisoned_variables instread of checking flags.
27275 (gimplify_target_expr): Likewise.
27276 (gimplify_expr): Likewise.
27277 (gimplify_function_tree): Conditionally initialize
27278 asan_poisoned_variables.
27280 2017-01-04 Jeff Law <law@redhat.com>
27282 PR tree-optimizatin/78812
27283 * rtl.h (contains_mem_rtx_p): Prototype.
27284 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27285 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27286 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27287 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27288 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27290 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27292 * input.c (assert_char_at_range): Default-initialize actual_range.
27294 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27296 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27297 to match the caller.
27299 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27301 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27302 insns after final jump in test to emit dummy move.
27304 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27306 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27307 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27309 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27311 * multiple_target.c (create_dispatcher_calls): Init e_next.
27312 * tree-ssa-loop-split.c (split_loop): Init border.
27313 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27316 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27321 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27322 (altivec_register_operand): Do not return true if the operand
27323 contains a SUBREG mixing SImode and SFmode.
27324 (vsx_register_operand): Likewise.
27325 (vsx_reg_sfsubreg_ok): New predicate.
27326 (vfloat_operand): Do not return true if the operand contains a
27327 SUBREG mixing SImode and SFmode.
27328 (vint_operand): Likewise.
27329 (vlogical_operand): Likewise.
27330 (gpc_reg_operand): Likewise.
27331 (int_reg_operand): Likewise.
27332 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27333 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27334 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27336 (rs6000_emit_move_si_sf_subreg): New helper function.
27337 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27338 fixup SUBREGs involving SImode and SFmode.
27339 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27340 numbers for the new peephole2 optimization.
27341 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27342 the GLIBC math library that do AND/IOR/XOR operations on single
27343 precision floating point.
27344 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27345 target macros to say whether we need to avoid SUBREGs mixing
27347 (TARGET_ALLOW_SF_SUBREG): Likewise.
27348 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27349 (UNSPEC_SI_FROM_SF): Likewise.
27350 (iorxor): Change spacing.
27351 (and_ior_xor): New iterator for AND, IOR, and XOR.
27352 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27353 (movdi_from_sf_zero_ext): Likewise.
27354 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27355 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27356 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27357 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27358 (fms<mode>4): Likewise.
27359 (fnma<mode>4): Likewise.
27360 (fnms<mode>4): Likewise.
27361 (nfma<mode>4): Likewise.
27362 (nfms<mode>4): Likewise.
27364 2017-01-04 Marek Polacek <polacek@redhat.com>
27367 * doc/invoke.texi: Document -Wpointer-compare.
27369 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27371 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27374 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27375 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27378 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27381 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27382 documentation of the powerpc_popcntb_ok attribute.
27383 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27384 code to issue warning messages if a requested CPU configuration is
27385 not supported by the binary (assembler and loader) toolchain.
27386 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27387 made to define a built-in function that has been disabled.
27388 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27389 made to define a built-in function that has been disabled.
27390 (altivec_init_builtins): Add comment explaining why definition
27391 of the DST built-in functions is not preceded by an assertion
27392 check. Add assertions to prevent ICE if attempts are made to
27393 define an altivec predicate or an abs* built-in function that has
27395 (htm_init_builtins): Add comment explaining why definition of the
27396 htm built-in functions is not preceded by an assertion check.
27398 2017-01-04 Jeff Law <law@redhat.com>
27400 PR tree-optimizatin/67955
27401 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27402 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27403 the points-to solution does not include pt_null. Use DECL_PT_UID
27406 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27408 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27409 Use gen_int_mode instead of gen_lopwart for const_int operands.
27411 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27413 PR tree-optimization/71563
27414 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27415 out of range value - has low bits known to be zero.
27417 2017-01-04 Alan Modra <amodra@gmail.com>
27419 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27420 * configure: Regenerate.
27421 * config.in: Regenerate.
27423 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27426 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27427 a substring of the message, but strcmp with the whole message. Ifdef
27428 ENABLE_NLS, translate the message first using dgettext.
27430 2017-01-03 Jeff Law <law@redhat.com>
27432 PR tree-optimizatin/78856
27433 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27434 (mark_threaded_blocks): Remove code to truncate thread paths that
27435 cross multiple loop headers. Instead invalidate the cached loop
27436 iteration information and handle case of a thread path walking
27437 into an irreducible region.
27439 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27442 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27443 assertions. Add support for doing the signbit if the IEEE 128-bit
27444 floating point value is in a GPR.
27445 * config/rs6000/rs6000.md (Fsignbit): Delete.
27446 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27447 Update the length attribute if the value is in a GPR.
27448 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27449 the sign or zero extension instruction, since the value is always 0/1.
27450 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27453 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27454 extracting SImode to a GPR register so that we can generate a
27455 store, limit the vector to be in a traditional Altivec register
27456 for the vextuwrx instruction.
27458 2017-01-03 Ian Lance Taylor <iant@google.com>
27460 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27462 2017-01-03 Martin Sebor <msebor@redhat.com>
27464 PR tree-optimization/78696
27465 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27466 precision. Use MPFR for %f for greater fidelity. Correct handling
27468 (pass_sprintf_length::compute_format_length): Set width and precision
27469 specified by asrerisk to void_node for vararg functions.
27470 (try_substitute_return_value): Adjust dump output.
27472 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27474 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27476 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27478 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27479 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27480 -mlra/-mno-lra was passed to the compiler.
27482 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27484 PR rtl-optimization/65618
27485 * emit-rtl.c (try_split): Move initialization of "before" and
27486 "after" to just before the call to emit_insn_after_setloc.
27488 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27490 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27492 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27494 * dwarf2out.c (gen_enumeration_type_die): When
27495 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27497 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27499 PR tree-optimization/78965
27500 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27501 Change first argument from const call_info & to call_info &. For %n
27502 set info.nowrite to false.
27504 PR middle-end/78901
27505 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27506 possibly throwing calls.
27508 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27509 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27510 and fns handling, rather than in a separate case SSA_NAME.
27512 2017-01-02 Jeff Law <law@redhat.com>
27514 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27515 fixes for first_period and second_period variables.
27517 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27520 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27521 (*insvqi_1): New insn pattern.
27522 (*insvqi_1_mem_rex64): Ditto.
27523 (*insvqi_2): Ditto.
27524 (*insvqi_3): Rename from *insvqi.
27526 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27528 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27530 * doc/cfg.texi (Edges): Remove reference to Java.
27531 (Maintaining the CFG): Ditto.
27533 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27535 PR middle-end/77674
27536 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27537 transparent aliases.
27539 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27541 PR middle-end/77484
27542 * predict.def (PRED_CALL): Update hitrate.
27543 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27544 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27545 into direct/indirect/polymorphic variants.
27547 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27549 Update copyright years.
27551 * gcc.c (process_command): Update copyright notice dates.
27552 * gcov-dump.c (print_version): Ditto.
27553 * gcov.c (print_version): Ditto.
27554 * gcov-tool.c (print_version): Ditto.
27555 * gengtype.c (create_file): Ditto.
27556 * doc/cpp.texi: Bump @copying's copyright year.
27557 * doc/cppinternals.texi: Ditto.
27558 * doc/gcc.texi: Ditto.
27559 * doc/gccint.texi: Ditto.
27560 * doc/gcov.texi: Ditto.
27561 * doc/install.texi: Ditto.
27562 * doc/invoke.texi: Ditto.
27564 Copyright (C) 2017 Free Software Foundation, Inc.
27566 Copying and distribution of this file, with or without modification,
27567 are permitted in any medium without royalty provided the copyright
27568 notice and this notice are preserved.